xxq250
/
homework-jianmu
1
0
Fork 0
Code Issues Pull Requests 1 Actions Packages Projects Releases Wiki Activity

other: merge 3.0

This commit is contained in:
Haojun Liao 2023-11-30 14:36:18 +08:00
parent de82aba2e5 215dbeaa3d
commit 999afd97e5
162 changed files with 7056 additions and 4030 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
Jenkinsfile2
View File

@ -355,7 +355,7 @@ pipeline {
}
parallel {
stage('check docs') {
agent{label " slave1_47 || slave1_48 || slave1_49 || slave1_52 || worker03 || slave215 || slave217 || slave219 || Mac_catalina "}
agent{label " slave1_47 || slave1_48 || slave1_49 || slave1_50 || slave1_52 || slave1_59 || slave1_63 || worker03 || slave215 || slave217 || slave219 || Mac_catalina "}
steps {
check_docs()
}
@ -401,7 +401,7 @@ pipeline {
}
}
stage('linux test') {
agent{label " slave1_47 || slave1_48 || slave1_49 || slave1_52 || worker03 || slave215 || slave217 || slave219 "}
agent{label " slave1_47 || slave1_48 || slave1_49 || slave1_50 || slave1_52 || slave1_59 || slave1_63 || worker03 || slave215 || slave217 || slave219 "}
options { skipDefaultCheckout() }
when {
changeRequest()

1
cmake/cmake.options
View File

@ -151,6 +151,7 @@ IF(${BUILD_S3})
IF(${BUILD_WITH_S3})
add_definitions(-DUSE_S3)
option(BUILD_WITH_COS "If build with cos" OFF)
ELSE ()

2
docs/en/05-get-started/03-package.md
View File

@ -243,7 +243,7 @@ You can use the TDengine CLI to monitor your TDengine deployment and execute ad
taos
```
The TDengine CLI displays a welcome message and version information to indicate that its connection to the TDengine service was successful. If an error message is displayed, see the [FAQ](/train-faq/faq) for troubleshooting information. At the following prompt, you can execute SQL commands.
The TDengine CLI displays a welcome message and version information to indicate that its connection to the TDengine service was successful. If an error message is displayed, see the [FAQ](../../train-faq/faq) for troubleshooting information. At the following prompt, you can execute SQL commands.
```cmd
taos>

2
docs/en/05-get-started/_pkg_install.mdx
View File

@ -10,7 +10,7 @@ Between official releases, beta versions may be released that contain new featur
<PkgList type={0}/>
For information about installing TDengine, see [Install and Uninstall](/operation/pkg-install).
For information about installing TDengine, see [Install and Uninstall](../../operation/pkg-install).
For information about TDengine releases, see [All Downloads](https://tdengine.com/all-downloads)

2
docs/en/05-get-started/index.md
View File

@ -12,7 +12,7 @@ import StackOverflowSVG from './stackoverflow.svg'
You can install and run TDengine on Linux/Windows/macOS machines as well as Docker containers. You can also deploy TDengine as a managed service with TDengine Cloud.
The full package of TDengine includes the TDengine Server (`taosd`), TDengine Client (`taosc`), taosAdapter for connecting with third-party systems and providing a RESTful interface, a command-line interface, and some tools. In addition to connectors for multiple languages, TDengine also provides a [RESTful interface](/reference/rest-api) through [taosAdapter](/reference/taosadapter).
The full package of TDengine includes the TDengine Server (`taosd`), TDengine Client (`taosc`), taosAdapter for connecting with third-party systems and providing a RESTful interface, a command-line interface, and some tools. In addition to connectors for multiple languages, TDengine also provides a [RESTful interface](../reference/rest-api) through [taosAdapter](../reference/taosadapter).
```mdx-code-block
import DocCardList from '@theme/DocCardList';

2
docs/en/07-develop/01-connect/_connect_java.mdx
View File

@ -12,4 +12,4 @@ When using REST connection, the feature of bulk pulling can be enabled if the si
{{#include docs/examples/java/src/main/java/com/taos/example/WSConnectExample.java:main}}
```
More configuration about connection, please refer to [Java Connector](/reference/connector/java)
More configuration about connection, please refer to [Java Connector](../../reference/connector/java)

4
docs/en/07-develop/01-connect/index.md
View File

@ -22,7 +22,7 @@ import VerifyLinux from "../../14-reference/03-connector/_verify_linux.mdx";
import VerifyWindows from "../../14-reference/03-connector/_verify_windows.mdx";
import VerifyMacOS from "../../14-reference/03-connector/_verify_macos.mdx";
Any application running on any platform can access TDengine through the REST API provided by TDengine. For information, see [REST API](/reference/rest-api/). Applications can also use the connectors for various programming languages, including C/C++, Java, Python, Go, Node.js, C#, and Rust, to access TDengine. These connectors support connecting to TDengine clusters using both native interfaces (taosc). Some connectors also support connecting over a REST interface. Community developers have also contributed several unofficial connectors, such as the ADO.NET connector, the Lua connector, and the PHP connector.
Any application running on any platform can access TDengine through the REST API provided by TDengine. For information, see [REST API](../../reference/rest-api/). Applications can also use the connectors for various programming languages, including C/C++, Java, Python, Go, Node.js, C#, and Rust, to access TDengine. These connectors support connecting to TDengine clusters using both native interfaces (taosc). Some connectors also support connecting over a REST interface. Community developers have also contributed several unofficial connectors, such as the ADO.NET connector, the Lua connector, and the PHP connector.
## Establish Connection
@ -36,7 +36,7 @@ For REST and native connections, connectors provide similar APIs for performing
Key differences:
3. The REST connection is more accessible with cross-platform support, however it results in a 30% performance downgrade.
1. The TDengine client driver (taosc) has the highest performance with all the features of TDengine like [Parameter Binding](/reference/connector/cpp#parameter-binding-api), [Subscription](/reference/connector/cpp#subscription-and-consumption-api), etc.
1. The TDengine client driver (taosc) has the highest performance with all the features of TDengine like [Parameter Binding](../../reference/connector/cpp#parameter-binding-api), [Subscription](../../reference/connector/cpp#subscription-and-consumption-api), etc.
## Install Client Driver taosc

14
docs/en/07-develop/02-model/index.mdx
View File

@ -3,9 +3,9 @@ title: Data Model
description: This document describes the data model of TDengine.
---
The data model employed by TDengine is similar to that of a relational database. You have to create databases and tables. You must design the data model based on your own business and application requirements. You should design the [STable](/concept/#super-table-stable) (an abbreviation for super table) schema to fit your data. This chapter will explain the big picture without getting into syntactical details.
The data model employed by TDengine is similar to that of a relational database. You have to create databases and tables. You must design the data model based on your own business and application requirements. You should design the [STable](../../concept/#super-table-stable) (an abbreviation for super table) schema to fit your data. This chapter will explain the big picture without getting into syntactical details.
Note: before you read this chapter, please make sure you have already read through [Key Concepts](/concept/), since TDengine introduces new concepts like "one table for one [data collection point](/concept/#data-collection-point)" and "[super table](/concept/#super-table-stable)".
Note: before you read this chapter, please make sure you have already read through [Key Concepts](../../concept/), since TDengine introduces new concepts like "one table for one [data collection point](../../concept/#data-collection-point)" and "[super table](../../concept/#super-table-stable)".
## Create Database
@ -22,7 +22,7 @@ In the above SQL statement:
- a new data file will be created every 10 days
- the size of the write cache pool on each VNode is 16 MB
- the number of vgroups is 100
- WAL is enabled but fsync is disabled For more details please refer to [Database](/taos-sql/database).
- WAL is enabled but fsync is disabled For more details please refer to [Database](../../taos-sql/database).
After creating a database, the current database in use can be switched using SQL command `USE`. For example the SQL statement below switches the current database to `power`.
@ -41,13 +41,13 @@ Without the current database specified, table name must be preceded with the cor
## Create STable
In a time-series application, there may be multiple kinds of data collection points. For example, in the electrical power system there are meters, transformers, bus bars, switches, etc. For easy and efficient aggregation of multiple tables, one STable needs to be created for each kind of data collection point. For example, for the meters in [table 1](/concept/#model_table1), the SQL statement below can be used to create the super table.
In a time-series application, there may be multiple kinds of data collection points. For example, in the electrical power system there are meters, transformers, bus bars, switches, etc. For easy and efficient aggregation of multiple tables, one STable needs to be created for each kind of data collection point. For example, for the meters in [table 1](../../concept/#model_table1), the SQL statement below can be used to create the super table.
```sql
CREATE STABLE meters (ts timestamp, current float, voltage int, phase float) TAGS (location binary(64), groupId int);
```
Similar to creating a regular table, when creating a STable, the name and schema need to be provided. In the STable schema, the first column must always be a timestamp (like ts in the example), and the other columns (like current, voltage and phase in the example) are the data collected. The remaining columns can [contain data of type](/taos-sql/data-type/) integer, float, double, string etc. In addition, the schema for tags, like location and groupId in the example, must be provided. The tag type can be integer, float, string, etc. Tags are essentially the static properties of a data collection point. For example, properties like the location, device type, device group ID, manager ID are tags. Tags in the schema can be added, removed or updated. Please refer to [STable](/taos-sql/stable) for more details.
Similar to creating a regular table, when creating a STable, the name and schema need to be provided. In the STable schema, the first column must always be a timestamp (like ts in the example), and the other columns (like current, voltage and phase in the example) are the data collected. The remaining columns can [contain data of type](../../taos-sql/data-type/) integer, float, double, string etc. In addition, the schema for tags, like location and groupId in the example, must be provided. The tag type can be integer, float, string, etc. Tags are essentially the static properties of a data collection point. For example, properties like the location, device type, device group ID, manager ID are tags. Tags in the schema can be added, removed or updated. Please refer to [STable](../../taos-sql/stable) for more details.
For each kind of data collection point, a corresponding STable must be created. There may be many STables in an application. For electrical power system, we need to create a STable respectively for meters, transformers, busbars, switches. There may be multiple kinds of data collection points on a single device, for example there may be one data collection point for electrical data like current and voltage and another data collection point for environmental data like temperature, humidity and wind direction. Multiple STables are required for these kinds of devices.
@ -61,7 +61,7 @@ A specific table needs to be created for each data collection point. Similar to
CREATE TABLE d1001 USING meters TAGS ("California.SanFrancisco", 2);
```
In the above SQL statement, "d1001" is the table name, "meters" is the STable name, followed by the value of tag "Location" and the value of tag "groupId", which are "California.SanFrancisco" and "2" respectively in the example. The tag values can be updated after the table is created. Please refer to [Tables](/taos-sql/table) for details.
In the above SQL statement, "d1001" is the table name, "meters" is the STable name, followed by the value of tag "Location" and the value of tag "groupId", which are "California.SanFrancisco" and "2" respectively in the example. The tag values can be updated after the table is created. Please refer to [Tables](../../taos-sql/table) for details.
It's suggested to use the globally unique ID of a data collection point as the table name. For example the device serial number could be used as a unique ID. If a unique ID doesn't exist, multiple IDs that are not globally unique can be combined to form a globally unique ID. It's not recommended to use a globally unique ID as tag value.
@ -75,7 +75,7 @@ INSERT INTO d1001 USING meters TAGS ("California.SanFrancisco", 2) VALUES (now,
In the above SQL statement, a row with value `(now, 10.2, 219, 0.32)` will be inserted into table "d1001". If table "d1001" doesn't exist, it will be created automatically using STable "meters" as template with tag value `"California.SanFrancisco", 2`.
For more details please refer to [Create Table Automatically](/taos-sql/insert#automatically-create-table-when-inserting).
For more details please refer to [Create Table Automatically](../../taos-sql/insert#automatically-create-table-when-inserting).
## Single Column vs Multiple Column

8
docs/en/07-develop/03-insert-data/01-sql-writing.mdx
View File

@ -33,7 +33,7 @@ The below SQL statement is used to insert one row into table "d1001".
INSERT INTO d1001 VALUES (ts1, 10.3, 219, 0.31);
```
`ts1` is Unix timestamp, the timestamps which is larger than the difference between current time and KEEP in config is only allowed. For further detail, refer to [TDengine SQL insert timestamp section](/taos-sql/insert).
`ts1` is Unix timestamp, the timestamps which is larger than the difference between current time and KEEP in config is only allowed. For further detail, refer to [TDengine SQL insert timestamp section](../../../taos-sql/insert).
### Insert Multiple Rows
@ -43,7 +43,7 @@ Multiple rows can be inserted in a single SQL statement. The example below inser
INSERT INTO d1001 VALUES (ts2, 10.2, 220, 0.23) (ts2, 10.3, 218, 0.25);
```
`ts1` and `ts2` is Unix timestamp, the timestamps which is larger than the difference between current time and KEEP in config is only allowed. For further detail, refer to [TDengine SQL insert timestamp section](/taos-sql/insert).
`ts1` and `ts2` is Unix timestamp, the timestamps which is larger than the difference between current time and KEEP in config is only allowed. For further detail, refer to [TDengine SQL insert timestamp section](../../../taos-sql/insert).
### Insert into Multiple Tables
@ -53,9 +53,9 @@ Data can be inserted into multiple tables in the same SQL statement. The example
INSERT INTO d1001 VALUES (ts1, 10.3, 219, 0.31) (ts2, 12.6, 218, 0.33) d1002 VALUES (ts3, 12.3, 221, 0.31);
```
`ts1`, `ts2` and `ts3` is Unix timestamp, the timestamps which is larger than the difference between current time and KEEP in config is only allowed. For further detail, refer to [TDengine SQL insert timestamp section](/taos-sql/insert).
`ts1`, `ts2` and `ts3` is Unix timestamp, the timestamps which is larger than the difference between current time and KEEP in config is only allowed. For further detail, refer to [TDengine SQL insert timestamp section](../../../taos-sql/insert).
For more details about `INSERT` please refer to [INSERT](/taos-sql/insert).
For more details about `INSERT` please refer to [INSERT](../../../taos-sql/insert).
:::info

2
docs/en/07-develop/03-insert-data/20-kafka-writting.mdx
View File

@ -35,7 +35,7 @@ bin/kafka-topics.sh --bootstrap-server=localhost:9092 --describe
## Insert into TDengine
We can write data into TDengine via SQL or Schemaless. For more information, please refer to [Insert Using SQL](/develop/insert-data/sql-writing/) or [High Performance Writing](/develop/insert-data/high-volume/) or [Schemaless Writing](/reference/schemaless/).
We can write data into TDengine via SQL or Schemaless. For more information, please refer to [Insert Using SQL](../sql-writing/) or [High Performance Writing](../high-volume/) or [Schemaless Writing](../../../reference/schemaless/).
## Examples

2
docs/en/07-develop/03-insert-data/30-influxdb-line.mdx
View File

@ -46,7 +46,7 @@ meters,location=California.LosAngeles,groupid=2 current=13.4,voltage=223,phase=0
:::
For more details please refer to [InfluxDB Line Protocol](https://docs.influxdata.com/influxdb/v2.0/reference/syntax/line-protocol/) and [TDengine Schemaless](/reference/schemaless/#Schemaless-Line-Protocol)
For more details please refer to [InfluxDB Line Protocol](https://docs.influxdata.com/influxdb/v2.0/reference/syntax/line-protocol/) and [TDengine Schemaless](../../../reference/schemaless/#Schemaless-Line-Protocol)
## Examples

2
docs/en/07-develop/04-query-data/index.mdx
View File

@ -128,7 +128,7 @@ For more information, see [Aggregate by Window](../../taos-sql/distinguished).
### Query
In the section describing [Insert](/develop/insert-data/sql-writing), a database named `power` is created and some data are inserted into STable `meters`. Below sample code demonstrates how to query the data in this STable.
In the section describing [Insert](../insert-data/sql-writing), a database named `power` is created and some data are inserted into STable `meters`. Below sample code demonstrates how to query the data in this STable.
<Tabs defaultValue="java" groupId="lang">
<TabItem label="Java" value="java">

4
docs/en/07-develop/index.md
View File

@ -6,7 +6,7 @@ description: This document describes how to use the various components of TDengi
Before creating an application to process time-series data with TDengine, consider the following:
1. Choose the method to connect to TDengine. TDengine offers a REST API that can be used with any programming language. It also has connectors for a variety of languages.
2. Design the data model based on your own use cases. Consider the main [concepts](/concept/) of TDengine, including "one table per data collection point" and the supertable. Learn about static labels, collected metrics, and subtables. Depending on the characteristics of your data and your requirements, you decide to create one or more databases and design a supertable schema that fit your data.
2. Design the data model based on your own use cases. Consider the main [concepts](../concept/) of TDengine, including "one table per data collection point" and the supertable. Learn about static labels, collected metrics, and subtables. Depending on the characteristics of your data and your requirements, you decide to create one or more databases and design a supertable schema that fit your data.
3. Decide how you will insert data. TDengine supports writing using standard SQL, but also supports schemaless writing, so that data can be written directly without creating tables manually.
4. Based on business requirements, find out what SQL query statements need to be written. You may be able to repurpose any existing SQL.
5. If you want to run real-time analysis based on time series data, including various dashboards, use the TDengine stream processing component instead of deploying complex systems such as Spark or Flink.
@ -14,7 +14,7 @@ Before creating an application to process time-series data with TDengine, consid
7. In many use cases (such as fleet management), the application needs to obtain the latest status of each data collection point. It is recommended that you use the cache function of TDengine instead of deploying Redis separately.
8. If you find that the SQL functions of TDengine cannot meet your requirements, then you can use user-defined functions to solve the problem.
This section is organized in the order described above. For ease of understanding, TDengine provides sample code for each supported programming language for each function. If you want to learn more about the use of SQL, please read the [SQL manual](/taos-sql/). For a more in-depth understanding of the use of each connector, please read the [Connector Reference Guide](/reference/connector/). If you also want to integrate TDengine with third-party systems, such as Grafana, please refer to the [third-party tools](/third-party/).
This section is organized in the order described above. For ease of understanding, TDengine provides sample code for each supported programming language for each function. If you want to learn more about the use of SQL, please read the [SQL manual](../taos-sql/). For a more in-depth understanding of the use of each connector, please read the [Connector Reference Guide](../reference/connector/). If you also want to integrate TDengine with third-party systems, such as Grafana, please refer to the [third-party tools](../third-party/).
If you encounter any problems during the development process, please click ["Submit an issue"](https://github.com/taosdata/TDengine/issues/new/choose) at the bottom of each page and submit it on GitHub right away.

4
docs/en/10-deployment/01-deploy.md
View File

@ -72,7 +72,7 @@ For all the dnodes in a TDengine cluster, the below parameters must be configure
## Start Cluster
The first dnode can be started following the instructions in [Get Started](/get-started/). Then TDengine CLI `taos` can be launched to execute command `show dnodes`, the output is as following for example:
The first dnode can be started following the instructions in [Get Started](../../get-started/). Then TDengine CLI `taos` can be launched to execute command `show dnodes`, the output is as following for example:
```
taos> show dnodes;
@ -90,7 +90,7 @@ From the above output, it is shown that the end point of the started dnode is "h
There are a few steps necessary to add other dnodes in the cluster.
Second, we can start `taosd` as instructed in [Get Started](/get-started/).
Second, we can start `taosd` as instructed in [Get Started](../../get-started/).
Then, on the first dnode i.e. h1.tdengine.com in our example, use TDengine CLI `taos` to execute the following command:

2
docs/en/12-taos-sql/26-udf.md
View File

@ -53,7 +53,7 @@ CREATE AGGREGATE FUNCTION function_name AS library_path OUTPUTTYPE output_type [
CREATE AGGREGATE FUNCTION l2norm AS "/home/taos/udf_example/libl2norm.so" OUTPUTTYPE DOUBLE bufsize 64;
```
For more information about user-defined functions, see [User-Defined Functions](/develop/udf).
For more information about user-defined functions, see [User-Defined Functions](../../develop/udf).
## Manage UDF

2
docs/en/13-operation/10-monitor.md
View File

@ -41,7 +41,7 @@ Launch `TDinsight.sh` with the command above and restart Grafana, then open Dash
## log database
The data of tdinsight dashboard is stored in `log` database (default. You can change it in taoskeeper's config file. For more infrmation, please reference to [taoskeeper document](/reference/taosKeeper)). The taoskeeper will create log database on taoskeeper startup.
The data of tdinsight dashboard is stored in `log` database (default. You can change it in taoskeeper's config file. For more infrmation, please reference to [taoskeeper document](../../reference/taosKeeper)). The taoskeeper will create log database on taoskeeper startup.
### cluster\_info table

2
docs/en/14-reference/02-rest-api/02-rest-api.mdx
View File

@ -514,4 +514,4 @@ Response body:
## Reference
[taosAdapter](/reference/taosadapter/)
[taosAdapter](../taosadapter/)

6
docs/en/14-reference/03-connector/03-cpp.mdx
View File

@ -24,7 +24,7 @@ The dynamic libraries for the TDengine client driver are located in.
## Supported platforms
Please refer to [list of supported platforms](/reference/connector#supported-platforms)
Please refer to [list of supported platforms](../#supported-platforms)
## Supported versions
@ -32,7 +32,7 @@ The version number of the TDengine client driver and the version number of the T
## Installation Steps
Please refer to the [Installation Steps](/reference/connector#installation-steps) for TDengine client driver installation
Please refer to the [Installation Steps](../#installation-steps) for TDengine client driver installation
## Establishing a connection
@ -394,7 +394,7 @@ The specific functions related to the interface are as follows (see also the [pr
### Schemaless Writing API
In addition to writing data using the SQL method or the parameter binding API, writing can also be done using schemaless writing, which eliminates the need to create a super table/data sub-table structure in advance and writes the data directly. The TDengine system automatically creates and maintains the required table structure based on the written data content. The use of schemaless writing is described in the chapter [Schemaless Writing](/reference/schemaless/), and the C/C++ API used with it is described here.
In addition to writing data using the SQL method or the parameter binding API, writing can also be done using schemaless writing, which eliminates the need to create a super table/data sub-table structure in advance and writes the data directly. The TDengine system automatically creates and maintains the required table structure based on the written data content. The use of schemaless writing is described in the chapter [Schemaless Writing](../../schemaless/), and the C/C++ API used with it is described here.
- `TAOS_RES* taos_schemaless_insert(TAOS* taos, const char* lines[], int numLines, int protocol, int precision)`

6
docs/en/14-reference/03-connector/04-java.mdx
View File

@ -148,7 +148,7 @@ TDengine currently supports timestamp, number, character, Boolean type, and the
**Note**: Only TAG supports JSON types
Due to historical reasons, the BINARY type data in TDengine is not truly binary data and is no longer recommended for use. Please use VARBINARY type instead.
GEOMETRY type is binary data in little endian byte order, which complies with the WKB specification. For detailed information, please refer to [Data Type] (/tao-sql/data-type/#Data Types)
GEOMETRY type is binary data in little endian byte order, which complies with the WKB specification. For detailed information, please refer to [Data Type](../../../taos-sql/data-type/)
For WKB specifications, please refer to [Well Known Binary (WKB)]（ https://libgeos.org/specifications/wkb/ ）
For Java connector, the jts library can be used to easily create GEOMETRY type objects, serialize them, and write them to TDengine. Here is an example [Geometry example]（https://github.com/taosdata/TDengine/blob/3.0/examples/JDBC/JDBCDemo/src/main/java/com/taosdata/example/GeometryDemo.java）
@ -160,7 +160,7 @@ For Java connector, the jts library can be used to easily create GEOMETRY type o
Before using Java Connector to connect to the database, the following conditions are required.
- Java 1.8 or above runtime environment and Maven 3.6 or above installed
- TDengine client driver installed (required for native connections, not required for REST connections), please refer to [Installing Client Driver](/reference/connector#Install-Client-Driver)
- TDengine client driver installed (required for native connections, not required for REST connections), please refer to [Installing Client Driver](../#Install-Client-Driver)
### Install the connectors
@ -368,7 +368,7 @@ The configuration parameters in properties are as follows.
- TSDBDriver.PROPERTY_KEY_MESSAGE_WAIT_TIMEOUT: message transmission timeout in milliseconds, the default value is 60000 ms. It only takes effect when using JDBC REST connection and batchfetch is true.
- TSDBDriver.PROPERTY_KEY_USE_SSL: connecting Securely Using SSL. true: using SSL connection, false: not using SSL connection. It only takes effect when using JDBC REST connection.
- TSDBDriver.HTTP_POOL_SIZE: size of REST concurrent requests. The default value is 20.
For JDBC native connections, you can specify other parameters, such as log level, SQL length, etc., by specifying URL and Properties. For more detailed configuration, please refer to [Client Configuration](/reference/config/#Client-Only).
For JDBC native connections, you can specify other parameters, such as log level, SQL length, etc., by specifying URL and Properties. For more detailed configuration, please refer to [Client Configuration](../../config/#Client-Only).
### Priority of configuration parameters

2
docs/en/14-reference/03-connector/05-go.mdx
View File

@ -74,7 +74,7 @@ If it is a TDengine error, you can get the error code and error information in t
### Pre-installation preparation
* Install Go development environment (Go 1.14 and above, GCC 4.8.5 and above)
* If you use the native connector, please install the TDengine client driver. Please refer to [Install Client Driver](/reference/connector/#install-client-driver) for specific steps
* If you use the native connector, please install the TDengine client driver. Please refer to [Install Client Driver](../#install-client-driver) for specific steps
Configure the environment variables and check the command.

2
docs/en/14-reference/03-connector/06-rust.mdx
View File

@ -80,7 +80,7 @@ Note: Only TAG supports JSON types
### Pre-installation preparation
* Install the Rust development toolchain
* If using the native connection, please install the TDengine client driver. Please refer to [install client driver](/reference/connector#install-client-driver)
* If using the native connection, please install the TDengine client driver. Please refer to [install client driver](../#install-client-driver)
### Install the connectors

12
docs/en/14-reference/03-connector/07-python.mdx
View File

@ -7,7 +7,7 @@ description: This document describes taospy, the TDengine Python connector.
import Tabs from "@theme/Tabs";
import TabItem from "@theme/TabItem";
`taospy` is the official Python connector for TDengine. taospy provides a rich API that makes it easy for Python applications to use TDengine. `taospy` wraps both the [native interface](/reference/connector/cpp) and [REST interface](/reference/rest-api) of TDengine, which correspond to the `taos` and `taosrest` modules of the `taospy` package, respectively.
`taospy` is the official Python connector for TDengine. taospy provides a rich API that makes it easy for Python applications to use TDengine. `taospy` wraps both the [native interface](../cpp) and [REST interface](../../rest-api) of TDengine, which correspond to the `taos` and `taosrest` modules of the `taospy` package, respectively.
In addition to wrapping the native and REST interfaces, `taospy` also provides a set of programming interfaces that conforms to the [Python Data Access Specification (PEP 249)](https://peps.python.org/pep-0249/). It is easy to integrate `taospy` with many third-party tools, such as [SQLAlchemy](https://www.sqlalchemy.org/) and [pandas](https://pandas.pydata.org/).
`taos-ws-py` is an optional package to enable using WebSocket to connect TDengine.
@ -17,7 +17,7 @@ The direct connection to the server using the native interface provided by the c
The source code for the Python connector is hosted on [GitHub](https://github.com/taosdata/taos-connector-python).
## Supported platforms
- The [supported platforms](/reference/connector/#supported-platforms) for the native connection are the same as the ones supported by the TDengine client.
- The [supported platforms](../#supported-platforms) for the native connection are the same as the ones supported by the TDengine client.
- REST connections are supported on all platforms that can run Python.
### Supported features
@ -95,7 +95,7 @@ TDengine currently supports timestamp, number, character, Boolean type, and the
1. Install Python. The recent taospy package requires Python 3.6.2+. The earlier versions of taospy require Python 3.7+. The taos-ws-py package requires Python 3.7+. If Python is not available on your system, refer to the [Python BeginnersGuide](https://wiki.python.org/moin/BeginnersGuide/Download) to install it.
2. Install [pip](https://pypi.org/project/pip/). In most cases, the Python installer comes with the pip utility. If not, please refer to [pip documentation](https://pip.pypa.io/en/stable/installation/) to install it.
If you use a native connection, you will also need to [Install Client Driver](/reference/connector#Install-Client-Driver). The client install package includes the TDengine client dynamic link library (`libtaos.so` or `taos.dll`) and the TDengine CLI.
If you use a native connection, you will also need to [Install Client Driver](../#Install-Client-Driver). The client install package includes the TDengine client dynamic link library (`libtaos.so` or `taos.dll`) and the TDengine CLI.
### Install via pip
@ -444,7 +444,7 @@ The best practice for TaosCursor is to create a cursor at the beginning of a que
##### Use of the RestClient class
The `RestClient` class is a direct wrapper for the [REST API](/reference/rest-api). It contains only a `sql()` method for executing arbitrary SQL statements and returning the result.
The `RestClient` class is a direct wrapper for the [REST API](../../rest-api). It contains only a `sql()` method for executing arbitrary SQL statements and returning the result.
```python title="Use of RestClient"
{{#include docs/examples/python/rest_client_example.py}}
@ -501,7 +501,7 @@ The queried results can only be fetched once. For example, only one of `fetch_al
<TabItem value="rest" label="REST connection">
The `RestClient` class is a direct wrapper for the [REST API](/reference/rest-api). It contains only a `sql()` method for executing arbitrary SQL statements and returning the result.
The `RestClient` class is a direct wrapper for the [REST API](../../rest-api). It contains only a `sql()` method for executing arbitrary SQL statements and returning the result.
```python
{{#include docs/examples/python/rest_client_example.py}}
@ -561,7 +561,7 @@ The `TaosConnection` class and the `TaosResult` class already implement all the
##### Use of the RestClient class
The `RestClient` class is a direct wrapper for the [REST API](/reference/rest-api). It contains only a `sql()` method for executing arbitrary SQL statements and returning the result.
The `RestClient` class is a direct wrapper for the [REST API](../../rest-api). It contains only a `sql()` method for executing arbitrary SQL statements and returning the result.
```python title="Use of RestClient"
{{#include docs/examples/python/rest_client_with_req_id_example.py}}

4
docs/en/14-reference/03-connector/08-node.mdx
View File

@ -28,7 +28,7 @@ The REST connector supports all platforms that can run Node.js.
## Version support
Please refer to [version support list](/reference/connector#version-support)
Please refer to [version support list](../#version-support)
## Supported features
@ -58,7 +58,7 @@ Please refer to [version support list](/reference/connector#version-support)
### Pre-installation preparation
- Install the Node.js development environment
- If you are using the REST connector, skip this step. However, if you use the native connector, please install the TDengine client driver. Please refer to [Install Client Driver](/reference/connector#Install-Client-Driver) for more details. We use [node-gyp](https://github.com/nodejs/node-gyp) to interact with TDengine instances and also need to install some dependencies mentioned below depending on the specific OS.
- If you are using the REST connector, skip this step. However, if you use the native connector, please install the TDengine client driver. Please refer to [Install Client Driver](../#Install-Client-Driver) for more details. We use [node-gyp](https://github.com/nodejs/node-gyp) to interact with TDengine instances and also need to install some dependencies mentioned below depending on the specific OS.
<Tabs defaultValue="Linux">
<TabItem value="Linux" label="Linux system installation dependencies">

4
docs/en/14-reference/03-connector/09-csharp.mdx
View File

@ -36,7 +36,7 @@ Please note TDengine does not support 32bit Windows any more.
## Version support
Please refer to [version support list](/reference/connector#version-support)
Please refer to [version support list](../#version-support)
## Supported features
@ -69,7 +69,7 @@ Please refer to [version support list](/reference/connector#version-support)
* Install the [.NET SDK](https://dotnet.microsoft.com/download)
* [Nuget Client](https://docs.microsoft.com/en-us/nuget/install-nuget-client-tools) (optional installation)
* Install TDengine client driver, please refer to [Install client driver](/reference/connector/#install-client-driver) for details
* Install TDengine client driver, please refer to [Install client driver](../#install-client-driver) for details
### Install `TDengine.Connector`

2
docs/en/14-reference/03-connector/80-php.mdx
View File

@ -40,7 +40,7 @@ Because the version of TDengine client driver is tightly associated with that of
### Install TDengine Client Driver
Regarding how to install TDengine client driver please refer to [Install Client Driver](/reference/connector#installation-steps)
Regarding how to install TDengine client driver please refer to [Install Client Driver](../#installation-steps)
### Install php-tdengine

2
docs/en/14-reference/03-connector/_preparation.mdx
View File

@ -2,7 +2,7 @@
:::info
Since the TDengine client driver is written in C, using the native connection requires loading the client driver shared library file, which is usually included in the TDengine installer. You can install either standard TDengine server installation package or [TDengine client installation package](/get-started/). For Windows development, you need to install the corresponding Windows client, please refer to [Install TDengine](../../get-started/package).
Since the TDengine client driver is written in C, using the native connection requires loading the client driver shared library file, which is usually included in the TDengine installer. You can install either standard TDengine server installation package or [TDengine client installation package](../../get-started/). For Windows development, you need to install the corresponding Windows client, please refer to [Install TDengine](../../get-started/package).
- libtaos.so: After successful installation of TDengine on a Linux system, the dependent Linux version of the client driver `libtaos.so` file will be automatically linked to `/usr/lib/libtaos.so`, which is included in the Linux scannable path and does not need to be specified separately.
- taos.dll: After installing the client on Windows, the dependent Windows version of the client driver taos.dll file will be automatically copied to the system default search path C:/Windows/System32, again without the need to specify it separately.

4
docs/en/14-reference/04-taosadapter.md
View File

@ -186,7 +186,7 @@ See [example/config/taosadapter.toml](https://github.com/taosdata/taosadapter/bl
### TDengine RESTful interface
You can use any client that supports the http protocol to write data to or query data from TDengine by accessing the REST interface address `http://<fqdn>:6041/rest/sql`. See the [official documentation](/reference/rest-api/) for details.
You can use any client that supports the http protocol to write data to or query data from TDengine by accessing the REST interface address `http://<fqdn>:6041/rest/sql`. See the [official documentation](../rest-api/) for details.
### InfluxDB
@ -202,7 +202,7 @@ Support InfluxDB query parameters as follows.
- `precision` The time precision used by TDengine
- `u` TDengine user name
- `p` TDengine password
- `ttl` The time to live of automatically created sub-table. This value cannot be updated. TDengine will use the ttl value of the first data of sub-table to create sub-table. For more information, please refer [Create Table](/taos-sql/table/#create-table)
- `ttl` The time to live of automatically created sub-table. This value cannot be updated. TDengine will use the ttl value of the first data of sub-table to create sub-table. For more information, please refer [Create Table](../../taos-sql/table/#create-table)
Note: InfluxDB token authorization is not supported at present. Only Basic authorization and query parameter validation are supported.
Example: curl --request POST http://127.0.0.1:6041/influxdb/v1/write?db=test --user "root:taosdata" --data-binary "measurement,host=host1 field1=2i,field2=2.0 1577836800000000000"

2
docs/en/14-reference/06-taosdump.md
View File

@ -31,7 +31,7 @@ There are two ways to install taosdump:
2. backup multiple specified databases: use `-D db1,db2,... ` parameters;
3. back up some super or normal tables in the specified database: use `dbname stbname1 stbname2 tbname1 tbname2 ... ` parameters. Note that the first parameter of this input sequence is the database name, and only one database is supported. The second and subsequent parameters are the names of super or normal tables in that database, separated by spaces.
4. back up the system log database: TDengine clusters usually contain a system database named `log`. The data in this database is the data that TDengine runs itself, and the taosdump will not back up the log database by default. If users need to back up the log database, users can use the `-a` or `-allow-sys` command-line parameter.
5. Loose mode backup: taosdump version 1.4.1 onwards provides `-n` and `-L` parameters for backing up data without using escape characters and "loose" mode, which can reduce the number of backups if table names, column names, tag names do not use escape characters. This can also reduce the backup data time and backup data footprint. If you are unsure about using `-n` and `-L` conditions, please use the default parameters for "strict" mode backup. See the [official documentation](/taos-sql/escape) for a description of escaped characters.
5. Loose mode backup: taosdump version 1.4.1 onwards provides `-n` and `-L` parameters for backing up data without using escape characters and "loose" mode, which can reduce the number of backups if table names, column names, tag names do not use escape characters. This can also reduce the backup data time and backup data footprint. If you are unsure about using `-n` and `-L` conditions, please use the default parameters for "strict" mode backup. See the [official documentation](../../taos-sql/escape) for a description of escaped characters.
:::tip
- taosdump versions after 1.4.1 provide the `-I` argument for parsing Avro file schema and data. If users specify `-s` then only taosdump will parse schema.

4
docs/en/14-reference/08-taos-shell.md
View File

@ -8,7 +8,7 @@ The TDengine command-line interface (hereafter referred to as `TDengine CLI`) is
## Installation
If executed on the TDengine server-side, there is no need for additional installation steps to install TDengine CLI as it is already included and installed automatically. To run TDengine CLI in an environment where no TDengine server is running, the TDengine client installation package needs to be installed first. For details, please refer to [Connector](/reference/connector/).
If executed on the TDengine server-side, there is no need for additional installation steps to install TDengine CLI as it is already included and installed automatically. To run TDengine CLI in an environment where no TDengine server is running, the TDengine client installation package needs to be installed first. For details, please refer to [Connector](../connector/).
## Execution
@ -18,7 +18,7 @@ To access the TDengine CLI, you can execute `taos` command-line utility from a t
taos
```
TDengine CLI will display a welcome message and version information if it successfully connected to the TDengine service. If it fails, TDengine CLI will print an error message. See [FAQ](/train-faq/faq) to solve the problem of terminal connection failure to the server. The TDengine CLI prompts as follows:
TDengine CLI will display a welcome message and version information if it successfully connected to the TDengine service. If it fails, TDengine CLI will print an error message. See [FAQ](../../train-faq/faq) to solve the problem of terminal connection failure to the server. The TDengine CLI prompts as follows:
```cmd
taos>

2
docs/en/14-reference/12-config/index.md
View File

@ -87,7 +87,7 @@ Ensure that your firewall rules do not block TCP port 6042 on any host in the c
| Protocol | Default Port | Description | How to configure |
| :------- | :----------- | :-------------------------------------------------------------------------------------------------------- | :--------------------------------------------------------------------------------------------- |
| TCP | 6030 | Communication between client and server. In a multi-node cluster, communication between nodes. serverPort |
| TCP | 6041 | REST connection between client and server | Prior to 2.4.0.0: serverPort+11; After 2.4.0.0 refer to [taosAdapter](/reference/taosadapter/) |
| TCP | 6041 | REST connection between client and server | Prior to 2.4.0.0: serverPort+11; After 2.4.0.0 refer to [taosAdapter](../taosadapter/) |
| TCP | 6043 | Service Port of taosKeeper | The parameter of taosKeeper |
| TCP | 6044 | Data access port for StatsD | Configurable through taosAdapter parameters. |
| UDP | 6045 | Data access for statsd | Configurable through taosAdapter parameters. |

2
docs/en/14-reference/13-schemaless/13-schemaless.md
View File

@ -116,7 +116,7 @@ You can configure smlChildTableName in taos.cfg to specify table names, for exam
10. Taos.cfg adds the configuration of smlTsDefaultName (with a string value), which only works on the client side. After configuration, the time column name of the schemaless automatic table creation can be set through this configuration. If not configured, defaults to _ts.
11. Super table name or child table name are case sensitive.
:::tip
All processing logic of schemaless will still follow TDengine's underlying restrictions on data structures, such as the total length of each row of data cannot exceed 48 KB(64 KB since version 3.0.5.0) and the total length of a tag value cannot exceed 16 KB. See [TDengine SQL Boundary Limits](/taos-sql/limit) for specific constraints in this area.
All processing logic of schemaless will still follow TDengine's underlying restrictions on data structures, such as the total length of each row of data cannot exceed 48 KB(64 KB since version 3.0.5.0) and the total length of a tag value cannot exceed 16 KB. See [TDengine SQL Boundary Limits](../../taos-sql/limit) for specific constraints in this area.
:::
## Time resolution recognition

2
docs/en/20-third-party/02-prometheus.md vendored
View File

@ -16,7 +16,7 @@ Prometheus data can be stored in TDengine via the `remote_write` interface with
To write Prometheus data to TDengine requires the following preparations.
- The TDengine cluster is deployed and functioning properly
- taosAdapter is installed and running properly. Please refer to the [taosAdapter manual](/reference/taosadapter) for details.
- taosAdapter is installed and running properly. Please refer to the [taosAdapter manual](../../reference/taosadapter) for details.
- Prometheus has been installed. Please refer to the [official documentation](https://prometheus.io/docs/prometheus/latest/installation/) for installing Prometheus
## Configuration steps

4
docs/en/20-third-party/03-telegraf.md vendored
View File

@ -14,7 +14,7 @@ Telegraf's data can be written to TDengine by simply adding the output configura
To write Telegraf data to TDengine requires the following preparations.
- The TDengine cluster is deployed and functioning properly
- taosAdapter is installed and running properly. Please refer to the [taosAdapter manual](/reference/taosadapter) for details.
- taosAdapter is installed and running properly. Please refer to the [taosAdapter manual](../../reference/taosadapter) for details.
- Telegraf has been installed. Please refer to the [official documentation](https://docs.influxdata.com/telegraf/v1.22/install/) for Telegraf installation.
- Telegraf collects the running status measurements of current system. You can enable [input plugins](https://docs.influxdata.com/telegraf/v1.22/plugins/) to insert [other formats](https://docs.influxdata.com/telegraf/v1.24/data_formats/input/) data to Telegraf then forward to TDengine.
@ -73,6 +73,6 @@ Query OK, 3 row(s) in set (0.013269s)
- TDengine take influxdb format data and create unique ID for table names by the rule.
The user can configure `smlChildTableName` parameter to generate specified table names if he/she needs. And he/she also need to insert data with specified data format.
For example, Add `smlChildTableName=tname` in the taos.cfg file. Insert data `st,tname=cpu1,t1=4 c1=3 1626006833639000000` then the table name will be cpu1. If there are multiple lines has same tname but different tag_set, the first line's tag_set will be used to automatically creating table and ignore other lines. Please refer to [TDengine Schemaless](/reference/schemaless/#Schemaless-Line-Protocol)
For example, Add `smlChildTableName=tname` in the taos.cfg file. Insert data `st,tname=cpu1,t1=4 c1=3 1626006833639000000` then the table name will be cpu1. If there are multiple lines has same tname but different tag_set, the first line's tag_set will be used to automatically creating table and ignore other lines. Please refer to [TDengine Schemaless](../../reference/schemaless/#Schemaless-Line-Protocol)
:::

2
docs/en/20-third-party/05-collectd.md vendored
View File

@ -15,7 +15,7 @@ You can write the data collected by collectd to TDengine by simply modifying the
Writing collectd data to the TDengine requires several preparations.
- The TDengine cluster is deployed and running properly
- taosAdapter is installed and running, please refer to [taosAdapter's manual](/reference/taosadapter) for details
- taosAdapter is installed and running, please refer to [taosAdapter's manual](../../reference/taosadapter) for details
- collectd has been installed. Please refer to the [official documentation](https://collectd.org/download.shtml) to install collectd
## Configuration steps

2
docs/en/20-third-party/07-icinga2.md vendored
View File

@ -14,7 +14,7 @@ You can write the data collected by icinga2 to TDengine by simply modifying the
To write icinga2 data to TDengine requires the following preparations.
- The TDengine cluster is deployed and working properly
- taosAdapter is installed and running properly. Please refer to the [taosAdapter manual](/reference/taosadapter) for details.
- taosAdapter is installed and running properly. Please refer to the [taosAdapter manual](../../reference/taosadapter) for details.
- icinga2 has been installed. Please refer to the [official documentation](https://icinga.com/docs/icinga-2/latest/doc/02-installation/) for icinga2 installation
## Configuration steps

2
docs/en/20-third-party/08-tcollector.md vendored
View File

@ -14,7 +14,7 @@ You can write the data collected by TCollector to TDengine by simply changing th
To write data to the TDengine via TCollector requires the following preparations.
- The TDengine cluster has been deployed and is working properly
- taosAdapter is installed and running properly. Please refer to the [taosAdapter manual](/reference/taosadapter) for details.
- taosAdapter is installed and running properly. Please refer to the [taosAdapter manual](../../reference/taosadapter) for details.
- TCollector has been installed. Please refer to [official documentation](http://opentsdb.net/docs/build/html/user_guide/utilities/tcollector.html#installation-of-tcollector) for TCollector installation
## Configuration steps

2
docs/en/20-third-party/09-emq-broker.md vendored
View File

@ -82,7 +82,7 @@ Edit the resource configuration to add the key/value pairing for Authorization.
Basic cm9vdDp0YW9zZGF0YQ==
```
Please refer to the [ TDengine REST API documentation ](/reference/rest-api/) for the authorization in details.
Please refer to the [ TDengine REST API documentation ](../../reference/rest-api/) for the authorization in details.
Enter the rule engine replacement template in the message body:

4
docs/en/20-third-party/11-kafka.md vendored
View File

@ -94,7 +94,7 @@ The output as bellow:
The role of the TDengine Sink Connector is to synchronize the data of the specified topic to TDengine. Users do not need to create databases and super tables in advance. The name of the target database can be specified manually (see the configuration parameter connection.database), or it can be generated according to specific rules (see the configuration parameter connection.database.prefix).
TDengine Sink Connector internally uses TDengine [modeless write interface](/reference/connector/cpp#modeless write-api) to write data to TDengine, currently supports data in three formats: [InfluxDB line protocol format](/develop /insert-data/influxdb-line), [OpenTSDB Telnet protocol format](/develop/insert-data/opentsdb-telnet), and [OpenTSDB JSON protocol format](/develop/insert-data/opentsdb-json).
TDengine Sink Connector internally uses TDengine [modeless write interface](../../reference/connector/cpp#modeless write-api) to write data to TDengine, currently supports data in three formats: [InfluxDB line protocol format](../../develop/insert-data/influxdb-line), [OpenTSDB Telnet protocol format](../../develop/insert-data/opentsdb-telnet), and [OpenTSDB JSON protocol format](../../develop/insert-data/opentsdb-json).
The following example synchronizes the data of the topic meters to the target database power. The data format is the InfluxDB Line protocol format.
@ -213,7 +213,7 @@ If you see the above data, the synchronization is successful. If not, check the
The role of the TDengine Source Connector is to push all the data of a specific TDengine database after a particular time to Kafka. The implementation principle of TDengine Source Connector is to first pull historical data in batches and then synchronize incremental data with the strategy of the regular query. At the same time, the changes in the table will be monitored, and the newly added table can be automatically synchronized. If Kafka Connect is restarted, synchronization will resume where it left off.
TDengine Source Connector will convert the data in TDengine data table into [InfluxDB Line protocol format](/develop/insert-data/influxdb-line/) or [OpenTSDB JSON protocol format](/develop/insert-data/opentsdb-json ) and then write to Kafka.
TDengine Source Connector will convert the data in TDengine data table into [InfluxDB Line protocol format](../../develop/insert-data/influxdb-line/) or [OpenTSDB JSON protocol format](../../develop/insert-data/opentsdb-json ) and then write to Kafka.
The following sample program synchronizes the data in the database test to the topic tdengine-test-meters.

2
docs/en/20-third-party/_deploytaosadapter.mdx vendored
View File

@ -14,4 +14,4 @@ Check the running status of taosAdapter.
systemctl status taosadapter
```
taosAdapter Please refer to the `taosadapter --help` command output and [reference documentation](/reference/taosadapter) for detailed configuration parameters and usage of taosAdapter.
taosAdapter Please refer to the `taosadapter --help` command output and [reference documentation](../../reference/taosadapter) for detailed configuration parameters and usage of taosAdapter.

2
docs/en/25-application/01-telegraf.md
View File

@ -41,7 +41,7 @@ Download and install the [latest version of TDengine](https://docs.tdengine.com/
### Install Grafana Plugin and Configure Data Source
Please refer to [Install Grafana Plugin and Configure Data Source](/third-party/grafana/#install-grafana-plugin-and-configure-data-source)
Please refer to [Install Grafana Plugin and Configure Data Source](../../third-party/grafana/#install-grafana-plugin-and-configure-data-source)
### Modify /etc/telegraf/telegraf.conf

2
docs/en/25-application/02-collectd.md
View File

@ -44,7 +44,7 @@ Download and install the [latest version of TDengine](https://docs.tdengine.com/
### Install Grafana Plugin and Configure Data Source
Please refer to [Install Grafana Plugin and Configure Data Source](/third-party/grafana/#install-grafana-plugin-and-configure-data-source)
Please refer to [Install Grafana Plugin and Configure Data Source](../../third-party/grafana/#install-grafana-plugin-and-configure-data-source)
### Configure collectd

4
docs/en/25-application/_03-immigrate.md
View File

@ -70,7 +70,7 @@ You can use collectd and push the data to taosAdapter utilizing the write_tsdb p
- **Tuning the Dashboard system**
After writing the data to TDengine, you can configure Grafana to visualize the data written to TDengine. To obtain and use the Grafana plugin provided by TDengine, please refer to [Links to other tools](/third-party/grafana).
After writing the data to TDengine, you can configure Grafana to visualize the data written to TDengine. To obtain and use the Grafana plugin provided by TDengine, please refer to [Links to other tools](../../third-party/grafana).
TDengine provides two sets of Dashboard templates by default, and users only need to import the templates from the Grafana directory into Grafana to activate their use.
@ -396,7 +396,7 @@ Hard disk writing performance has little effect on TDengine. The TDengine writin
### Computational resource estimates
Due to the characteristics of IoT data, when the frequency of data generation is consistent, the writing process of TDengine maintains a relatively fixed amount of resource consumption (computing and storage). According to the [TDengine Operation and Maintenance Guide](/operation/) description, the system consumes less than 1 CPU core at 22,000 writes per second.
Due to the characteristics of IoT data, when the frequency of data generation is consistent, the writing process of TDengine maintains a relatively fixed amount of resource consumption (computing and storage). According to the [TDengine Operation and Maintenance Guide](../../operation/) description, the system consumes less than 1 CPU core at 22,000 writes per second.
In estimating the CPU resources consumed by the query, assuming that the application requires the database to provide 10,000 QPS, the CPU time consumed by each query is about 1 ms. The query provided by each core per second is 1,000 QPS, which satisfies 10,000 QPS. The query request requires at least 10 cores. For the system as a whole system to have less than 50% CPU load, the entire cluster needs twice as many cores i.e. 20 cores.

1
include/common/cos.h
View File

@ -43,6 +43,7 @@ int32_t s3GetObjectBlock(const char *object_name, int64_t offset, int64_t size,
int32_t s3GetObjectsByPrefix(const char *prefix, const char *path);
void s3EvictCache(const char *path, long object_size);
long s3Size(const char *object_name);
int32_t s3GetObjectToFile(const char *object_name, char *fileName);
#ifdef __cplusplus
}

17
include/common/tglobal.h
View File

@ -75,12 +75,15 @@ extern int32_t tsElectInterval;
extern int32_t tsHeartbeatInterval;
extern int32_t tsHeartbeatTimeout;
// vnode
extern int64_t tsVndCommitMaxIntervalMs;
// snode
extern int32_t tsRsyncPort;
extern char tsCheckpointBackupDir[];
extern char tsCheckpointBackupDir[];
// vnode checkpoint
extern char tsSnodeAddress[]; //127.0.0.1:873
extern char tsSnodeAddress[]; // 127.0.0.1:873
// mnode
extern int64_t tsMndSdbWriteDelta;
@ -104,8 +107,8 @@ extern int32_t tsMonitorMaxLogs;
extern bool tsMonitorComp;
// audit
extern bool tsEnableAudit;
extern bool tsEnableAuditCreateTable;
extern bool tsEnableAudit;
extern bool tsEnableAuditCreateTable;
// telem
extern bool tsEnableTelem;
@ -113,9 +116,9 @@ extern int32_t tsTelemInterval;
extern char tsTelemServer[];
extern uint16_t tsTelemPort;
extern bool tsEnableCrashReport;
extern char *tsTelemUri;
extern char *tsClientCrashReportUri;
extern char *tsSvrCrashReportUri;
extern char * tsTelemUri;
extern char * tsClientCrashReportUri;
extern char * tsSvrCrashReportUri;
// query buffer management
extern int32_t tsQueryBufferSize; // maximum allowed usage buffer size in MB for each data node during query processing

23
include/common/tmsgdef.h
View File

@ -67,7 +67,7 @@ enum {
enum { // WARN: new msg should be appended to segment tail
#endif
TD_NEW_MSG_SEG(TDMT_DND_MSG)
TD_NEW_MSG_SEG(TDMT_DND_MSG) // 0<<8
TD_DEF_MSG_TYPE(TDMT_DND_CREATE_MNODE, "dnode-create-mnode", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_DND_DROP_MNODE, "dnode-drop-mnode", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_DND_CREATE_QNODE, "dnode-create-qnode", NULL, NULL)
@ -87,7 +87,7 @@ enum { // WARN: new msg should be appended to segment tail
TD_DEF_MSG_TYPE(TDMT_DND_ALTER_VNODE_TYPE, "dnode-alter-vnode-type", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_DND_CHECK_VNODE_LEARNER_CATCHUP, "dnode-check-vnode-learner-catchup", NULL, NULL)
TD_NEW_MSG_SEG(TDMT_MND_MSG)
TD_NEW_MSG_SEG(TDMT_MND_MSG) // 1<<8
TD_DEF_MSG_TYPE(TDMT_MND_CONNECT, "connect", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_MND_CREATE_ACCT, "create-acct", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_MND_ALTER_ACCT, "alter-acct", NULL, NULL)
@ -186,6 +186,7 @@ enum { // WARN: new msg should be appended to segment tail
TD_DEF_MSG_TYPE(TDMT_MND_RESUME_STREAM, "resume-stream", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_MND_STREAM_CHECKPOINT_TIMER, "stream-checkpoint-tmr", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_MND_STREAM_BEGIN_CHECKPOINT, "stream-begin-checkpoint", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_MND_STREAM_CHECKPOINT_CANDIDITATE, "stream-checkpoint-remain", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_MND_STREAM_NODECHANGE_CHECK, "stream-nodechange-check", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_MND_TRIM_DB_TIMER, "trim-db-tmr", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_MND_GRANT_NOTIFY, "grant-notify", NULL, NULL)
@ -194,7 +195,7 @@ enum { // WARN: new msg should be appended to segment tail
TD_DEF_MSG_TYPE(TDMT_MND_VIEW_META, "view-meta", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_MND_MAX_MSG, "mnd-max", NULL, NULL)
TD_NEW_MSG_SEG(TDMT_VND_MSG)
TD_NEW_MSG_SEG(TDMT_VND_MSG) // 2<<8
TD_DEF_MSG_TYPE(TDMT_VND_SUBMIT, "submit", SSubmitReq, SSubmitRsp)
TD_DEF_MSG_TYPE(TDMT_VND_CREATE_TABLE, "create-table", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_VND_ALTER_TABLE, "alter-table", NULL, NULL)
@ -243,7 +244,7 @@ enum { // WARN: new msg should be appended to segment tail
TD_DEF_MSG_TYPE(TDMT_VND_DISABLE_WRITE, "vnode-disable-write", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_VND_MAX_MSG, "vnd-max", NULL, NULL)
TD_NEW_MSG_SEG(TDMT_SCH_MSG)
TD_NEW_MSG_SEG(TDMT_SCH_MSG) // 3<<8
TD_DEF_MSG_TYPE(TDMT_SCH_QUERY, "query", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_SCH_MERGE_QUERY, "merge-query", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_SCH_QUERY_CONTINUE, "query-continue", NULL, NULL)
@ -258,13 +259,13 @@ enum { // WARN: new msg should be appended to segment tail
TD_DEF_MSG_TYPE(TDMT_SCH_MAX_MSG, "sch-max", NULL, NULL)
TD_NEW_MSG_SEG(TDMT_STREAM_MSG)
TD_DEF_MSG_TYPE(TDMT_STREAM_TASK_DEPLOY, "stream-task-deploy", SStreamTaskDeployReq, SStreamTaskDeployRsp)
TD_NEW_MSG_SEG(TDMT_STREAM_MSG) //4 << 8
TD_DEF_MSG_TYPE(TDMT_STREAM_TASK_DEPLOY, "stream-task-deploy", SStreamTaskDeployReq, SStreamTaskDeployRsp) //1025 1026
TD_DEF_MSG_TYPE(TDMT_STREAM_TASK_DROP, "stream-task-drop", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_STREAM_TASK_RUN, "stream-task-run", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_STREAM_TASK_DISPATCH, "stream-task-dispatch", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_STREAM_UNUSED1, "stream-unused1", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_STREAM_RETRIEVE, "stream-retrieve", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_STREAM_RETRIEVE, "stream-retrieve", NULL, NULL) //1035 1036
TD_DEF_MSG_TYPE(TDMT_STREAM_TASK_CHECKPOINT_READY, "stream-checkpoint-ready", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_STREAM_TASK_REPORT_CHECKPOINT, "stream-report-checkpoint", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_STREAM_TASK_RESTORE_CHECKPOINT, "stream-restore-checkpoint", NULL, NULL)
@ -274,10 +275,10 @@ enum { // WARN: new msg should be appended to segment tail
TD_DEF_MSG_TYPE(TDMT_STREAM_HTASK_DROP, "stream-htask-drop", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_STREAM_MAX_MSG, "stream-max", NULL, NULL)
TD_NEW_MSG_SEG(TDMT_MON_MSG)
TD_NEW_MSG_SEG(TDMT_MON_MSG) //5 << 8
TD_DEF_MSG_TYPE(TDMT_MON_MAX_MSG, "monitor-max", NULL, NULL)
TD_NEW_MSG_SEG(TDMT_SYNC_MSG)
TD_NEW_MSG_SEG(TDMT_SYNC_MSG) //6 << 8
TD_DEF_MSG_TYPE(TDMT_SYNC_TIMEOUT, "sync-timer", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_SYNC_TIMEOUT_ELECTION, "sync-elect", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_SYNC_PING_REPLY, "sync-ping-reply", NULL, NULL) // no longer used
@ -308,7 +309,7 @@ enum { // WARN: new msg should be appended to segment tail
TD_DEF_MSG_TYPE(TDMT_SYNC_MAX_MSG, "sync-max", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_SYNC_FORCE_FOLLOWER, "sync-force-become-follower", NULL, NULL)
TD_NEW_MSG_SEG(TDMT_VND_STREAM_MSG)
TD_NEW_MSG_SEG(TDMT_VND_STREAM_MSG) //7 << 8
TD_DEF_MSG_TYPE(TDMT_VND_STREAM_SCAN_HISTORY, "vnode-stream-scan-history", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_VND_STREAM_SCAN_HISTORY_FINISH, "vnode-stream-scan-history-finish", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_VND_STREAM_CHECK_POINT_SOURCE, "vnode-stream-checkpoint-source", NULL, NULL)
@ -317,7 +318,7 @@ enum { // WARN: new msg should be appended to segment tail
TD_DEF_MSG_TYPE(TDMT_VND_STREAM_TASK_CHECK, "vnode-stream-task-check", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_VND_STREAM_MAX_MSG, "vnd-stream-max", NULL, NULL)
TD_NEW_MSG_SEG(TDMT_VND_TMQ_MSG)
TD_NEW_MSG_SEG(TDMT_VND_TMQ_MSG) //8 << 8
TD_DEF_MSG_TYPE(TDMT_VND_TMQ_SUBSCRIBE, "vnode-tmq-subscribe", SMqRebVgReq, SMqRebVgRsp)
TD_DEF_MSG_TYPE(TDMT_VND_TMQ_DELETE_SUB, "vnode-tmq-delete-sub", SMqVDeleteReq, SMqVDeleteRsp)
TD_DEF_MSG_TYPE(TDMT_VND_TMQ_COMMIT_OFFSET, "vnode-tmq-commit-offset", STqOffset, STqOffset)

18
include/dnode/vnode/stream.h
View File

@ -1,18 +0,0 @@
//
// Created by mingming wanng on 2023/11/15.
//
#ifndef TDENGINE_STREAM_H
#define TDENGINE_STREAM_H
#define STREAM_EXEC_EXTRACT_DATA_IN_WAL_ID (-1)
#define STREAM_EXEC_START_ALL_TASKS_ID (-2)
#define STREAM_EXEC_RESTART_ALL_TASKS_ID (-3)
typedef struct STaskUpdateEntry {
int64_t streamId;
int32_t taskId;
int32_t transId;
} STaskUpdateEntry;
#endif // TDENGINE_STREAM_H

36
include/dnode/vnode/tqCommon.h Normal file
View File

@ -0,0 +1,36 @@
/*
* Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
*
* This program is free software: you can use, redistribute, and/or modify
* it under the terms of the GNU Affero General Public License, version 3
* or later ("AGPL"), as published by the Free Software Foundation.
*
* This program is distributed in the hope that it will be useful, but WITHOUT
* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
* FITNESS FOR A PARTICULAR PURPOSE.
*
* You should have received a copy of the GNU Affero General Public License
* along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#ifndef TDENGINE_TQ_COMMON_H
#define TDENGINE_TQ_COMMON_H
// message process
int32_t tqStreamTaskStartAsync(SStreamMeta* pMeta, SMsgCb* cb, bool restart);
int32_t tqStreamTaskProcessUpdateReq(SStreamMeta* pMeta, SMsgCb* cb, SRpcMsg* pMsg, bool restored);
int32_t tqStreamTaskProcessDispatchReq(SStreamMeta* pMeta, SRpcMsg* pMsg);
int32_t tqStreamTaskProcessDispatchRsp(SStreamMeta* pMeta, SRpcMsg* pMsg);
int32_t tqStreamTaskProcessRetrieveReq(SStreamMeta* pMeta, SRpcMsg* pMsg);
int32_t tqStreamTaskProcessScanHistoryFinishReq(SStreamMeta* pMeta, SRpcMsg* pMsg);
int32_t tqStreamTaskProcessScanHistoryFinishRsp(SStreamMeta* pMeta, SRpcMsg* pMsg);
int32_t tqStreamTaskProcessCheckReq(SStreamMeta* pMeta, SRpcMsg* pMsg);
int32_t tqStreamTaskProcessCheckRsp(SStreamMeta* pMeta, SRpcMsg* pMsg, bool isLeader);
int32_t tqStreamTaskProcessCheckpointReadyMsg(SStreamMeta* pMeta, SRpcMsg* pMsg);
int32_t tqStreamTaskProcessDeployReq(SStreamMeta* pMeta, int64_t sversion, char* msg, int32_t msgLen, bool isLeader, bool restored);
int32_t tqStreamTaskProcessDropReq(SStreamMeta* pMeta, char* msg, int32_t msgLen);
int32_t tqStreamTaskProcessRunReq(SStreamMeta* pMeta, SRpcMsg* pMsg, bool isLeader);
int32_t startStreamTasks(SStreamMeta* pMeta);
int32_t resetStreamTaskStatus(SStreamMeta* pMeta);
#endif // TDENGINE_TQ_COMMON_H

1
include/libs/nodes/querynodes.h
View File

@ -449,6 +449,7 @@ typedef struct SVnodeModifyOpStmt {
SHashObj* pSubTableHashObj; // SHashObj<table_name, STableMeta*>
SHashObj* pTableNameHashObj; // set of table names for refreshing meta, sync mode
SHashObj* pDbFNameHashObj; // set of db names for refreshing meta, sync mode
SHashObj* pTableCxtHashObj; // temp SHashObj<tuid, STableDataCxt*> for single request
SArray* pVgDataBlocks; // SArray<SVgroupDataCxt*>
SVCreateTbReq* pCreateTblReq;
TdFilePtr fp;

3
include/libs/stream/streamState.h
View File

@ -35,6 +35,7 @@ int32_t streamStateBegin(SStreamState* pState);
int32_t streamStateCommit(SStreamState* pState);
void streamStateDestroy(SStreamState* pState, bool remove);
int32_t streamStateDeleteCheckPoint(SStreamState* pState, TSKEY mark);
int32_t streamStateDelTaskDb(SStreamState* pState);
int32_t streamStateFuncPut(SStreamState* pState, const SWinKey* key, const void* value, int32_t vLen);
int32_t streamStateFuncGet(SStreamState* pState, const SWinKey* key, void** ppVal, int32_t* pVLen);
@ -133,4 +134,4 @@ char* streamStateIntervalDump(SStreamState* pState);
}
#endif
#endif /* ifndef _STREAM_STATE_H_ */
#endif /* ifndef _STREAM_STATE_H_ */

46
include/libs/stream/tstream.h
View File

@ -50,11 +50,17 @@ extern "C" {
(_t)->hTaskInfo.id.streamId = 0; \
} while (0)
#define STREAM_EXEC_EXTRACT_DATA_IN_WAL_ID (-1)
#define STREAM_EXEC_START_ALL_TASKS_ID (-2)
#define STREAM_EXEC_RESTART_ALL_TASKS_ID (-3)
typedef struct SStreamTask SStreamTask;
typedef struct SStreamQueue SStreamQueue;
typedef struct SStreamTaskSM SStreamTaskSM;
#define SSTREAM_TASK_VER 2
#define SSTREAM_TASK_VER 2
#define SSTREAM_TASK_INCOMPATIBLE_VER 1
#define SSTREAM_TASK_NEED_CONVERT_VER 2
enum {
STREAM_STATUS__NORMAL = 0,
@ -106,6 +112,7 @@ typedef enum {
TASK_LEVEL__SOURCE = 1,
TASK_LEVEL__AGG,
TASK_LEVEL__SINK,
TASK_LEVEL_SMA,
} ETASK_LEVEL;
enum {
@ -300,11 +307,16 @@ typedef struct SStreamTaskId {
typedef struct SCheckpointInfo {
int64_t startTs;
int64_t checkpointId;
int64_t checkpointVer; // latest checkpointId version
int64_t processedVer; // already processed ver, that has generated results version.
int64_t checkpointVer; // latest checkpointId version
int64_t processedVer;
int64_t nextProcessVer; // current offset in WAL, not serialize it
int64_t failedId; // record the latest failed checkpoint id
int64_t checkpointingId;
int32_t downstreamAlignNum;
int32_t checkpointNotReadyTasks;
bool dispatchCheckpointTrigger;
int64_t msgVer;
} SCheckpointInfo;
typedef struct SStreamStatus {
@ -443,12 +455,11 @@ struct SStreamTask {
int64_t checkReqId;
SArray* checkReqIds; // shuffle
int32_t refCnt;
int64_t checkpointingId;
int32_t checkpointAlignCnt;
int32_t checkpointNotReadyTasks;
int32_t transferStateAlignCnt;
struct SStreamMeta* pMeta;
SSHashObj* pNameMap;
void* pBackend;
int64_t backendRefId;
char reserve[256];
};
@ -486,20 +497,25 @@ typedef struct SStreamMeta {
int32_t walScanCounter;
void* streamBackend;
int64_t streamBackendRid;
SHashObj* pTaskBackendUnique;
SHashObj* pTaskDbUnique;
TdThreadMutex backendMutex;
SMetaHbInfo* pHbInfo;
STaskUpdateInfo updateInfo;
SHashObj* pUpdateTaskSet;
int32_t numOfStreamTasks; // this value should be increased when a new task is added into the meta
int32_t numOfPausedTasks;
int32_t chkptNotReadyTasks;
int64_t rid;
int64_t chkpId;
int32_t chkpCap;
SArray* chkpSaved;
SArray* chkpInUse;
int32_t chkpCap;
SRWLatch chkpDirLock;
void* qHandle;
int32_t pauseTaskNum;
void* bkdChkptMgt;
} SStreamMeta;
int32_t tEncodeStreamEpInfo(SEncoder* pEncoder, const SStreamChildEpInfo* pInfo);
@ -655,7 +671,7 @@ int32_t tDecodeStreamCheckpointReadyMsg(SDecoder* pDecoder, SStreamCheckpointRea
typedef struct STaskStatusEntry {
STaskId id;
int32_t status;
int32_t statusLastDuration; // to record the last duration of current status
int32_t statusLastDuration; // to record the last duration of current status
int64_t stage;
int32_t nodeId;
int64_t verStart; // start version in WAL, only valid for source task
@ -668,8 +684,8 @@ typedef struct STaskStatusEntry {
int64_t inputQUnchangeCounter;
double inputQUsed; // in MiB
double inputRate;
double sinkQuota; // existed quota size for sink task
double sinkDataSize; // sink to dst data size
double sinkQuota; // existed quota size for sink task
double sinkDataSize; // sink to dst data size
} STaskStatusEntry;
typedef struct SStreamHbMsg {
@ -830,8 +846,10 @@ int32_t streamMetaReopen(SStreamMeta* pMeta);
void streamMetaInitBackend(SStreamMeta* pMeta);
int32_t streamMetaCommit(SStreamMeta* pMeta);
int32_t streamMetaLoadAllTasks(SStreamMeta* pMeta);
int32_t streamMetaReloadAllTasks(SStreamMeta* pMeta);
int64_t streamMetaGetLatestCheckpointId(SStreamMeta* pMeta);
void streamMetaNotifyClose(SStreamMeta* pMeta);
int32_t streamTaskSetDb(SStreamMeta* pMeta, void* pTask, char* key);
void streamMetaStartHb(SStreamMeta* pMeta);
bool streamMetaTaskInTimer(SStreamMeta* pMeta);
int32_t streamMetaUpdateTaskDownstreamStatus(SStreamTask* pTask, int64_t startTs, int64_t endTs, bool succ);
@ -853,8 +871,10 @@ int32_t streamAddCheckpointSourceRspMsg(SStreamCheckpointSourceReq* pReq, SRpcHa
int32_t buildCheckpointSourceRsp(SStreamCheckpointSourceReq* pReq, SRpcHandleInfo* pRpcInfo, SRpcMsg* pMsg,
int8_t isSucceed);
SStreamTaskSM* streamCreateStateMachine(SStreamTask* pTask);
void* streamDestroyStateMachine(SStreamTaskSM* pSM);
#ifdef __cplusplus
}
#endif
#endif /* ifndef _STREAM_H_ */
#endif /* ifndef _STREAM_H_ */

2
include/os/osFile.h
View File

@ -64,6 +64,8 @@ typedef struct TdFile *TdFilePtr;
#define TD_FILE_EXCL 0x0080
#define TD_FILE_STREAM 0x0100 // Only support taosFprintfFile, taosGetLineFile, taosEOFFile
#define TD_FILE_WRITE_THROUGH 0x0200
#define TD_FILE_CLOEXEC 0x0400
TdFilePtr taosOpenFile(const char *path, int32_t tdFileOptions);
TdFilePtr taosCreateFile(const char *path, int32_t tdFileOptions);

1
include/util/taoserror.h
View File

@ -124,6 +124,7 @@ int32_t* taosGetErrno();
#define TSDB_CODE_INVALID_CFG_VALUE TAOS_DEF_ERROR_CODE(0, 0x0133)
#define TSDB_CODE_IP_NOT_IN_WHITE_LIST TAOS_DEF_ERROR_CODE(0, 0x0134)
#define TSDB_CODE_FAILED_TO_CONNECT_S3 TAOS_DEF_ERROR_CODE(0, 0x0135)
//client
#define TSDB_CODE_TSC_INVALID_OPERATION TAOS_DEF_ERROR_CODE(0, 0x0200)

1
packaging/tools/set_core.sh
View File

@ -38,3 +38,4 @@ source /etc/profile
${csudo}mkdir -p ${corePath} ||:
${csudo}sysctl -w kernel.core_pattern=${corePath}/core-%e-%p ||:
${csudo}echo "${corePath}/core-%e-%p" | ${csudo}tee /proc/sys/kernel/core_pattern ||:
${csudo}echo "kernel.core_pattern = ${corePath}/core_%e-%p" >> /etc/sysctl.conf ||:

3
source/client/src/clientEnv.c
View File

@ -555,6 +555,9 @@ static void *tscCrashReportThreadFp(void *param) {
if (pFile) {
taosReleaseCrashLogFile(pFile, false);
pFile = NULL;
taosMsleep(sleepTime);
loopTimes = 0;
continue;
}
} else {

1
source/client/src/clientMain.c
View File

@ -876,6 +876,7 @@ int taos_get_current_db(TAOS *taos, char *database, int len, int *required) {
code = 0;
}
taosThreadMutexUnlock(&pTscObj->mutex);
releaseTscObj(*(int64_t *)taos);
return code;
}

14
source/client/src/clientTmq.c
View File

@ -1216,7 +1216,7 @@ int32_t tmq_subscribe(tmq_t* tmq, const tmq_list_t* topic_list) {
}
int32_t retryCnt = 0;
while (TSDB_CODE_MND_CONSUMER_NOT_READY == syncAskEp(tmq)) {
while (syncAskEp(tmq) != 0) {
if (retryCnt++ > MAX_RETRY_COUNT) {
tscError("consumer:0x%" PRIx64 ", mnd not ready for subscribe, retry more than 2 minutes", tmq->consumerId);
code = TSDB_CODE_MND_CONSUMER_NOT_READY;
@ -1454,6 +1454,9 @@ static bool doUpdateLocalEp(tmq_t* tmq, int32_t epoch, const SMqAskEpRsp* pRsp)
if (epoch < tmq->epoch || (epoch == tmq->epoch && topicNumGet == 0)) {
tscInfo("consumer:0x%" PRIx64 " no update ep epoch from %d to epoch %d, incoming topics:%d",
tmq->consumerId, tmq->epoch, epoch, topicNumGet);
if(atomic_load_8(&tmq->status) == TMQ_CONSUMER_STATUS__RECOVER){
atomic_store_8(&tmq->status, TMQ_CONSUMER_STATUS__READY);
}
return false;
}
@ -1965,9 +1968,14 @@ TAOS_RES* tmq_consumer_poll(tmq_t* tmq, int64_t timeout) {
return NULL;
}
while (atomic_load_8(&tmq->status) == TMQ_CONSUMER_STATUS__RECOVER) {
while (1) {
if(atomic_load_8(&tmq->status) != TMQ_CONSUMER_STATUS__RECOVER){
break;
}
tscInfo("consumer:0x%" PRIx64 " tmq status is recover", tmq->consumerId);
int32_t retryCnt = 0;
while (TSDB_CODE_MND_CONSUMER_NOT_READY == syncAskEp(tmq)) {
while (syncAskEp(tmq) != 0) {
if (retryCnt++ > 40) {
return NULL;
}

90
source/common/src/cos.c
View File

@ -86,7 +86,7 @@ typedef struct {
char err_msg[128];
S3Status status;
uint64_t content_length;
char *buf;
char * buf;
int64_t buf_pos;
} TS3SizeCBD;
@ -270,7 +270,7 @@ typedef struct list_parts_callback_data {
typedef struct MultipartPartData {
put_object_callback_data put_object_data;
int seq;
UploadManager *manager;
UploadManager * manager;
} MultipartPartData;
static int putObjectDataCallback(int bufferSize, char *buffer, void *callbackData) {
@ -317,7 +317,7 @@ S3Status MultipartResponseProperiesCallback(const S3ResponseProperties *properti
MultipartPartData *data = (MultipartPartData *)callbackData;
int seq = data->seq;
const char *etag = properties->eTag;
const char * etag = properties->eTag;
data->manager->etags[seq - 1] = strdup(etag);
data->manager->next_etags_pos = seq;
return S3StatusOK;
@ -450,10 +450,10 @@ int32_t s3PutObjectFromFile2(const char *file, const char *object) {
int32_t code = 0;
const char *key = object;
// const char *uploadId = 0;
const char *filename = 0;
const char * filename = 0;
uint64_t contentLength = 0;
const char *cacheControl = 0, *contentType = 0, *md5 = 0;
const char *contentDispositionFilename = 0, *contentEncoding = 0;
const char * cacheControl = 0, *contentType = 0, *md5 = 0;
const char * contentDispositionFilename = 0, *contentEncoding = 0;
int64_t expires = -1;
S3CannedAcl cannedAcl = S3CannedAclPrivate;
int metaPropertiesCount = 0;
@ -467,6 +467,7 @@ int32_t s3PutObjectFromFile2(const char *file, const char *object) {
// data.infileFD = NULL;
// data.noStatus = noStatus;
// uError("ERROR: %s stat file %s: ", __func__, file);
if (taosStatFile(file, &contentLength, NULL, NULL) < 0) {
uError("ERROR: %s Failed to stat file %s: ", __func__, file);
code = TAOS_SYSTEM_ERROR(errno);
@ -647,7 +648,7 @@ typedef struct list_bucket_callback_data {
char nextMarker[1024];
int keyCount;
int allDetails;
SArray *objectArray;
SArray * objectArray;
} list_bucket_callback_data;
static S3Status listBucketCallback(int isTruncated, const char *nextMarker, int contentsCount,
@ -692,11 +693,11 @@ static void s3FreeObjectKey(void *pItem) {
static SArray *getListByPrefix(const char *prefix) {
S3BucketContext bucketContext = {0, tsS3BucketName, protocolG, uriStyleG, tsS3AccessKeyId, tsS3AccessKeySecret,
0, awsRegionG};
0, awsRegionG};
S3ListBucketHandler listBucketHandler = {{&responsePropertiesCallbackNull, &responseCompleteCallback},
&listBucketCallback};
const char *marker = 0, *delimiter = 0;
const char * marker = 0, *delimiter = 0;
int maxkeys = 0, allDetails = 0;
list_bucket_callback_data data;
data.objectArray = taosArrayInit(32, sizeof(void *));
@ -737,7 +738,7 @@ static SArray *getListByPrefix(const char *prefix) {
void s3DeleteObjects(const char *object_name[], int nobject) {
S3BucketContext bucketContext = {0, tsS3BucketName, protocolG, uriStyleG, tsS3AccessKeyId, tsS3AccessKeySecret,
0, awsRegionG};
0, awsRegionG};
S3ResponseHandler responseHandler = {0, &responseCompleteCallback};
for (int i = 0; i < nobject; ++i) {
@ -788,7 +789,7 @@ int32_t s3GetObjectBlock(const char *object_name, int64_t offset, int64_t size,
const char *ifMatch = 0, *ifNotMatch = 0;
S3BucketContext bucketContext = {0, tsS3BucketName, protocolG, uriStyleG, tsS3AccessKeyId, tsS3AccessKeySecret,
0, awsRegionG};
0, awsRegionG};
S3GetConditions getConditions = {ifModifiedSince, ifNotModifiedSince, ifMatch, ifNotMatch};
S3GetObjectHandler getObjectHandler = {{&responsePropertiesCallback, &responseCompleteCallback},
&getObjectDataCallback};
@ -826,7 +827,7 @@ int32_t s3GetObjectToFile(const char *object_name, char *fileName) {
const char *ifMatch = 0, *ifNotMatch = 0;
S3BucketContext bucketContext = {0, tsS3BucketName, protocolG, uriStyleG, tsS3AccessKeyId, tsS3AccessKeySecret,
0, awsRegionG};
0, awsRegionG};
S3GetConditions getConditions = {ifModifiedSince, ifNotModifiedSince, ifMatch, ifNotMatch};
S3GetObjectHandler getObjectHandler = {{&responsePropertiesCallbackNull, &responseCompleteCallback},
&getObjectCallback};
@ -857,7 +858,7 @@ int32_t s3GetObjectsByPrefix(const char *prefix, const char *path) {
if (objectArray == NULL) return -1;
for (size_t i = 0; i < taosArrayGetSize(objectArray); i++) {
char *object = taosArrayGetP(objectArray, i);
char * object = taosArrayGetP(objectArray, i);
const char *tmp = strchr(object, '/');
tmp = (tmp == NULL) ? object : tmp + 1;
char fileName[PATH_MAX] = {0};
@ -891,6 +892,8 @@ long s3Size(const char *object_name) {
if ((cbd.status != S3StatusOK) && (cbd.status != S3StatusErrorPreconditionFailed)) {
s3PrintError(__FILE__, __LINE__, __func__, cbd.status, cbd.err_msg);
return -1;
}
size = cbd.content_length;
@ -946,12 +949,12 @@ static void s3InitRequestOptions(cos_request_options_t *options, int is_cname) {
int32_t s3PutObjectFromFile(const char *file_str, const char *object_str) {
int32_t code = 0;
cos_pool_t *p = NULL;
cos_pool_t * p = NULL;
int is_cname = 0;
cos_status_t *s = NULL;
cos_status_t * s = NULL;
cos_request_options_t *options = NULL;
cos_string_t bucket, object, file;
cos_table_t *resp_headers;
cos_table_t * resp_headers;
// int traffic_limit = 0;
cos_pool_create(&p, NULL);
@ -982,14 +985,14 @@ int32_t s3PutObjectFromFile(const char *file_str, const char *object_str) {
int32_t s3PutObjectFromFile2(const char *file_str, const char *object_str) {
int32_t code = 0;
cos_pool_t *p = NULL;
cos_pool_t * p = NULL;
int is_cname = 0;
cos_status_t *s = NULL;
cos_request_options_t *options = NULL;
cos_status_t * s = NULL;
cos_request_options_t * options = NULL;
cos_string_t bucket, object, file;
cos_table_t *resp_headers;
cos_table_t * resp_headers;
int traffic_limit = 0;
cos_table_t *headers = NULL;
cos_table_t * headers = NULL;
cos_resumable_clt_params_t *clt_params = NULL;
cos_pool_create(&p, NULL);
@ -1022,11 +1025,11 @@ int32_t s3PutObjectFromFile2(const char *file_str, const char *object_str) {
}
void s3DeleteObjectsByPrefix(const char *prefix_str) {
cos_pool_t *p = NULL;
cos_pool_t * p = NULL;
cos_request_options_t *options = NULL;
int is_cname = 0;
cos_string_t bucket;
cos_status_t *s = NULL;
cos_status_t * s = NULL;
cos_string_t prefix;
cos_pool_create(&p, NULL);
@ -1041,10 +1044,10 @@ void s3DeleteObjectsByPrefix(const char *prefix_str) {
}
void s3DeleteObjects(const char *object_name[], int nobject) {
cos_pool_t *p = NULL;
cos_pool_t * p = NULL;
int is_cname = 0;
cos_string_t bucket;
cos_table_t *resp_headers = NULL;
cos_table_t * resp_headers = NULL;
cos_request_options_t *options = NULL;
cos_list_t object_list;
cos_list_t deleted_object_list;
@ -1078,14 +1081,14 @@ void s3DeleteObjects(const char *object_name[], int nobject) {
bool s3Exists(const char *object_name) {
bool ret = false;
cos_pool_t *p = NULL;
cos_pool_t * p = NULL;
int is_cname = 0;
cos_status_t *s = NULL;
cos_request_options_t *options = NULL;
cos_status_t * s = NULL;
cos_request_options_t * options = NULL;
cos_string_t bucket;
cos_string_t object;
cos_table_t *resp_headers;
cos_table_t *headers = NULL;
cos_table_t * resp_headers;
cos_table_t * headers = NULL;
cos_object_exist_status_e object_exist;
cos_pool_create(&p, NULL);
@ -1112,15 +1115,15 @@ bool s3Exists(const char *object_name) {
bool s3Get(const char *object_name, const char *path) {
bool ret = false;
cos_pool_t *p = NULL;
cos_pool_t * p = NULL;
int is_cname = 0;
cos_status_t *s = NULL;
cos_status_t * s = NULL;
cos_request_options_t *options = NULL;
cos_string_t bucket;
cos_string_t object;
cos_string_t file;
cos_table_t *resp_headers = NULL;
cos_table_t *headers = NULL;
cos_table_t * resp_headers = NULL;
cos_table_t * headers = NULL;
int traffic_limit = 0;
//创建内存池
@ -1156,15 +1159,15 @@ bool s3Get(const char *object_name, const char *path) {
int32_t s3GetObjectBlock(const char *object_name, int64_t offset, int64_t block_size, bool check, uint8_t **ppBlock) {
(void)check;
int32_t code = 0;
cos_pool_t *p = NULL;
cos_pool_t * p = NULL;
int is_cname = 0;
cos_status_t *s = NULL;
cos_status_t * s = NULL;
cos_request_options_t *options = NULL;
cos_string_t bucket;
cos_string_t object;
cos_table_t *resp_headers;
cos_table_t *headers = NULL;
cos_buf_t *content = NULL;
cos_table_t * resp_headers;
cos_table_t * headers = NULL;
cos_buf_t * content = NULL;
// cos_string_t file;
// int traffic_limit = 0;
char range_buf[64];
@ -1258,7 +1261,7 @@ void s3EvictCache(const char *path, long object_size) {
terrno = TAOS_SYSTEM_ERROR(errno);
vError("failed to open %s since %s", dir_name, terrstr());
}
SArray *evict_files = taosArrayInit(16, sizeof(SEvictFile));
SArray * evict_files = taosArrayInit(16, sizeof(SEvictFile));
tdbDirEntryPtr pDirEntry;
while ((pDirEntry = taosReadDir(pDir)) != NULL) {
char *name = taosGetDirEntryName(pDirEntry);
@ -1300,13 +1303,13 @@ void s3EvictCache(const char *path, long object_size) {
long s3Size(const char *object_name) {
long size = 0;
cos_pool_t *p = NULL;
cos_pool_t * p = NULL;
int is_cname = 0;
cos_status_t *s = NULL;
cos_status_t * s = NULL;
cos_request_options_t *options = NULL;
cos_string_t bucket;
cos_string_t object;
cos_table_t *resp_headers = NULL;
cos_table_t * resp_headers = NULL;
//创建内存池
cos_pool_create(&p, NULL);
@ -1352,5 +1355,6 @@ int32_t s3GetObjectBlock(const char *object_name, int64_t offset, int64_t size,
void s3EvictCache(const char *path, long object_size) {}
long s3Size(const char *object_name) { return 0; }
int32_t s3GetObjectsByPrefix(const char *prefix, const char *path) { return 0; }
int32_t s3GetObjectToFile(const char *object_name, char *fileName) { return 0; }
#endif

2
source/common/src/systable.c
View File

@ -140,7 +140,7 @@ static const SSysDbTableSchema userStbsSchema[] = {
{.name = "columns", .bytes = 4, .type = TSDB_DATA_TYPE_INT, .sysInfo = false},
{.name = "tags", .bytes = 4, .type = TSDB_DATA_TYPE_INT, .sysInfo = false},
{.name = "last_update", .bytes = 8, .type = TSDB_DATA_TYPE_TIMESTAMP, .sysInfo = false},
{.name = "table_comment", .bytes = TSDB_TB_COMMENT_LEN + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_VARCHAR, .sysInfo = false},
{.name = "table_comment", .bytes = TSDB_TB_COMMENT_LEN - 1 + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_VARCHAR, .sysInfo = false},
{.name = "watermark", .bytes = 64 + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_VARCHAR, .sysInfo = false},
{.name = "max_delay", .bytes = 64 + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_VARCHAR, .sysInfo = false},
{.name = "rollup", .bytes = 128 + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_VARCHAR, .sysInfo = false},

21
source/common/src/tglobal.c
View File

@ -107,7 +107,7 @@ bool tsEnableTelem = true;
int32_t tsTelemInterval = 43200;
char tsTelemServer[TSDB_FQDN_LEN] = "telemetry.tdengine.com";
uint16_t tsTelemPort = 80;
char *tsTelemUri = "/report";
char * tsTelemUri = "/report";
#ifdef TD_ENTERPRISE
bool tsEnableCrashReport = false;
@ -354,16 +354,24 @@ static int32_t taosLoadCfg(SConfig *pCfg, const char **envCmd, const char *input
char cfgFile[PATH_MAX + 100] = {0};
taosExpandDir(inputCfgDir, cfgDir, PATH_MAX);
char lastC = cfgDir[strlen(cfgDir) - 1];
char lastC = cfgDir[strlen(cfgDir) - 1];
char *tdDirsep = TD_DIRSEP;
if (lastC == '\\' || lastC == '/') {
tdDirsep = "";
}
if (taosIsDir(cfgDir)) {
#ifdef CUS_PROMPT
snprintf(cfgFile, sizeof(cfgFile), "%s" "%s" "%s.cfg", cfgDir, tdDirsep, CUS_PROMPT);
snprintf(cfgFile, sizeof(cfgFile),
"%s"
"%s"
"%s.cfg",
cfgDir, tdDirsep, CUS_PROMPT);
#else
snprintf(cfgFile, sizeof(cfgFile), "%s" "%s" "taos.cfg", cfgDir, tdDirsep);
snprintf(cfgFile, sizeof(cfgFile),
"%s"
"%s"
"taos.cfg",
cfgDir, tdDirsep);
#endif
} else {
tstrncpy(cfgFile, cfgDir, sizeof(cfgDir));
@ -729,6 +737,7 @@ static int32_t taosAddServerCfg(SConfig *pCfg) {
if (cfgAddBool(pCfg, "disableStream", tsDisableStream, CFG_SCOPE_SERVER, CFG_DYN_ENT_SERVER) != 0) return -1;
if (cfgAddInt64(pCfg, "streamBufferSize", tsStreamBufferSize, 0, INT64_MAX, CFG_SCOPE_SERVER, CFG_DYN_NONE) != 0)
return -1;
if (cfgAddInt32(pCfg, "checkpointInterval", tsStreamCheckpointInterval, 60, 1200, CFG_SCOPE_SERVER,
CFG_DYN_ENT_SERVER) != 0)
return -1;
@ -1347,7 +1356,7 @@ void taosCleanupCfg() {
typedef struct {
const char *optionName;
void *optionVar;
void * optionVar;
} OptionNameAndVar;
static int32_t taosCfgSetOption(OptionNameAndVar *pOptions, int32_t optionSize, SConfigItem *pItem, bool isDebugflag) {
@ -1360,7 +1369,7 @@ static int32_t taosCfgSetOption(OptionNameAndVar *pOptions, int32_t optionSize,
switch (pItem->dtype) {
case CFG_DTYPE_BOOL: {
int32_t flag = pItem->i32;
bool *pVar = pOptions[d].optionVar;
bool * pVar = pOptions[d].optionVar;
uInfo("%s set from %d to %d", optName, *pVar, flag);
*pVar = flag;
terrno = TSDB_CODE_SUCCESS;

3
source/dnode/mgmt/mgmt_dnode/src/dmWorker.c
View File

@ -127,6 +127,9 @@ static void *dmCrashReportThreadFp(void *param) {
if (pFile) {
taosReleaseCrashLogFile(pFile, false);
pFile = NULL;
taosMsleep(sleepTime);
loopTimes = 0;
continue;
}
} else {

6
source/dnode/mgmt/mgmt_snode/src/smWorker.c
View File

@ -193,10 +193,6 @@ int32_t smPutNodeMsgToStreamQueue(SSnodeMgmt *pMgmt, SRpcMsg *pMsg) {
SSingleWorker *pWorker = &pMgmt->streamWorker;
dTrace("msg:%p, put into worker %s", pMsg, pWorker->name);
if (pMsg->msgType == TDMT_STREAM_TASK_DISPATCH) {
sndEnqueueStreamDispatch(pMgmt->pSnode, pMsg);
} else {
taosWriteQitem(pWorker->queue, pMsg);
}
taosWriteQitem(pWorker->queue, pMsg);
return 0;
}

4
source/dnode/mgmt/mgmt_vnode/src/vmHandle.c
View File

@ -281,8 +281,8 @@ int32_t vmProcessCreateVnodeReq(SVnodeMgmt *pMgmt, SRpcMsg *pMsg) {
vmGenerateWrapperCfg(pMgmt, &req, &wrapperCfg);
SVnodeObj *pVnode = vmAcquireVnode(pMgmt, req.vgId);
if (pVnode != NULL) {
SVnodeObj *pVnode = vmAcquireVnodeImpl(pMgmt, req.vgId, false);
if (pVnode != NULL && (req.replica == 1 || !pVnode->failed)) {
dError("vgId:%d, already exist", req.vgId);
tFreeSCreateVnodeReq(&req);
vmReleaseVnode(pMgmt, pVnode);

6
source/dnode/mgmt/mgmt_vnode/src/vmWorker.c
View File

@ -206,11 +206,7 @@ static int32_t vmPutMsgToQueue(SVnodeMgmt *pMgmt, SRpcMsg *pMsg, EQueueType qtyp
break;
case STREAM_QUEUE:
dGTrace("vgId:%d, msg:%p put into vnode-stream queue", pVnode->vgId, pMsg);
if (pMsg->msgType == TDMT_STREAM_TASK_DISPATCH) {
vnodeEnqueueStreamMsg(pVnode->pImpl, pMsg);
} else {
taosWriteQitem(pVnode->pStreamQ, pMsg);
}
taosWriteQitem(pVnode->pStreamQ, pMsg);
break;
case FETCH_QUEUE:
dGTrace("vgId:%d, msg:%p put into vnode-fetch queue", pVnode->vgId, pMsg);

2
source/dnode/mgmt/node_util/src/dmFile.c
View File

@ -149,7 +149,7 @@ TdFilePtr dmCheckRunning(const char *dataDir) {
char filepath[PATH_MAX] = {0};
snprintf(filepath, sizeof(filepath), "%s%s.running", dataDir, TD_DIRSEP);
TdFilePtr pFile = taosOpenFile(filepath, TD_FILE_CREATE | TD_FILE_WRITE | TD_FILE_TRUNC);
TdFilePtr pFile = taosOpenFile(filepath, TD_FILE_CREATE | TD_FILE_WRITE | TD_FILE_TRUNC | TD_FILE_CLOEXEC);
if (pFile == NULL) {
terrno = TAOS_SYSTEM_ERROR(errno);
dError("failed to open file:%s since %s", filepath, terrstr());

8
source/dnode/mnode/impl/inc/mndDef.h
View File

@ -700,6 +700,11 @@ typedef struct {
} SStreamObj;
typedef struct SStreamSeq {
char name[24];
uint64_t seq;
SRWLatch lock;
} SStreamSeq;
int32_t tEncodeSStreamObj(SEncoder* pEncoder, const SStreamObj* pObj);
int32_t tDecodeSStreamObj(SDecoder* pDecoder, SStreamObj* pObj, int32_t sver);
void tFreeStreamObj(SStreamObj* pObj);
@ -731,14 +736,13 @@ typedef struct {
int8_t type;
int32_t numOfCols;
SSchema* pSchema;
SRWLatch lock;
SRWLatch lock;
} SViewObj;
int32_t tEncodeSViewObj(SEncoder* pEncoder, const SViewObj* pObj);
int32_t tDecodeSViewObj(SDecoder* pDecoder, SViewObj* pObj, int32_t sver);
void tFreeSViewObj(SViewObj* pObj);
#ifdef __cplusplus
}
#endif

27
source/dnode/mnode/impl/inc/mndStream.h
View File

@ -28,18 +28,28 @@ typedef struct SStreamTransInfo {
const char *name;
} SStreamTransInfo;
// time to generated the checkpoint, if now() - checkpointTs >= tsCheckpointInterval, this checkpoint will be discard
// to avoid too many checkpoints for a taskk in the waiting list
typedef struct SCheckpointCandEntry {
char * pName;
int64_t streamId;
int64_t checkpointTs;
int64_t checkpointId;
} SCheckpointCandEntry;
typedef struct SStreamTransMgmt {
SHashObj *pDBTrans;
SHashObj *pWaitingList; // stream id list, of which timed checkpoint failed to be issued due to the trans conflict.
} SStreamTransMgmt;
typedef struct SStreamExecInfo {
SArray *pNodeList;
int64_t ts; // snapshot ts
SStreamTransMgmt transMgmt;
int64_t activeCheckpoint; // active check point id
SHashObj * pTaskMap;
SArray * pTaskList;
TdThreadMutex lock;
SArray * pNodeList;
int64_t ts; // snapshot ts
SStreamTransMgmt transMgmt;
int64_t activeCheckpoint; // active check point id
SHashObj * pTaskMap;
SArray * pTaskList;
TdThreadMutex lock;
} SStreamExecInfo;
extern SStreamExecInfo execInfo;
@ -51,7 +61,8 @@ void mndReleaseStream(SMnode *pMnode, SStreamObj *pStream);
int32_t mndDropStreamByDb(SMnode *pMnode, STrans *pTrans, SDbObj *pDb);
int32_t mndPersistStream(SMnode *pMnode, STrans *pTrans, SStreamObj *pStream);
int32_t mndStreamRegisterTrans(STrans* pTrans, const char* pName, const char* pSrcDb, const char* pDstDb);
int32_t mndStreamRegisterTrans(STrans *pTrans, const char *pName, const char *pSrcDb, const char *pDstDb);
int32_t mndAddtoCheckpointWaitingList(SStreamObj *pStream, int64_t checkpointId);
bool streamTransConflictOtherTrans(SMnode *pMnode, const char *pSrcDb, const char *pDstDb, bool lock);
// for sma

13
source/dnode/mnode/impl/src/mndMain.c
View File

@ -146,6 +146,15 @@ static void mndStreamCheckpointTick(SMnode *pMnode, int64_t sec) {
}
}
static void mndStreamCheckpointRemain(SMnode* pMnode) {
int32_t contLen = 0;
void *pReq = mndBuildCheckpointTickMsg(&contLen, 0);
if (pReq != NULL) {
SRpcMsg rpcMsg = {.msgType = TDMT_MND_STREAM_CHECKPOINT_CANDIDITATE, .pCont = pReq, .contLen = contLen};
tmsgPutToQueue(&pMnode->msgCb, READ_QUEUE, &rpcMsg);
}
}
static void mndStreamCheckNode(SMnode* pMnode) {
int32_t contLen = 0;
void *pReq = mndBuildTimerMsg(&contLen);
@ -286,6 +295,10 @@ static void *mndThreadFp(void *param) {
mndStreamCheckpointTick(pMnode, sec);
}
if (sec % 5 == 0) {
mndStreamCheckpointRemain(pMnode);
}
if (sec % tsStreamNodeCheckInterval == 0) {
mndStreamCheckNode(pMnode);
}

34
source/dnode/mnode/impl/src/mndStb.c
View File

@ -1014,6 +1014,20 @@ static int32_t mndFindSuperTableColumnIndex(const SStbObj *pStb, const char *col
return -1;
}
static bool mndValidateSchema(SSchema *pSchemas, int32_t nSchema, SArray *pFields, int32_t maxLen) {
int32_t rowLen = 0;
for (int32_t i = 0; i < nSchema; ++i) {
rowLen += (pSchemas + i)->bytes;
}
int32_t nField = taosArrayGetSize(pFields);
for (int32_t i = 0; i < nField; ++i) {
rowLen += ((SField *)TARRAY_GET_ELEM(pFields, i))->bytes;
}
return rowLen <= maxLen;
}
static int32_t mndBuildStbFromAlter(SStbObj *pStb, SStbObj *pDst, SMCreateStbReq *createReq) {
taosRLockLatch(&pStb->lock);
memcpy(pDst, pStb, sizeof(SStbObj));
@ -1269,6 +1283,11 @@ static int32_t mndAddSuperTableTag(const SStbObj *pOld, SStbObj *pNew, SArray *p
return -1;
}
if (!mndValidateSchema(pOld->pTags, pOld->numOfTags, pFields, TSDB_MAX_TAGS_LEN)) {
terrno = TSDB_CODE_PAR_INVALID_TAGS_LENGTH;
return -1;
}
pNew->numOfTags = pNew->numOfTags + ntags;
if (mndAllocStbSchemas(pOld, pNew) != 0) {
return -1;
@ -1558,6 +1577,16 @@ static int32_t mndAlterStbTagBytes(SMnode *pMnode, const SStbObj *pOld, SStbObj
return -1;
}
uint32_t nLen = 0;
for (int32_t i = 0; i < pOld->numOfTags; ++i) {
nLen += (pOld->pTags[i].colId == colId) ? pField->bytes : pOld->pTags[i].bytes;
}
if (nLen > TSDB_MAX_TAGS_LEN) {
terrno = TSDB_CODE_PAR_INVALID_TAGS_LENGTH;
return -1;
}
if (mndAllocStbSchemas(pOld, pNew) != 0) {
return -1;
}
@ -1592,6 +1621,11 @@ static int32_t mndAddSuperTableColumn(const SStbObj *pOld, SStbObj *pNew, SArray
return -1;
}
if (!mndValidateSchema(pOld->pColumns, pOld->numOfColumns, pFields, TSDB_MAX_BYTES_PER_ROW)) {
terrno = TSDB_CODE_PAR_INVALID_ROW_LENGTH;
return -1;
}
pNew->numOfColumns = pNew->numOfColumns + ncols;
if (mndAllocStbSchemas(pOld, pNew) != 0) {
return -1;

545
source/dnode/mnode/impl/src/mndStream.c
View File

@ -21,19 +21,18 @@
#include "mndPrivilege.h"
#include "mndScheduler.h"
#include "mndShow.h"
#include "mndSnode.h"
#include "mndStb.h"
#include "mndTopic.h"
#include "mndTrans.h"
#include "mndUser.h"
#include "mndVgroup.h"
#include "osMemory.h"
#include "parser.h"
#include "tmisce.h"
#include "tname.h"
#define MND_STREAM_VER_NUMBER 4
#define MND_STREAM_RESERVE_SIZE 64
#define MND_STREAM_MAX_NUM 60
#define MND_STREAM_VER_NUMBER 4
#define MND_STREAM_RESERVE_SIZE 64
#define MND_STREAM_MAX_NUM 60
#define MND_STREAM_CHECKPOINT_NAME "stream-checkpoint"
#define MND_STREAM_PAUSE_NAME "stream-pause"
@ -64,6 +63,7 @@ static int32_t mndProcessCreateStreamReq(SRpcMsg *pReq);
static int32_t mndProcessDropStreamReq(SRpcMsg *pReq);
static int32_t mndProcessStreamCheckpointTmr(SRpcMsg *pReq);
static int32_t mndProcessStreamDoCheckpoint(SRpcMsg *pReq);
static int32_t mndProcessStreamCheckpointInCandid(SRpcMsg *pReq);
static int32_t mndProcessStreamHb(SRpcMsg *pReq);
static int32_t mndRetrieveStream(SRpcMsg *pReq, SShowObj *pShow, SSDataBlock *pBlock, int32_t rows);
static void mndCancelGetNextStream(SMnode *pMnode, void *pIter);
@ -82,7 +82,7 @@ static SStreamObj *mndGetStreamObj(SMnode *pMnode, int64_t streamId);
static SVgroupChangeInfo mndFindChangedNodeInfo(SMnode *pMnode, const SArray *pPrevNodeList, const SArray *pNodeList);
static STrans *doCreateTrans(SMnode *pMnode, SStreamObj *pStream, SRpcMsg *pReq, const char *name, const char* pMsg);
static STrans *doCreateTrans(SMnode *pMnode, SStreamObj *pStream, SRpcMsg *pReq, const char *name, const char *pMsg);
static int32_t mndPersistTransLog(SStreamObj *pStream, STrans *pTrans);
static void initTransAction(STransAction *pAction, void *pCont, int32_t contLen, int32_t msgType, const SEpSet *pEpset,
int32_t retryCode);
@ -90,12 +90,19 @@ static int32_t createStreamUpdateTrans(SStreamObj *pStream, SVgroupChangeInfo *p
static void removeStreamTasksInBuf(SStreamObj *pStream, SStreamExecInfo *pExecNode);
static void saveStreamTasksInfo(SStreamObj *pStream, SStreamExecInfo *pExecNode);
static int32_t removeExpirednodeEntryAndTask(SArray *pNodeSnapshot);
static int32_t killActiveCheckpointTrans(SMnode *pMnode, const char* pDbName, size_t len);
static int32_t killActiveCheckpointTrans(SMnode *pMnode, const char *pDbName, size_t len);
static int32_t setNodeEpsetExpiredFlag(const SArray *pNodeList);
static void freeCheckpointCandEntry(void *);
static SSdbRaw *mndStreamActionEncode(SStreamObj *pStream);
static SSdbRow *mndStreamActionDecode(SSdbRaw *pRaw);
SSdbRaw * mndStreamSeqActionEncode(SStreamObj *pStream);
SSdbRow * mndStreamSeqActionDecode(SSdbRaw *pRaw);
static int32_t mndStreamSeqActionInsert(SSdb *pSdb, SStreamSeq *pStream);
static int32_t mndStreamSeqActionDelete(SSdb *pSdb, SStreamSeq *pStream);
static int32_t mndStreamSeqActionUpdate(SSdb *pSdb, SStreamSeq *pOldStream, SStreamSeq *pNewStream);
int32_t mndInitStream(SMnode *pMnode) {
SSdbTable table = {
.sdbType = SDB_STREAM,
@ -106,6 +113,15 @@ int32_t mndInitStream(SMnode *pMnode) {
.updateFp = (SdbUpdateFp)mndStreamActionUpdate,
.deleteFp = (SdbDeleteFp)mndStreamActionDelete,
};
SSdbTable tableSeq = {
.sdbType = SDB_STREAM_SEQ,
.keyType = SDB_KEY_BINARY,
.encodeFp = (SdbEncodeFp)mndStreamSeqActionEncode,
.decodeFp = (SdbDecodeFp)mndStreamSeqActionDecode,
.insertFp = (SdbInsertFp)mndStreamSeqActionInsert,
.updateFp = (SdbUpdateFp)mndStreamSeqActionUpdate,
.deleteFp = (SdbDeleteFp)mndStreamSeqActionDelete,
};
mndSetMsgHandle(pMnode, TDMT_MND_CREATE_STREAM, mndProcessCreateStreamReq);
mndSetMsgHandle(pMnode, TDMT_MND_DROP_STREAM, mndProcessDropStreamReq);
@ -122,6 +138,7 @@ int32_t mndInitStream(SMnode *pMnode) {
mndSetMsgHandle(pMnode, TDMT_VND_STREAM_CHECK_POINT_SOURCE_RSP, mndTransProcessRsp);
mndSetMsgHandle(pMnode, TDMT_MND_STREAM_CHECKPOINT_TIMER, mndProcessStreamCheckpointTmr);
mndSetMsgHandle(pMnode, TDMT_MND_STREAM_BEGIN_CHECKPOINT, mndProcessStreamDoCheckpoint);
mndSetMsgHandle(pMnode, TDMT_MND_STREAM_CHECKPOINT_CANDIDITATE, mndProcessStreamCheckpointInCandid);
mndSetMsgHandle(pMnode, TDMT_MND_STREAM_HEARTBEAT, mndProcessStreamHb);
mndSetMsgHandle(pMnode, TDMT_STREAM_TASK_REPORT_CHECKPOINT, mndTransProcessRsp);
mndSetMsgHandle(pMnode, TDMT_MND_STREAM_NODECHANGE_CHECK, mndProcessNodeCheckReq);
@ -140,14 +157,23 @@ int32_t mndInitStream(SMnode *pMnode) {
execInfo.pTaskList = taosArrayInit(4, sizeof(STaskId));
execInfo.pTaskMap = taosHashInit(64, fn, true, HASH_NO_LOCK);
execInfo.transMgmt.pDBTrans = taosHashInit(32, fn, true, HASH_NO_LOCK);
execInfo.transMgmt.pWaitingList = taosHashInit(32, fn, true, HASH_NO_LOCK);
taosHashSetFreeFp(execInfo.transMgmt.pWaitingList, freeCheckpointCandEntry);
return sdbSetTable(pMnode->pSdb, table);
if (sdbSetTable(pMnode->pSdb, table) != 0) {
return -1;
}
if (sdbSetTable(pMnode->pSdb, tableSeq) != 0) {
return -1;
}
return 0;
}
void mndCleanupStream(SMnode *pMnode) {
taosArrayDestroy(execInfo.pTaskList);
taosHashCleanup(execInfo.pTaskMap);
taosHashCleanup(execInfo.transMgmt.pDBTrans);
taosHashCleanup(execInfo.transMgmt.pWaitingList);
taosThreadMutexDestroy(&execInfo.lock);
mDebug("mnd stream exec info cleanup");
}
@ -194,7 +220,8 @@ STREAM_ENCODE_OVER:
return NULL;
}
mTrace("stream:%s, encode to raw:%p, row:%p", pStream->name, pRaw, pStream);
mTrace("stream:%s, encode to raw:%p, row:%p, checkpoint:%" PRId64 "", pStream->name, pRaw, pStream,
pStream->checkpointId);
return pRaw;
}
@ -247,7 +274,8 @@ STREAM_DECODE_OVER:
return NULL;
}
mTrace("stream:%s, decode from raw:%p, row:%p", pStream->name, pRaw, pStream);
mTrace("stream:%s, decode from raw:%p, row:%p, checkpoint:%" PRId64 "", pStream->name, pRaw, pStream,
pStream->checkpointId);
return pRow;
}
@ -273,6 +301,8 @@ static int32_t mndStreamActionUpdate(SSdb *pSdb, SStreamObj *pOldStream, SStream
pOldStream->status = pNewStream->status;
pOldStream->updateTime = pNewStream->updateTime;
pOldStream->checkpointId = pNewStream->checkpointId;
pOldStream->checkpointFreq = pNewStream->checkpointFreq;
taosWUnLockLatch(&pOldStream->lock);
return 0;
@ -307,6 +337,12 @@ static void mndShowStreamStatus(char *dst, SStreamObj *pStream) {
}
}
SSdbRaw *mndStreamSeqActionEncode(SStreamObj *pStream) { return NULL; }
SSdbRow *mndStreamSeqActionDecode(SSdbRaw *pRaw) { return NULL; }
int32_t mndStreamSeqActionInsert(SSdb *pSdb, SStreamSeq *pStream) { return 0; }
int32_t mndStreamSeqActionDelete(SSdb *pSdb, SStreamSeq *pStream) { return 0; }
int32_t mndStreamSeqActionUpdate(SSdb *pSdb, SStreamSeq *pOldStream, SStreamSeq *pNewStream) { return 0; }
static void mndShowStreamTrigger(char *dst, SStreamObj *pStream) {
int8_t trigger = pStream->conf.trigger;
if (trigger == STREAM_TRIGGER_AT_ONCE) {
@ -685,7 +721,7 @@ _OVER:
return -1;
}
static int32_t mndPersistTaskDropReq(SMnode* pMnode, STrans *pTrans, SStreamTask *pTask) {
static int32_t mndPersistTaskDropReq(SMnode *pMnode, STrans *pTrans, SStreamTask *pTask) {
SVDropStreamTaskReq *pReq = taosMemoryCalloc(1, sizeof(SVDropStreamTaskReq));
if (pReq == NULL) {
terrno = TSDB_CODE_OUT_OF_MEMORY;
@ -697,9 +733,15 @@ static int32_t mndPersistTaskDropReq(SMnode* pMnode, STrans *pTrans, SStreamTask
pReq->streamId = pTask->id.streamId;
STransAction action = {0};
SVgObj *pVgObj = mndAcquireVgroup(pMnode, pTask->info.nodeId);
SEpSet epset = mndGetVgroupEpset(pMnode, pVgObj);
mndReleaseVgroup(pMnode, pVgObj);
SEpSet epset = {0};
if (pTask->info.nodeId == SNODE_HANDLE) {
SSnodeObj *pObj = mndAcquireSnode(pMnode, pTask->info.nodeId);
addEpIntoEpSet(&epset, pObj->pDnode->fqdn, pObj->pDnode->port);
} else {
SVgObj *pVgObj = mndAcquireVgroup(pMnode, pTask->info.nodeId);
epset = mndGetVgroupEpset(pMnode, pVgObj);
mndReleaseVgroup(pMnode, pVgObj);
}
// The epset of nodeId of this task may have been expired now, let's use the newest epset from mnode.
initTransAction(&action, pReq, sizeof(SVDropStreamTaskReq), TDMT_STREAM_TASK_DROP, &epset, 0);
@ -729,7 +771,7 @@ int32_t mndDropStreamTasks(SMnode *pMnode, STrans *pTrans, SStreamObj *pStream)
static int32_t checkForNumOfStreams(SMnode *pMnode, SStreamObj *pStreamObj) { // check for number of existed tasks
int32_t numOfStream = 0;
SStreamObj *pStream = NULL;
void *pIter = NULL;
void * pIter = NULL;
while ((pIter = sdbFetch(pMnode->pSdb, SDB_STREAM, pIter, (void **)&pStream)) != NULL) {
if (pStream->sourceDbUid == pStreamObj->sourceDbUid) {
@ -760,7 +802,7 @@ static int32_t mndProcessCreateStreamReq(SRpcMsg *pReq) {
SMnode *pMnode = pReq->info.node;
SStreamObj *pStream = NULL;
SStreamObj streamObj = {0};
char *sql = NULL;
char * sql = NULL;
int32_t sqlLen = 0;
terrno = TSDB_CODE_SUCCESS;
@ -910,8 +952,11 @@ int64_t mndStreamGenChkpId(SMnode *pMnode) {
if (pIter == NULL) break;
maxChkpId = TMAX(maxChkpId, pStream->checkpointId);
mDebug("stream %p checkpoint %" PRId64 "", pStream, pStream->checkpointId);
sdbRelease(pSdb, pStream);
}
mDebug("generated checkpoint %" PRId64 "", maxChkpId + 1);
return maxChkpId + 1;
}
@ -931,6 +976,22 @@ static int32_t mndProcessStreamCheckpointTmr(SRpcMsg *pReq) {
return 0;
}
static int32_t mndProcessStreamRemainChkptTmr(SRpcMsg *pReq) {
SMnode *pMnode = pReq->info.node;
SSdb * pSdb = pMnode->pSdb;
if (sdbGetSize(pSdb, SDB_STREAM) <= 0) {
return 0;
}
SMStreamDoCheckpointMsg *pMsg = rpcMallocCont(sizeof(SMStreamDoCheckpointMsg));
pMsg->checkpointId = 0;
int32_t size = sizeof(SMStreamDoCheckpointMsg);
SRpcMsg rpcMsg = {.msgType = TDMT_MND_STREAM_CHECKPOINT_CANDIDITATE, .pCont = pMsg, .contLen = size};
tmsgPutToQueue(&pMnode->msgCb, WRITE_QUEUE, &rpcMsg);
return 0;
}
static int32_t mndBuildStreamCheckpointSourceReq2(void **pBuf, int32_t *pLen, int32_t nodeId, int64_t checkpointId,
int64_t streamId, int32_t taskId) {
SStreamCheckpointSourceReq req = {0};
@ -973,107 +1034,104 @@ static int32_t mndBuildStreamCheckpointSourceReq2(void **pBuf, int32_t *pLen, in
return 0;
}
// static int32_t mndProcessStreamCheckpointTrans(SMnode *pMnode, SStreamObj *pStream, int64_t checkpointId) {
// int64_t timestampMs = taosGetTimestampMs();
// if (timestampMs - pStream->checkpointFreq < tsStreamCheckpointInterval * 1000) {
// return -1;
// }
// STrans *pTrans = mndTransCreate(pMnode, TRN_POLICY_RETRY, TRN_CONFLICT_DB_INSIDE, NULL, "stream-checkpoint");
// if (pTrans == NULL) return -1;
// mndTransSetDbName(pTrans, pStream->sourceDb, pStream->targetDb);
// if (mndTrancCheckConflict(pMnode, pTrans) != 0) {
// mError("failed to checkpoint of stream name%s, checkpointId: %" PRId64 ", reason:%s", pStream->name,
// checkpointId,
// tstrerror(TSDB_CODE_MND_TRANS_CONFLICT));
// mndTransDrop(pTrans);
// return -1;
// }
// mDebug("start to trigger checkpoint for stream:%s, checkpoint: %" PRId64 "", pStream->name, checkpointId);
// atomic_store_64(&pStream->currentTick, 1);
// taosWLockLatch(&pStream->lock);
// // 1. redo action: broadcast checkpoint source msg for all source vg
// int32_t totLevel = taosArrayGetSize(pStream->tasks);
// for (int32_t i = 0; i < totLevel; i++) {
// SArray *pLevel = taosArrayGetP(pStream->tasks, i);
// SStreamTask *pTask = taosArrayGetP(pLevel, 0);
// if (pTask->info.taskLevel == TASK_LEVEL__SOURCE) {
// int32_t sz = taosArrayGetSize(pLevel);
// for (int32_t j = 0; j < sz; j++) {
// SStreamTask *pTask = taosArrayGetP(pLevel, j);
// /*A(pTask->info.nodeId > 0);*/
// SVgObj *pVgObj = mndAcquireVgroup(pMnode, pTask->info.nodeId);
// if (pVgObj == NULL) {
// taosWUnLockLatch(&pStream->lock);
// mndTransDrop(pTrans);
// return -1;
// }
static int32_t mndProcessStreamCheckpointTrans(SMnode *pMnode, SStreamObj *pStream, int64_t checkpointId) {
int32_t code = -1;
int64_t timestampMs = taosGetTimestampMs();
if (timestampMs - pStream->checkpointFreq < tsStreamCheckpointInterval * 1000) {
return -1;
}
// void *buf;
// int32_t tlen;
// if (mndBuildStreamCheckpointSourceReq2(&buf, &tlen, pTask->info.nodeId, checkpointId, pTask->id.streamId,
// pTask->id.taskId) < 0) {
// mndReleaseVgroup(pMnode, pVgObj);
// taosWUnLockLatch(&pStream->lock);
// mndTransDrop(pTrans);
// return -1;
// }
bool conflict = streamTransConflictOtherTrans(pMnode, pStream->sourceDb, pStream->targetDb, true);
if (conflict) {
mndAddtoCheckpointWaitingList(pStream, checkpointId);
mWarn("checkpoint conflict with other trans in %s, ignore the checkpoint for stream:%s %" PRIx64, pStream->sourceDb,
pStream->name, pStream->uid);
return -1;
}
// STransAction action = {0};
// action.epSet = mndGetVgroupEpset(pMnode, pVgObj);
// action.pCont = buf;
// action.contLen = tlen;
// action.msgType = TDMT_VND_STREAM_CHECK_POINT_SOURCE;
STrans *pTrans = mndTransCreate(pMnode, TRN_POLICY_RETRY, TRN_CONFLICT_DB_INSIDE, NULL, MND_STREAM_CHECKPOINT_NAME);
if (pTrans == NULL) {
return -1;
}
// mndReleaseVgroup(pMnode, pVgObj);
mndStreamRegisterTrans(pTrans, MND_STREAM_CHECKPOINT_NAME, pStream->sourceDb, pStream->targetDb);
// if (mndTransAppendRedoAction(pTrans, &action) != 0) {
// taosMemoryFree(buf);
// taosWUnLockLatch(&pStream->lock);
// mndReleaseStream(pMnode, pStream);
// mndTransDrop(pTrans);
// return -1;
// }
// }
// }
// }
// // 2. reset tick
// pStream->checkpointFreq = checkpointId;
// pStream->checkpointId = checkpointId;
// pStream->checkpointFreq = taosGetTimestampMs();
// atomic_store_64(&pStream->currentTick, 0);
// // 3. commit log: stream checkpoint info
// pStream->version = pStream->version + 1;
// taosWUnLockLatch(&pStream->lock);
mndTransSetDbName(pTrans, pStream->sourceDb, pStream->targetDb);
if (mndTrancCheckConflict(pMnode, pTrans) != 0) {
mError("failed to checkpoint of stream name%s, checkpointId: %" PRId64 ", reason:%s", pStream->name, checkpointId,
tstrerror(TSDB_CODE_MND_TRANS_CONFLICT));
goto _ERR;
}
// // // code condtion
mDebug("start to trigger checkpoint for stream:%s, checkpoint: %" PRId64 "", pStream->name, checkpointId);
// SSdbRaw *pCommitRaw = mndStreamActionEncode(pStream);
// if (pCommitRaw == NULL) {
// mError("failed to prepare trans rebalance since %s", terrstr());
// goto _ERR;
// }
// if (mndTransAppendCommitlog(pTrans, pCommitRaw) != 0) {
// sdbFreeRaw(pCommitRaw);
// mError("failed to prepare trans rebalance since %s", terrstr());
// goto _ERR;
// }
// if (sdbSetRawStatus(pCommitRaw, SDB_STATUS_READY) != 0) {
// sdbFreeRaw(pCommitRaw);
// mError("failed to prepare trans rebalance since %s", terrstr());
// goto _ERR;
// }
taosWLockLatch(&pStream->lock);
pStream->currentTick = 1;
// if (mndTransPrepare(pMnode, pTrans) != 0) {
// mError("failed to prepare trans rebalance since %s", terrstr());
// goto _ERR;
// }
// mndTransDrop(pTrans);
// return 0;
// _ERR:
// mndTransDrop(pTrans);
// return -1;
// }
// 1. redo action: broadcast checkpoint source msg for all source vg
int32_t totLevel = taosArrayGetSize(pStream->tasks);
for (int32_t i = 0; i < totLevel; i++) {
SArray * pLevel = taosArrayGetP(pStream->tasks, i);
SStreamTask *p = taosArrayGetP(pLevel, 0);
if (p->info.taskLevel == TASK_LEVEL__SOURCE) {
int32_t sz = taosArrayGetSize(pLevel);
for (int32_t j = 0; j < sz; j++) {
SStreamTask *pTask = taosArrayGetP(pLevel, j);
SVgObj *pVgObj = mndAcquireVgroup(pMnode, pTask->info.nodeId);
if (pVgObj == NULL) {
taosWUnLockLatch(&pStream->lock);
goto _ERR;
}
void * buf;
int32_t tlen;
if (mndBuildStreamCheckpointSourceReq2(&buf, &tlen, pTask->info.nodeId, checkpointId, pTask->id.streamId,
pTask->id.taskId) < 0) {
mndReleaseVgroup(pMnode, pVgObj);
taosWUnLockLatch(&pStream->lock);
goto _ERR;
}
STransAction act = {0};
SEpSet epset = mndGetVgroupEpset(pMnode, pVgObj);
mndReleaseVgroup(pMnode, pVgObj);
initTransAction(&act, buf, tlen, TDMT_VND_STREAM_CHECK_POINT_SOURCE, &epset, TSDB_CODE_SYN_PROPOSE_NOT_READY);
if (mndTransAppendRedoAction(pTrans, &act) != 0) {
taosMemoryFree(buf);
taosWUnLockLatch(&pStream->lock);
goto _ERR;
}
}
}
}
// 2. reset tick
pStream->checkpointId = checkpointId;
pStream->checkpointFreq = taosGetTimestampMs();
pStream->currentTick = 0;
// 3. commit log: stream checkpoint info
pStream->version = pStream->version + 1;
taosWUnLockLatch(&pStream->lock);
if ((code = mndPersistTransLog(pStream, pTrans)) != TSDB_CODE_SUCCESS) {
return code;
}
if ((code = mndTransPrepare(pMnode, pTrans)) != TSDB_CODE_SUCCESS) {
mError("failed to prepare trans rebalance since %s", terrstr());
goto _ERR;
}
code = 0;
_ERR:
mndTransDrop(pTrans);
return code;
}
static int32_t mndAddStreamCheckpointToTrans(STrans *pTrans, SStreamObj *pStream, SMnode *pMnode, int64_t chkptId) {
taosWLockLatch(&pStream->lock);
@ -1149,23 +1207,7 @@ static int32_t mndAddStreamCheckpointToTrans(STrans *pTrans, SStreamObj *pStream
return 0;
}
static const char *mndGetStreamDB(SMnode *pMnode) {
SSdb * pSdb = pMnode->pSdb;
SStreamObj *pStream = NULL;
void * pIter = NULL;
pIter = sdbFetch(pSdb, SDB_STREAM, pIter, (void **)&pStream);
if (pIter == NULL) {
return NULL;
}
const char *p = taosStrdup(pStream->sourceDb);
mndReleaseStream(pMnode, pStream);
sdbCancelFetch(pSdb, pIter);
return p;
}
static int32_t initStreamNodeList(SMnode* pMnode) {
static int32_t initStreamNodeList(SMnode *pMnode) {
if (execInfo.pNodeList == NULL || (taosArrayGetSize(execInfo.pNodeList) == 0)) {
execInfo.pNodeList = taosArrayDestroy(execInfo.pNodeList);
execInfo.pNodeList = extractNodeListFromStream(pMnode);
@ -1174,11 +1216,11 @@ static int32_t initStreamNodeList(SMnode* pMnode) {
return taosArrayGetSize(execInfo.pNodeList);
}
static bool taskNodeIsUpdated(SMnode* pMnode) {
static bool taskNodeIsUpdated(SMnode *pMnode) {
// check if the node update happens or not
taosThreadMutexLock(&execInfo.lock);
int32_t numOfNodes = initStreamNodeList(pMnode);
int32_t numOfNodes = initStreamNodeList(pMnode);
if (numOfNodes == 0) {
mDebug("stream task node change checking done, no vgroups exist, do nothing");
execInfo.ts = taosGetTimestampSec();
@ -1218,6 +1260,38 @@ static bool taskNodeIsUpdated(SMnode* pMnode) {
return nodeUpdated;
}
static int32_t mndCheckNodeStatus(SMnode *pMnode) {
bool ready = true;
int64_t ts = taosGetTimestampSec();
if (taskNodeIsUpdated(pMnode)) {
return -1;
}
taosThreadMutexLock(&execInfo.lock);
if (taosArrayGetSize(execInfo.pNodeList) == 0) {
mDebug("stream task node change checking done, no vgroups exist, do nothing");
execInfo.ts = ts;
}
for (int32_t i = 0; i < taosArrayGetSize(execInfo.pTaskList); ++i) {
STaskId * p = taosArrayGet(execInfo.pTaskList, i);
STaskStatusEntry *pEntry = taosHashGet(execInfo.pTaskMap, p, sizeof(*p));
if (pEntry == NULL) {
continue;
}
if (pEntry->status != TASK_STATUS__READY) {
mDebug("s-task:0x%" PRIx64 "-0x%x (nodeId:%d) status:%s not ready, checkpoint msg not issued",
pEntry->id.streamId, (int32_t)pEntry->id.taskId, 0, streamTaskGetStatusStr(pEntry->status));
ready = false;
break;
}
}
taosThreadMutexUnlock(&execInfo.lock);
return ready ? 0 : -1;
}
static int32_t mndProcessStreamDoCheckpoint(SRpcMsg *pReq) {
SMnode * pMnode = pReq->info.node;
SSdb * pSdb = pMnode->pSdb;
@ -1225,90 +1299,65 @@ static int32_t mndProcessStreamDoCheckpoint(SRpcMsg *pReq) {
SStreamObj *pStream = NULL;
int32_t code = 0;
// check if the node update happens or not
bool updated = taskNodeIsUpdated(pMnode);
if (updated) {
mWarn("checkpoint ignore, stream task nodes update detected");
return -1;
}
{ // check if all tasks are in TASK_STATUS__READY status
bool ready = true;
taosThreadMutexLock(&execInfo.lock);
// no streams exists, abort
int32_t numOfTasks = taosArrayGetSize(execInfo.pTaskList);
if (numOfTasks <= 0) {
taosThreadMutexUnlock(&execInfo.lock);
return 0;
}
for (int32_t i = 0; i < taosArrayGetSize(execInfo.pTaskList); ++i) {
STaskId * p = taosArrayGet(execInfo.pTaskList, i);
STaskStatusEntry *pEntry = taosHashGet(execInfo.pTaskMap, p, sizeof(*p));
if (pEntry == NULL) {
continue;
}
if (pEntry->status != TASK_STATUS__READY) {
mDebug("s-task:0x%" PRIx64 "-0x%x (nodeId:%d) status:%s not ready, checkpoint msg not issued",
pEntry->id.streamId, (int32_t)pEntry->id.taskId, 0, streamTaskGetStatusStr(pEntry->status));
ready = false;
break;
}
}
taosThreadMutexUnlock(&execInfo.lock);
if (!ready) {
return 0;
}
if ((code = mndCheckNodeStatus(pMnode)) != 0) {
return code;
}
SMStreamDoCheckpointMsg *pMsg = (SMStreamDoCheckpointMsg *)pReq->pCont;
int64_t checkpointId = pMsg->checkpointId;
STrans *pTrans = mndTransCreate(pMnode, TRN_POLICY_RETRY, TRN_CONFLICT_DB_INSIDE, NULL, MND_STREAM_CHECKPOINT_NAME);
if (pTrans == NULL) {
mError("failed to trigger checkpoint, reason: %s", tstrerror(TSDB_CODE_OUT_OF_MEMORY));
return -1;
}
mDebug("start to trigger checkpoint, checkpointId: %" PRId64, checkpointId);
const char *pDb = mndGetStreamDB(pMnode);
mndTransSetDbName(pTrans, pDb, pDb);
mndStreamRegisterTrans(pTrans, MND_STREAM_CHECKPOINT_NAME, pDb, pDb);
taosMemoryFree((void *)pDb);
if (mndTransCheckConflict(pMnode, pTrans) != 0) {
mError("failed to trigger checkpoint, checkpointId: %" PRId64 ", reason:%s", checkpointId,
tstrerror(TSDB_CODE_MND_TRANS_CONFLICT));
mndTransDrop(pTrans);
return -1;
}
while (1) {
pIter = sdbFetch(pSdb, SDB_STREAM, pIter, (void **)&pStream);
if (pIter == NULL) break;
code = mndAddStreamCheckpointToTrans(pTrans, pStream, pMnode, checkpointId);
while ((pIter = sdbFetch(pSdb, SDB_STREAM, pIter, (void **)&pStream)) != NULL) {
code = mndProcessStreamCheckpointTrans(pMnode, pStream, pMsg->checkpointId);
sdbRelease(pSdb, pStream);
if (code == -1) {
break;
}
}
if (code == 0) {
if (mndTransPrepare(pMnode, pTrans) != 0) {
mError("failed to prepare trans rebalance since %s", terrstr());
return code;
}
static int32_t mndProcessStreamCheckpointInCandid(SRpcMsg *pReq) {
SMnode *pMnode = pReq->info.node;
void * pIter = NULL;
int32_t code = 0;
taosThreadMutexLock(&execInfo.lock);
int32_t num = taosHashGetSize(execInfo.transMgmt.pWaitingList);
taosThreadMutexUnlock(&execInfo.lock);
if (num == 0) {
return code;
}
if ((code = mndCheckNodeStatus(pMnode)) != 0) {
return code;
}
SArray *pList = taosArrayInit(4, sizeof(int64_t));
while ((pIter = taosHashIterate(execInfo.transMgmt.pWaitingList, pIter)) != NULL) {
SCheckpointCandEntry *pEntry = pIter;
SStreamObj *ps = mndAcquireStream(pMnode, pEntry->pName);
if (ps == NULL) {
continue;
}
mDebug("start to launch checkpoint for stream:%s %" PRIx64 " in candidate list", pEntry->pName, pEntry->streamId);
code = mndProcessStreamCheckpointTrans(pMnode, ps, pEntry->checkpointId);
mndReleaseStream(pMnode, ps);
if (code == TSDB_CODE_SUCCESS) {
taosArrayPush(pList, &pEntry->streamId);
}
}
mndTransDrop(pTrans);
for (int32_t i = 0; i < taosArrayGetSize(pList); ++i) {
int64_t *pId = taosArrayGet(pList, i);
// only one trans here
taosThreadMutexLock(&execInfo.lock);
execInfo.activeCheckpoint = checkpointId;
taosThreadMutexUnlock(&execInfo.lock);
taosHashRemove(execInfo.transMgmt.pWaitingList, pId, sizeof(*pId));
}
int32_t remain = taosHashGetSize(execInfo.transMgmt.pWaitingList);
mDebug("%d in candidate list generated checkpoint, remaining:%d", (int32_t)taosArrayGetSize(pList), remain);
taosArrayDestroy(pList);
return code;
}
@ -1546,7 +1595,7 @@ static int32_t mndRetrieveStream(SRpcMsg *pReq, SShowObj *pShow, SSDataBlock *pB
char dstStr[20] = {0};
STR_TO_VARSTR(dstStr, sinkQuota)
pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
colDataSetVal(pColInfo, numOfRows, (const char*) dstStr, false);
colDataSetVal(pColInfo, numOfRows, (const char *)dstStr, false);
char scanHistoryIdle[20 + VARSTR_HEADER_SIZE] = {0};
strcpy(scanHistoryIdle, "100a");
@ -1554,7 +1603,7 @@ static int32_t mndRetrieveStream(SRpcMsg *pReq, SShowObj *pShow, SSDataBlock *pB
memset(dstStr, 0, tListLen(dstStr));
STR_TO_VARSTR(dstStr, scanHistoryIdle)
pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
colDataSetVal(pColInfo, numOfRows, (const char*) dstStr, false);
colDataSetVal(pColInfo, numOfRows, (const char *)dstStr, false);
numOfRows++;
sdbRelease(pSdb, pStream);
@ -1569,7 +1618,7 @@ static void mndCancelGetNextStream(SMnode *pMnode, void *pIter) {
sdbCancelFetch(pSdb, pIter);
}
static void setTaskAttrInResBlock(SStreamObj* pStream, SStreamTask* pTask, SSDataBlock* pBlock, int32_t numOfRows) {
static void setTaskAttrInResBlock(SStreamObj *pStream, SStreamTask *pTask, SSDataBlock *pBlock, int32_t numOfRows) {
SColumnInfoData *pColInfo;
int32_t cols = 0;
@ -1623,7 +1672,7 @@ static void setTaskAttrInResBlock(SStreamObj* pStream, SStreamTask* pTask, SSDat
colDataSetVal(pColInfo, numOfRows, (const char *)level, false);
// status
char status[20 + VARSTR_HEADER_SIZE] = {0};
char status[20 + VARSTR_HEADER_SIZE] = {0};
STaskId id = {.streamId = pTask->id.streamId, .taskId = pTask->id.taskId};
STaskStatusEntry *pe = taosHashGet(execInfo.pTaskMap, &id, sizeof(id));
@ -1674,7 +1723,7 @@ static void setTaskAttrInResBlock(SStreamObj* pStream, SStreamTask* pTask, SSDat
colDataSetVal(pColInfo, numOfRows, (const char *)vbuf, false);
}
static int32_t getNumOfTasks(SArray* pTaskList) {
static int32_t getNumOfTasks(SArray *pTaskList) {
int32_t numOfLevels = taosArrayGetSize(pTaskList);
int32_t count = 0;
@ -1732,7 +1781,7 @@ static void mndCancelGetNextStreamTask(SMnode *pMnode, void *pIter) {
sdbCancelFetch(pSdb, pIter);
}
static int32_t mndPauseStreamTask(SMnode* pMnode, STrans *pTrans, SStreamTask *pTask) {
static int32_t mndPauseStreamTask(SMnode *pMnode, STrans *pTrans, SStreamTask *pTask) {
SVPauseStreamTaskReq *pReq = taosMemoryCalloc(1, sizeof(SVPauseStreamTaskReq));
if (pReq == NULL) {
mError("failed to malloc in pause stream, size:%" PRIzu ", code:%s", sizeof(SVPauseStreamTaskReq),
@ -1758,7 +1807,7 @@ static int32_t mndPauseStreamTask(SMnode* pMnode, STrans *pTrans, SStreamTask *p
return 0;
}
int32_t mndPauseAllStreamTasks(SMnode* pMnode, STrans *pTrans, SStreamObj *pStream) {
int32_t mndPauseAllStreamTasks(SMnode *pMnode, STrans *pTrans, SStreamObj *pStream) {
SArray *tasks = pStream->tasks;
int32_t size = taosArrayGetSize(tasks);
@ -1886,7 +1935,7 @@ static int32_t mndProcessPauseStreamReq(SRpcMsg *pReq) {
return TSDB_CODE_ACTION_IN_PROGRESS;
}
static int32_t mndResumeStreamTask(STrans *pTrans, SMnode* pMnode, SStreamTask *pTask, int8_t igUntreated) {
static int32_t mndResumeStreamTask(STrans *pTrans, SMnode *pMnode, SStreamTask *pTask, int8_t igUntreated) {
SVResumeStreamTaskReq *pReq = taosMemoryCalloc(1, sizeof(SVResumeStreamTaskReq));
if (pReq == NULL) {
terrno = TSDB_CODE_OUT_OF_MEMORY;
@ -1910,7 +1959,7 @@ static int32_t mndResumeStreamTask(STrans *pTrans, SMnode* pMnode, SStreamTask *
return 0;
}
int32_t mndResumeAllStreamTasks(STrans *pTrans, SMnode* pMnode, SStreamObj *pStream, int8_t igUntreated) {
int32_t mndResumeAllStreamTasks(STrans *pTrans, SMnode *pMnode, SStreamObj *pStream, int8_t igUntreated) {
int32_t size = taosArrayGetSize(pStream->tasks);
for (int32_t i = 0; i < size; i++) {
SArray *pTasks = taosArrayGetP(pStream->tasks, i);
@ -2147,8 +2196,8 @@ static bool isNodeEpsetChanged(const SEpSet *pPrevEpset, const SEpSet *pCurrent)
// 1. increase the replica does not affect the stream process.
// 2. decreasing the replica may affect the stream task execution in the way that there is one or more running stream
// tasks on the will be removed replica.
// 3. vgroup redistribution is an combination operation of first increase replica and then decrease replica. So we will
// handle it as mentioned in 1 & 2 items.
// 3. vgroup redistribution is an combination operation of first increase replica and then decrease replica. So we
// will handle it as mentioned in 1 & 2 items.
static SVgroupChangeInfo mndFindChangedNodeInfo(SMnode *pMnode, const SArray *pPrevNodeList, const SArray *pNodeList) {
SVgroupChangeInfo info = {
.pUpdateNodeList = taosArrayInit(4, sizeof(SNodeUpdateInfo)),
@ -2176,10 +2225,8 @@ static SVgroupChangeInfo mndFindChangedNodeInfo(SMnode *pMnode, const SArray *pP
epsetAssign(&updateInfo.prevEp, &pPrevEntry->epset);
epsetAssign(&updateInfo.newEp, &pCurrent->epset);
taosArrayPush(info.pUpdateNodeList, &updateInfo);
}
if(pCurrent->nodeId != SNODE_HANDLE){
if (pCurrent->nodeId != SNODE_HANDLE) {
SVgObj *pVgroup = mndAcquireVgroup(pMnode, pCurrent->nodeId);
taosHashPut(info.pDBMap, pVgroup->dbName, strlen(pVgroup->dbName), NULL, 0);
mndReleaseVgroup(pMnode, pVgroup);
@ -2392,7 +2439,6 @@ static int32_t doRemoveTasks(SStreamExecInfo *pExecNode, STaskId *pRemovedId) {
if (p == NULL) {
return TSDB_CODE_SUCCESS;
}
taosHashRemove(pExecNode->pTaskMap, pRemovedId, sizeof(*pRemovedId));
for (int32_t k = 0; k < taosArrayGetSize(pExecNode->pTaskList); ++k) {
@ -2430,7 +2476,7 @@ int32_t removeExpirednodeEntryAndTask(SArray *pNodeSnapshot) {
STaskId * pId = taosArrayGet(execInfo.pTaskList, i);
STaskStatusEntry *pEntry = taosHashGet(execInfo.pTaskMap, pId, sizeof(*pId));
if(pEntry->nodeId == SNODE_HANDLE) continue;
if (pEntry->nodeId == SNODE_HANDLE) continue;
bool existed = taskNodeExists(pNodeSnapshot, pEntry->nodeId);
if (!existed) {
@ -2447,9 +2493,9 @@ int32_t removeExpirednodeEntryAndTask(SArray *pNodeSnapshot) {
(int32_t)taosArrayGetSize(execInfo.pTaskList));
int32_t size = taosArrayGetSize(pNodeSnapshot);
SArray* pValidNodeEntryList = taosArrayInit(4, sizeof(SNodeEntry));
for(int32_t i = 0; i < taosArrayGetSize(execInfo.pNodeList); ++i) {
SNodeEntry* p = taosArrayGet(execInfo.pNodeList, i);
SArray *pValidNodeEntryList = taosArrayInit(4, sizeof(SNodeEntry));
for (int32_t i = 0; i < taosArrayGetSize(execInfo.pNodeList); ++i) {
SNodeEntry *p = taosArrayGet(execInfo.pNodeList, i);
for (int32_t j = 0; j < size; ++j) {
SNodeEntry *pEntry = taosArrayGet(pNodeSnapshot, j);
@ -2468,13 +2514,14 @@ int32_t removeExpirednodeEntryAndTask(SArray *pNodeSnapshot) {
return 0;
}
static void killAllCheckpointTrans(SMnode* pMnode, SVgroupChangeInfo* pChangeInfo) {
void* pIter = NULL;
while((pIter = taosHashIterate(pChangeInfo->pDBMap, pIter)) != NULL) {
char* pDb = (char*) pIter;
// kill all trans in the dst DB
static void killAllCheckpointTrans(SMnode *pMnode, SVgroupChangeInfo *pChangeInfo) {
void *pIter = NULL;
while ((pIter = taosHashIterate(pChangeInfo->pDBMap, pIter)) != NULL) {
char *pDb = (char *)pIter;
size_t len = 0;
void* pKey = taosHashGetKey(pDb, &len);
void * pKey = taosHashGetKey(pDb, &len);
killActiveCheckpointTrans(pMnode, pKey, len);
}
}
@ -2623,7 +2670,7 @@ void removeStreamTasksInBuf(SStreamObj *pStream, SStreamExecInfo *pExecNode) {
ASSERT(taosHashGetSize(pExecNode->pTaskMap) == taosArrayGetSize(pExecNode->pTaskList));
}
STrans *doCreateTrans(SMnode *pMnode, SStreamObj *pStream, SRpcMsg *pReq, const char *name, const char* pMsg) {
STrans *doCreateTrans(SMnode *pMnode, SStreamObj *pStream, SRpcMsg *pReq, const char *name, const char *pMsg) {
STrans *pTrans = mndTransCreate(pMnode, TRN_POLICY_RETRY, TRN_CONFLICT_DB_INSIDE, pReq, name);
if (pTrans == NULL) {
mError("failed to build trans:%s, reason: %s", name, tstrerror(TSDB_CODE_OUT_OF_MEMORY));
@ -2631,7 +2678,7 @@ STrans *doCreateTrans(SMnode *pMnode, SStreamObj *pStream, SRpcMsg *pReq, const
return NULL;
}
mDebug("s-task:0x%"PRIx64" start to build trans %s", pStream->uid, pMsg);
mDebug("s-task:0x%" PRIx64 " start to build trans %s", pStream->uid, pMsg);
mndTransSetDbName(pTrans, pStream->sourceDb, pStream->targetDb);
if (mndTransCheckConflict(pMnode, pTrans) != 0) {
@ -2667,6 +2714,7 @@ int32_t createStreamResetStatusTrans(SMnode *pMnode, SStreamObj *pStream) {
terrno = TSDB_CODE_OUT_OF_MEMORY;
mError("failed to malloc in reset stream, size:%" PRIzu ", code:%s", sizeof(SVResetStreamTaskReq),
tstrerror(TSDB_CODE_OUT_OF_MEMORY));
taosWUnLockLatch(&pStream->lock);
return terrno;
}
@ -2710,9 +2758,9 @@ int32_t createStreamResetStatusTrans(SMnode *pMnode, SStreamObj *pStream) {
return TSDB_CODE_ACTION_IN_PROGRESS;
}
int32_t killActiveCheckpointTrans(SMnode *pMnode, const char* pDBName, size_t len) {
int32_t killActiveCheckpointTrans(SMnode *pMnode, const char *pDBName, size_t len) {
// data in the hash table will be removed automatically, no need to remove it here.
SStreamTransInfo* pTransInfo = taosHashGet(execInfo.transMgmt.pDBTrans, pDBName, len);
SStreamTransInfo *pTransInfo = taosHashGet(execInfo.transMgmt.pDBTrans, pDBName, len);
if (pTransInfo == NULL) {
return TSDB_CODE_SUCCESS;
}
@ -2723,7 +2771,7 @@ int32_t killActiveCheckpointTrans(SMnode *pMnode, const char* pDBName, size_t le
return TSDB_CODE_SUCCESS;
}
STrans* pTrans = mndAcquireTrans(pMnode, pTransInfo->transId);
STrans *pTrans = mndAcquireTrans(pMnode, pTransInfo->transId);
if (pTrans != NULL) {
mInfo("kill checkpoint transId:%d in Db:%s", pTransInfo->transId, pDBName);
mndKillTrans(pMnode, pTrans);
@ -2734,7 +2782,7 @@ int32_t killActiveCheckpointTrans(SMnode *pMnode, const char* pDBName, size_t le
}
static int32_t mndResetStatusFromCheckpoint(SMnode *pMnode, int32_t transId) {
STrans* pTrans = mndAcquireTrans(pMnode, transId);
STrans *pTrans = mndAcquireTrans(pMnode, transId);
if (pTrans != NULL) {
mInfo("kill checkpoint transId:%d to reset task status", transId);
mndKillTrans(pMnode, pTrans);
@ -2753,8 +2801,8 @@ static int32_t mndResetStatusFromCheckpoint(SMnode *pMnode, int32_t transId) {
bool conflict = streamTransConflictOtherTrans(pMnode, pStream->sourceDb, pStream->targetDb, false);
if (conflict) {
mError("stream:%s other trans exists in DB:%s & %s failed to start reset-status trans",
pStream->name, pStream->sourceDb, pStream->targetDb);
mError("stream:%s other trans exists in DB:%s & %s failed to start reset-status trans", pStream->name,
pStream->sourceDb, pStream->targetDb);
continue;
}
@ -2765,11 +2813,10 @@ static int32_t mndResetStatusFromCheckpoint(SMnode *pMnode, int32_t transId) {
return code;
}
}
return 0;
}
static SStreamTask* mndGetStreamTask(STaskId* pId, SStreamObj* pStream) {
static SStreamTask *mndGetStreamTask(STaskId *pId, SStreamObj *pStream) {
for (int32_t i = 0; i < taosArrayGetSize(pStream->tasks); i++) {
SArray *pLevel = taosArrayGetP(pStream->tasks, i);
@ -2824,7 +2871,7 @@ static bool needDropRelatedFillhistoryTask(STaskStatusEntry *pTaskEntry, SStream
static int32_t mndDropRelatedFillhistoryTask(SMnode *pMnode, STaskStatusEntry *pTaskEntry, SStreamObj *pStream) {
SStreamTask *pTask = mndGetStreamTask(&pTaskEntry->id, pStream);
if (pTask == NULL) {
mError("failed to get the stream task:0x%x, may have been dropped", (int32_t) pTaskEntry->id.taskId);
mError("failed to get the stream task:0x%x, may have been dropped", (int32_t)pTaskEntry->id.taskId);
return -1;
}
@ -2859,12 +2906,12 @@ int32_t setNodeEpsetExpiredFlag(const SArray *pNodeList) {
mInfo("set node expired for %d nodes", num);
for (int k = 0; k < num; ++k) {
int32_t* pVgId = taosArrayGet(pNodeList, k);
int32_t *pVgId = taosArrayGet(pNodeList, k);
mInfo("set node expired for nodeId:%d, total:%d", *pVgId, num);
int32_t numOfNodes = taosArrayGetSize(execInfo.pNodeList);
for (int i = 0; i < numOfNodes; ++i) {
SNodeEntry* pNodeEntry = taosArrayGet(execInfo.pNodeList, i);
SNodeEntry *pNodeEntry = taosArrayGet(execInfo.pNodeList, i);
if (pNodeEntry->nodeId == *pVgId) {
mInfo("vgId:%d expired for some stream tasks, needs update nodeEp", *pVgId);
@ -2877,13 +2924,13 @@ int32_t setNodeEpsetExpiredFlag(const SArray *pNodeList) {
return TSDB_CODE_SUCCESS;
}
static void updateStageInfo(STaskStatusEntry* pTaskEntry, int64_t stage) {
static void updateStageInfo(STaskStatusEntry *pTaskEntry, int64_t stage) {
int32_t numOfNodes = taosArrayGetSize(execInfo.pNodeList);
for(int32_t j = 0; j < numOfNodes; ++j) {
SNodeEntry* pNodeEntry = taosArrayGet(execInfo.pNodeList, j);
for (int32_t j = 0; j < numOfNodes; ++j) {
SNodeEntry *pNodeEntry = taosArrayGet(execInfo.pNodeList, j);
if (pNodeEntry->nodeId == pTaskEntry->nodeId) {
mInfo("vgId:%d stage updated from %"PRId64 " to %"PRId64 ", nodeUpdate trigger by s-task:0x%" PRIx64, pTaskEntry->nodeId,
pTaskEntry->stage, stage, pTaskEntry->id.taskId);
mInfo("vgId:%d stage updated from %" PRId64 " to %" PRId64 ", nodeUpdate trigger by s-task:0x%" PRIx64,
pTaskEntry->nodeId, pTaskEntry->stage, stage, pTaskEntry->id.taskId);
pNodeEntry->stageUpdated = true;
pTaskEntry->stage = stage;
@ -2939,9 +2986,7 @@ int32_t mndProcessStreamHb(SRpcMsg *pReq) {
if (pTaskEntry->stage != p->stage && pTaskEntry->stage != -1) {
updateStageInfo(pTaskEntry, p->stage);
if(pTaskEntry->nodeId == SNODE_HANDLE) {
snodeChanged = true;
}
if (pTaskEntry->nodeId == SNODE_HANDLE) snodeChanged = true;
} else {
// task is idle for more than 50 sec.
if (fabs(pTaskEntry->inputQUsed - p->inputQUsed) <= DBL_EPSILON) {
@ -2981,7 +3026,7 @@ int32_t mndProcessStreamHb(SRpcMsg *pReq) {
if (p->status == TASK_STATUS__STREAM_SCAN_HISTORY) {
bool drop = needDropRelatedFillhistoryTask(pTaskEntry, &execInfo);
if(drop) {
if (drop) {
SStreamObj *pStreamObj = mndGetStreamObj(pMnode, pTaskEntry->id.streamId);
if (pStreamObj == NULL) {
mError("failed to acquire the streamObj:0x%" PRIx64 " it may have been dropped", pStreamObj->uid);
@ -3018,9 +3063,13 @@ int32_t mndProcessStreamHb(SRpcMsg *pReq) {
return TSDB_CODE_SUCCESS;
}
void freeCheckpointCandEntry(void *param) {
SCheckpointCandEntry *pEntry = param;
taosMemoryFreeClear(pEntry->pName);
}
SStreamObj *mndGetStreamObj(SMnode *pMnode, int64_t streamId) {
void *pIter = NULL;
SSdb *pSdb = pMnode->pSdb;
void * pIter = NULL;
SSdb * pSdb = pMnode->pSdb;
SStreamObj *pStream = NULL;
while ((pIter = sdbFetch(pSdb, SDB_STREAM, pIter, (void **)&pStream)) != NULL) {
@ -3031,4 +3080,4 @@ SStreamObj *mndGetStreamObj(SMnode *pMnode, int64_t streamId) {
}
return NULL;
}
}

21
source/dnode/mnode/impl/src/mndStreamTrans.c
View File

@ -13,8 +13,8 @@
* along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#include "mndTrans.h"
#include "mndStream.h"
#include "mndTrans.h"
typedef struct SKeyInfo {
void* pKey;
@ -110,5 +110,24 @@ bool streamTransConflictOtherTrans(SMnode* pMnode, const char* pSrcDb, const cha
return false;
}
int32_t mndAddtoCheckpointWaitingList(SStreamObj* pStream, int64_t checkpointId) {
SCheckpointCandEntry* pEntry = taosHashGet(execInfo.transMgmt.pWaitingList, &pStream->uid, sizeof(pStream->uid));
if (pEntry == NULL) {
SCheckpointCandEntry entry = {.streamId = pStream->uid,
.checkpointTs = taosGetTimestampMs(),
.checkpointId = checkpointId,
.pName = taosStrdup(pStream->name)};
taosHashPut(execInfo.transMgmt.pWaitingList, &pStream->uid, sizeof(pStream->uid), &entry, sizeof(entry));
int32_t size = taosHashGetSize(execInfo.transMgmt.pWaitingList);
mDebug("stream:%" PRIx64 " add into waiting list due to conflict, ts:%" PRId64 " , checkpointId: %" PRId64
", total in waitingList:%d",
pStream->uid, entry.checkpointTs, checkpointId, size);
} else {
mDebug("stream:%" PRIx64 " ts:%" PRId64 ", checkpointId:%" PRId64 " already in waiting list, no need to add into",
pStream->uid, pEntry->checkpointTs, checkpointId);
}
return TSDB_CODE_SUCCESS;
}

31
source/dnode/mnode/impl/src/mndUser.c
View File

@ -708,9 +708,8 @@ SSdbRaw *mndUserActionEncode(SUserObj *pUser) {
int32_t numOfAlterViews = taosHashGetSize(pUser->alterViews);
int32_t numOfTopics = taosHashGetSize(pUser->topics);
int32_t numOfUseDbs = taosHashGetSize(pUser->useDbs);
int32_t size = sizeof(SUserObj) + USER_RESERVE_SIZE +
(numOfReadDbs + numOfWriteDbs + numOfUseDbs) * TSDB_DB_FNAME_LEN + numOfTopics * TSDB_TOPIC_FNAME_LEN +
ipWhiteReserve;
int32_t size = sizeof(SUserObj) + USER_RESERVE_SIZE + (numOfReadDbs + numOfWriteDbs) * TSDB_DB_FNAME_LEN +
numOfTopics * TSDB_TOPIC_FNAME_LEN + ipWhiteReserve;
char *stb = taosHashIterate(pUser->readTbs, NULL);
while (stb != NULL) {
@ -720,7 +719,7 @@ SSdbRaw *mndUserActionEncode(SUserObj *pUser) {
size += keyLen;
size_t valueLen = 0;
valueLen = strlen(stb);
valueLen = strlen(stb) + 1;
size += sizeof(int32_t);
size += valueLen;
stb = taosHashIterate(pUser->readTbs, stb);
@ -734,7 +733,7 @@ SSdbRaw *mndUserActionEncode(SUserObj *pUser) {
size += keyLen;
size_t valueLen = 0;
valueLen = strlen(stb);
valueLen = strlen(stb) + 1;
size += sizeof(int32_t);
size += valueLen;
stb = taosHashIterate(pUser->writeTbs, stb);
@ -748,7 +747,7 @@ SSdbRaw *mndUserActionEncode(SUserObj *pUser) {
size += keyLen;
size_t valueLen = 0;
valueLen = strlen(stb);
valueLen = strlen(stb) + 1;
size += sizeof(int32_t);
size += valueLen;
stb = taosHashIterate(pUser->alterTbs, stb);
@ -762,7 +761,7 @@ SSdbRaw *mndUserActionEncode(SUserObj *pUser) {
size += keyLen;
size_t valueLen = 0;
valueLen = strlen(stb);
valueLen = strlen(stb) + 1;
size += sizeof(int32_t);
size += valueLen;
stb = taosHashIterate(pUser->readViews, stb);
@ -776,7 +775,7 @@ SSdbRaw *mndUserActionEncode(SUserObj *pUser) {
size += keyLen;
size_t valueLen = 0;
valueLen = strlen(stb);
valueLen = strlen(stb) + 1;
size += sizeof(int32_t);
size += valueLen;
stb = taosHashIterate(pUser->writeViews, stb);
@ -790,11 +789,21 @@ SSdbRaw *mndUserActionEncode(SUserObj *pUser) {
size += keyLen;
size_t valueLen = 0;
valueLen = strlen(stb);
valueLen = strlen(stb) + 1;
size += sizeof(int32_t);
size += valueLen;
stb = taosHashIterate(pUser->alterViews, stb);
}
}
int32_t *useDb = taosHashIterate(pUser->useDbs, NULL);
while (useDb != NULL) {
size_t keyLen = 0;
void *key = taosHashGetKey(useDb, &keyLen);
size += sizeof(int32_t);
size += keyLen;
size += sizeof(int32_t);
useDb = taosHashIterate(pUser->useDbs, useDb);
}
SSdbRaw *pRaw = sdbAllocRaw(SDB_USER, USER_VER_NUMBER, size);
if (pRaw == NULL) goto _OVER;
@ -925,7 +934,7 @@ SSdbRaw *mndUserActionEncode(SUserObj *pUser) {
stb = taosHashIterate(pUser->alterViews, stb);
}
int32_t *useDb = taosHashIterate(pUser->useDbs, NULL);
useDb = taosHashIterate(pUser->useDbs, NULL);
while (useDb != NULL) {
size_t keyLen = 0;
void *key = taosHashGetKey(useDb, &keyLen);

37
source/dnode/mnode/sdb/inc/sdb.h
View File

@ -149,7 +149,8 @@ typedef enum {
SDB_FUNC = 20,
SDB_IDX = 21,
SDB_VIEW = 22,
SDB_MAX = 23
SDB_STREAM_SEQ = 23,
SDB_MAX = 24
} ESdbType;
typedef struct SSdbRaw {
@ -169,11 +170,11 @@ typedef struct SSdbRow {
} SSdbRow;
typedef struct SSdb {
SMnode *pMnode;
SWal *pWal;
SMnode * pMnode;
SWal * pWal;
int64_t sync;
char *currDir;
char *tmpDir;
char * currDir;
char * tmpDir;
int64_t commitIndex;
int64_t commitTerm;
int64_t commitConfig;
@ -183,7 +184,7 @@ typedef struct SSdb {
int64_t tableVer[SDB_MAX];
int64_t maxId[SDB_MAX];
EKeyType keyTypes[SDB_MAX];
SHashObj *hashObjs[SDB_MAX];
SHashObj * hashObjs[SDB_MAX];
TdThreadRwlock locks[SDB_MAX];
SdbInsertFp insertFps[SDB_MAX];
SdbUpdateFp updateFps[SDB_MAX];
@ -198,25 +199,25 @@ typedef struct SSdb {
typedef struct SSdbIter {
TdFilePtr file;
int64_t total;
char *name;
char * name;
} SSdbIter;
typedef struct {
ESdbType sdbType;
EKeyType keyType;
SdbDeployFp deployFp;
SdbEncodeFp encodeFp;
SdbDecodeFp decodeFp;
SdbInsertFp insertFp;
SdbUpdateFp updateFp;
SdbDeleteFp deleteFp;
ESdbType sdbType;
EKeyType keyType;
SdbDeployFp deployFp;
SdbEncodeFp encodeFp;
SdbDecodeFp decodeFp;
SdbInsertFp insertFp;
SdbUpdateFp updateFp;
SdbDeleteFp deleteFp;
SdbValidateFp validateFp;
} SSdbTable;
typedef struct SSdbOpt {
const char *path;
SMnode *pMnode;
SWal *pWal;
SMnode * pMnode;
SWal * pWal;
int64_t sync;
} SSdbOpt;
@ -393,7 +394,7 @@ int32_t sdbGetRawSoftVer(SSdbRaw *pRaw, int8_t *sver);
int32_t sdbGetRawTotalSize(SSdbRaw *pRaw);
SSdbRow *sdbAllocRow(int32_t objSize);
void *sdbGetRowObj(SSdbRow *pRow);
void * sdbGetRowObj(SSdbRow *pRow);
void sdbFreeRow(SSdb *pSdb, SSdbRow *pRow, bool callFunc);
int32_t sdbStartRead(SSdb *pSdb, SSdbIter **ppIter, int64_t *index, int64_t *term, int64_t *config);

1
source/dnode/snode/CMakeLists.txt
View File

@ -17,4 +17,5 @@ target_link_libraries(
PRIVATE stream
PRIVATE wal
PRIVATE index
PRIVATE tqCommon
)

808
source/dnode/snode/src/snode.c
View File

@ -13,83 +13,44 @@
* along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#include "rsync.h"
#include "executor.h"
#include "rsync.h"
#include "sndInt.h"
#include "tstream.h"
#include "tqCommon.h"
#include "tuuid.h"
#include "stream.h"
#define sndError(...) \
do { \
do { \
if (sndDebugFlag & DEBUG_ERROR) { \
taosPrintLog("SND ERROR ", DEBUG_ERROR, sndDebugFlag, __VA_ARGS__); \
} \
} \
} while (0)
#define sndInfo(...) \
#define sndInfo(...) \
do { \
if (sndDebugFlag & DEBUG_INFO) { \
if (sndDebugFlag & DEBUG_INFO) { \
taosPrintLog("SND INFO ", DEBUG_INFO, sndDebugFlag, __VA_ARGS__); \
} \
} while (0)
#define sndDebug(...) \
do { \
do { \
if (sndDebugFlag & DEBUG_DEBUG) { \
taosPrintLog("SND ", DEBUG_DEBUG, sndDebugFlag, __VA_ARGS__); \
} \
} \
} while (0)
void sndEnqueueStreamDispatch(SSnode *pSnode, SRpcMsg *pMsg) {
char *msgStr = pMsg->pCont;
char *msgBody = POINTER_SHIFT(msgStr, sizeof(SMsgHead));
int32_t msgLen = pMsg->contLen - sizeof(SMsgHead);
int32_t code = 0;
SStreamDispatchReq req;
SDecoder decoder;
tDecoderInit(&decoder, msgBody, msgLen);
if (tDecodeStreamDispatchReq(&decoder, &req) < 0) {
code = TSDB_CODE_MSG_DECODE_ERROR;
tDecoderClear(&decoder);
goto FAIL;
}
tDecoderClear(&decoder);
SStreamTask *pTask = streamMetaAcquireTask(pSnode->pMeta, req.streamId, req.taskId);
if (pTask) {
SRpcMsg rsp = { .info = pMsg->info, .code = 0 };
streamProcessDispatchMsg(pTask, &req, &rsp);
tDeleteStreamDispatchReq(&req);
streamMetaReleaseTask(pSnode->pMeta, pTask);
rpcFreeCont(pMsg->pCont);
taosFreeQitem(pMsg);
return;
} else {
tDeleteStreamDispatchReq(&req);
return;
}
FAIL:
if (pMsg->info.handle == NULL) return;
SRpcMsg rsp = { .code = code, .info = pMsg->info};
tmsgSendRsp(&rsp);
rpcFreeCont(pMsg->pCont);
taosFreeQitem(pMsg);
}
int32_t sndExpandTask(SSnode *pSnode, SStreamTask *pTask, int64_t nextProcessVer) {
ASSERT(pTask->info.taskLevel == TASK_LEVEL__AGG && taosArrayGetSize(pTask->upstreamInfo.pList) != 0);
int32_t code = streamTaskInit(pTask, pSnode->pMeta, &pSnode->msgCb, nextProcessVer);
if (code != TSDB_CODE_SUCCESS) {
return code;
}
pTask->pBackend = NULL;
streamTaskOpenAllUpstreamInput(pTask);
SStreamTask* pSateTask = pTask;
SStreamTask *pSateTask = pTask;
SStreamTask task = {0};
if (pTask->info.fillHistory) {
task.id.streamId = pTask->streamTaskId.streamId;
@ -124,165 +85,35 @@ int32_t sndExpandTask(SSnode *pSnode, SStreamTask *pTask, int64_t nextProcessVer
streamTaskResetUpstreamStageInfo(pTask);
streamSetupScheduleTrigger(pTask);
SCheckpointInfo* pChkInfo = &pTask->chkInfo;
SCheckpointInfo *pChkInfo = &pTask->chkInfo;
// checkpoint ver is the kept version, handled data should be the next version.
if (pTask->chkInfo.checkpointId != 0) {
pTask->chkInfo.nextProcessVer = pTask->chkInfo.checkpointVer + 1;
sndInfo("s-task:%s restore from the checkpointId:%" PRId64 " ver:%" PRId64 " nextProcessVer:%" PRId64, pTask->id.idStr,
pChkInfo->checkpointId, pChkInfo->checkpointVer, pChkInfo->nextProcessVer);
} else {
if (pTask->chkInfo.nextProcessVer == -1) {
pTask->chkInfo.nextProcessVer = 0;
}
}
char* p = NULL;
char *p = NULL;
streamTaskGetStatus(pTask, &p);
sndInfo("snode:%d expand stream task, s-task:%s, checkpointId:%" PRId64 " checkpointVer:%" PRId64
" nextProcessVer:%" PRId64 " child id:%d, level:%d, status:%s fill-history:%d, trigger:%" PRId64 " ms",
SNODE_HANDLE, pTask->id.idStr, pChkInfo->checkpointId, pChkInfo->checkpointVer, pChkInfo->nextProcessVer,
pTask->info.selfChildId, pTask->info.taskLevel, p, pTask->info.fillHistory, pTask->info.triggerParam);
if (pTask->info.fillHistory) {
sndInfo("vgId:%d expand stream task, s-task:%s, checkpointId:%" PRId64 " checkpointVer:%" PRId64
" nextProcessVer:%" PRId64
" child id:%d, level:%d, status:%s fill-history:%d, related stream task:0x%x trigger:%" PRId64 " ms",
SNODE_HANDLE, pTask->id.idStr, pChkInfo->checkpointId, pChkInfo->checkpointVer, pChkInfo->nextProcessVer,
pTask->info.selfChildId, pTask->info.taskLevel, p, pTask->info.fillHistory,
(int32_t)pTask->streamTaskId.taskId, pTask->info.triggerParam);
} else {
sndInfo("vgId:%d expand stream task, s-task:%s, checkpointId:%" PRId64 " checkpointVer:%" PRId64
" nextProcessVer:%" PRId64
" child id:%d, level:%d, status:%s fill-history:%d, related fill-task:0x%x trigger:%" PRId64 " ms",
SNODE_HANDLE, pTask->id.idStr, pChkInfo->checkpointId, pChkInfo->checkpointVer, pChkInfo->nextProcessVer,
pTask->info.selfChildId, pTask->info.taskLevel, p, pTask->info.fillHistory,
(int32_t)pTask->hTaskInfo.id.taskId, pTask->info.triggerParam);
}
return 0;
}
int32_t sndStartStreamTasks(SSnode* pSnode) {
int32_t code = TSDB_CODE_SUCCESS;
int32_t vgId = SNODE_HANDLE;
SStreamMeta* pMeta = pSnode->pMeta;
int32_t numOfTasks = taosArrayGetSize(pMeta->pTaskList);
sndDebug("vgId:%d start to check all %d stream task(s) downstream status", vgId, numOfTasks);
if (numOfTasks == 0) {
return TSDB_CODE_SUCCESS;
}
SArray* pTaskList = NULL;
streamMetaWLock(pMeta);
pTaskList = taosArrayDup(pMeta->pTaskList, NULL);
taosHashClear(pMeta->startInfo.pReadyTaskSet);
taosHashClear(pMeta->startInfo.pFailedTaskSet);
pMeta->startInfo.startTs = taosGetTimestampMs();
streamMetaWUnLock(pMeta);
// broadcast the check downstream tasks msg
for (int32_t i = 0; i < numOfTasks; ++i) {
SStreamTaskId* pTaskId = taosArrayGet(pTaskList, i);
SStreamTask* pTask = streamMetaAcquireTask(pMeta, pTaskId->streamId, pTaskId->taskId);
if (pTask == NULL) {
continue;
}
// fill-history task can only be launched by related stream tasks.
if (pTask->info.fillHistory == 1) {
streamMetaReleaseTask(pMeta, pTask);
continue;
}
if (pTask->status.downstreamReady == 1) {
if (HAS_RELATED_FILLHISTORY_TASK(pTask)) {
sndDebug("s-task:%s downstream ready, no need to check downstream, check only related fill-history task",
pTask->id.idStr);
streamLaunchFillHistoryTask(pTask);
}
streamMetaUpdateTaskDownstreamStatus(pTask, pTask->execInfo.init, pTask->execInfo.start, true);
streamMetaReleaseTask(pMeta, pTask);
continue;
}
EStreamTaskEvent event = (HAS_RELATED_FILLHISTORY_TASK(pTask)) ? TASK_EVENT_INIT_STREAM_SCANHIST : TASK_EVENT_INIT;
int32_t ret = streamTaskHandleEvent(pTask->status.pSM, event);
if (ret != TSDB_CODE_SUCCESS) {
code = ret;
}
streamMetaReleaseTask(pMeta, pTask);
}
taosArrayDestroy(pTaskList);
return code;
}
int32_t sndResetStreamTaskStatus(SSnode* pSnode) {
SStreamMeta* pMeta = pSnode->pMeta;
int32_t vgId = pMeta->vgId;
int32_t numOfTasks = taosArrayGetSize(pMeta->pTaskList);
sndDebug("vgId:%d reset all %d stream task(s) status to be uninit", vgId, numOfTasks);
if (numOfTasks == 0) {
return TSDB_CODE_SUCCESS;
}
for (int32_t i = 0; i < numOfTasks; ++i) {
SStreamTaskId* pTaskId = taosArrayGet(pMeta->pTaskList, i);
STaskId id = {.streamId = pTaskId->streamId, .taskId = pTaskId->taskId};
SStreamTask** pTask = taosHashGet(pMeta->pTasksMap, &id, sizeof(id));
streamTaskResetStatus(*pTask);
}
return 0;
}
int32_t sndRestartStreamTasks(SSnode* pSnode) {
SStreamMeta* pMeta = pSnode->pMeta;
int32_t vgId = pMeta->vgId;
int32_t code = 0;
int64_t st = taosGetTimestampMs();
while(1) {
int32_t startVal = atomic_val_compare_exchange_32(&pMeta->startInfo.taskStarting, 0, 1);
if (startVal == 0) {
break;
}
sndDebug("vgId:%d in start stream tasks procedure, wait for 500ms and recheck", vgId);
taosMsleep(500);
}
terrno = 0;
sndInfo("vgId:%d tasks are all updated and stopped, restart all tasks, triggered by transId:%d", vgId,
pMeta->updateInfo.transId);
while (streamMetaTaskInTimer(pMeta)) {
sndDebug("vgId:%d some tasks in timer, wait for 100ms and recheck", pMeta->vgId);
taosMsleep(100);
}
streamMetaWLock(pMeta);
code = streamMetaReopen(pMeta);
if (code != TSDB_CODE_SUCCESS) {
sndError("vgId:%d failed to reopen stream meta", vgId);
streamMetaWUnLock(pMeta);
code = terrno;
return code;
}
streamMetaInitBackend(pMeta);
int64_t el = taosGetTimestampMs() - st;
sndInfo("vgId:%d close&reload state elapsed time:%.3fs", vgId, el/1000.);
code = streamMetaLoadAllTasks(pMeta);
if (code != TSDB_CODE_SUCCESS) {
sndError("vgId:%d failed to load stream tasks, code:%s", vgId, tstrerror(terrno));
streamMetaWUnLock(pMeta);
code = terrno;
return code;
}
sndInfo("vgId:%d restart all stream tasks after all tasks being updated", vgId);
sndResetStreamTaskStatus(pSnode);
streamMetaWUnLock(pMeta);
sndStartStreamTasks(pSnode);
code = terrno;
return code;
}
SSnode *sndOpen(const char *path, const SSnodeOpt *pOption) {
SSnode *pSnode = taosMemoryCalloc(1, sizeof(SSnode));
if (pSnode == NULL) {
@ -318,8 +149,8 @@ FAIL:
}
int32_t sndInit(SSnode * pSnode) {
sndResetStreamTaskStatus(pSnode);
sndStartStreamTasks(pSnode);
resetStreamTaskStatus(pSnode->pMeta);
startStreamTasks(pSnode->pMeta);
return 0;
}
@ -332,578 +163,30 @@ void sndClose(SSnode *pSnode) {
taosMemoryFree(pSnode);
}
int32_t sndGetLoad(SSnode *pSnode, SSnodeLoad *pLoad) { return 0; }
int32_t sndStartStreamTaskAsync(SSnode* pSnode, bool restart) {
SStreamMeta* pMeta = pSnode->pMeta;
int32_t vgId = pMeta->vgId;
int32_t numOfTasks = taosArrayGetSize(pMeta->pTaskList);
if (numOfTasks == 0) {
sndDebug("vgId:%d no stream tasks existed to run", vgId);
return 0;
}
SStreamTaskRunReq* pRunReq = rpcMallocCont(sizeof(SStreamTaskRunReq));
if (pRunReq == NULL) {
terrno = TSDB_CODE_OUT_OF_MEMORY;
sndError("vgId:%d failed to create msg to start wal scanning to launch stream tasks, code:%s", vgId, terrstr());
return -1;
}
sndDebug("vgId:%d start all %d stream task(s) async", vgId, numOfTasks);
pRunReq->head.vgId = vgId;
pRunReq->streamId = 0;
pRunReq->taskId = restart? STREAM_EXEC_RESTART_ALL_TASKS_ID:STREAM_EXEC_START_ALL_TASKS_ID;
SRpcMsg msg = {.msgType = TDMT_STREAM_TASK_RUN, .pCont = pRunReq, .contLen = sizeof(SStreamTaskRunReq)};
tmsgPutToQueue(&pSnode->msgCb, STREAM_QUEUE, &msg);
return 0;
}
int32_t sndProcessTaskDeployReq(SSnode *pSnode, char *msg, int32_t msgLen) {
int32_t code;
// 1.deserialize msg and build task
SStreamTask *pTask = taosMemoryCalloc(1, sizeof(SStreamTask));
if (pTask == NULL) {
return -1;
}
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t *)msg, msgLen);
code = tDecodeStreamTask(&decoder, pTask);
if (code < 0) {
tDecoderClear(&decoder);
taosMemoryFree(pTask);
return -1;
}
tDecoderClear(&decoder);
ASSERT(pTask->info.taskLevel == TASK_LEVEL__AGG);
// 2.save task
streamMetaWLock(pSnode->pMeta);
bool added = false;
code = streamMetaRegisterTask(pSnode->pMeta, -1, pTask, &added);
if (code < 0) {
streamMetaWUnLock(pSnode->pMeta);
return -1;
}
int32_t numOfTasks = streamMetaGetNumOfTasks(pSnode->pMeta);
streamMetaWUnLock(pSnode->pMeta);
char* p = NULL;
streamTaskGetStatus(pTask, &p);
sndDebug("snode:%d s-task:%s is deployed on snode and add into meta, status:%s, numOfTasks:%d", SNODE_HANDLE,
pTask->id.idStr, p, numOfTasks);
EStreamTaskEvent event = (HAS_RELATED_FILLHISTORY_TASK(pTask)) ? TASK_EVENT_INIT_STREAM_SCANHIST : TASK_EVENT_INIT;
streamTaskHandleEvent(pTask->status.pSM, event);
return 0;
}
int32_t sndProcessTaskDropReq(SSnode *pSnode, char *msg, int32_t msgLen) {
SVDropStreamTaskReq *pReq = (SVDropStreamTaskReq *)msg;
sndDebug("snode:%d receive msg to drop stream task:0x%x", pSnode->pMeta->vgId, pReq->taskId);
streamMetaUnregisterTask(pSnode->pMeta, pReq->streamId, pReq->taskId);
// commit the update
streamMetaWLock(pSnode->pMeta);
int32_t numOfTasks = streamMetaGetNumOfTasks(pSnode->pMeta);
sndDebug("vgId:%d task:0x%x dropped, remain tasks:%d", pSnode->pMeta->vgId, pReq->taskId, numOfTasks);
if (streamMetaCommit(pSnode->pMeta) < 0) {
// persist to disk
}
streamMetaWUnLock(pSnode->pMeta);
return 0;
}
int32_t sndProcessTaskRunReq(SSnode *pSnode, SRpcMsg *pMsg) {
SStreamTaskRunReq *pReq = pMsg->pCont;
int32_t taskId = pReq->taskId;
if (taskId == STREAM_EXEC_START_ALL_TASKS_ID) {
sndStartStreamTasks(pSnode);
return 0;
} else if (taskId == STREAM_EXEC_RESTART_ALL_TASKS_ID) {
sndRestartStreamTasks(pSnode);
return 0;
}
SStreamTask *pTask = streamMetaAcquireTask(pSnode->pMeta, pReq->streamId, pReq->taskId);
if (pTask) {
streamExecTask(pTask);
streamMetaReleaseTask(pSnode->pMeta, pTask);
return 0;
} else {
return -1;
}
}
int32_t sndProcessTaskDispatchReq(SSnode *pSnode, SRpcMsg *pMsg, bool exec) {
char *msgStr = pMsg->pCont;
char *msgBody = POINTER_SHIFT(msgStr, sizeof(SMsgHead));
int32_t msgLen = pMsg->contLen - sizeof(SMsgHead);
SStreamDispatchReq req;
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t *)msgBody, msgLen);
tDecodeStreamDispatchReq(&decoder, &req);
tDecoderClear(&decoder);
SStreamTask *pTask = streamMetaAcquireTask(pSnode->pMeta, req.streamId, req.taskId);
if (pTask) {
SRpcMsg rsp = {.info = pMsg->info, .code = 0};
streamProcessDispatchMsg(pTask, &req, &rsp);
tDeleteStreamDispatchReq(&req);
streamMetaReleaseTask(pSnode->pMeta, pTask);
return 0;
} else {
tDeleteStreamDispatchReq(&req);
return -1;
}
}
int32_t sndProcessTaskRetrieveReq(SSnode *pSnode, SRpcMsg *pMsg) {
char *msgStr = pMsg->pCont;
char *msgBody = POINTER_SHIFT(msgStr, sizeof(SMsgHead));
int32_t msgLen = pMsg->contLen - sizeof(SMsgHead);
SStreamRetrieveReq req;
SDecoder decoder;
tDecoderInit(&decoder, msgBody, msgLen);
tDecodeStreamRetrieveReq(&decoder, &req);
tDecoderClear(&decoder);
SStreamTask *pTask = streamMetaAcquireTask(pSnode->pMeta, req.streamId, req.dstTaskId);
if (pTask) {
SRpcMsg rsp = {.info = pMsg->info, .code = 0};
streamProcessRetrieveReq(pTask, &req, &rsp);
streamMetaReleaseTask(pSnode->pMeta, pTask);
tDeleteStreamRetrieveReq(&req);
return 0;
} else {
return -1;
}
}
int32_t sndProcessTaskDispatchRsp(SSnode *pSnode, SRpcMsg *pMsg) {
SStreamDispatchRsp *pRsp = POINTER_SHIFT(pMsg->pCont, sizeof(SMsgHead));
pRsp->upstreamTaskId = htonl(pRsp->upstreamTaskId);
pRsp->streamId = htobe64(pRsp->streamId);
pRsp->downstreamTaskId = htonl(pRsp->downstreamTaskId);
pRsp->downstreamNodeId = htonl(pRsp->downstreamNodeId);
pRsp->stage = htobe64(pRsp->stage);
pRsp->msgId = htonl(pRsp->msgId);
SStreamTask *pTask = streamMetaAcquireTask(pSnode->pMeta, pRsp->streamId, pRsp->upstreamTaskId);
if (pTask) {
streamProcessDispatchRsp(pTask, pRsp, pMsg->code);
streamMetaReleaseTask(pSnode->pMeta, pTask);
return 0;
} else {
return -1;
}
}
int32_t sndProcessTaskRetrieveRsp(SSnode *pSnode, SRpcMsg *pMsg) {
//
return 0;
}
int32_t sndProcessTaskScanHistoryFinishReq(SSnode *pSnode, SRpcMsg *pMsg) {
char *msg = POINTER_SHIFT(pMsg->pCont, sizeof(SMsgHead));
int32_t msgLen = pMsg->contLen - sizeof(SMsgHead);
// deserialize
SStreamScanHistoryFinishReq req;
SDecoder decoder;
tDecoderInit(&decoder, msg, msgLen);
tDecodeStreamScanHistoryFinishReq(&decoder, &req);
tDecoderClear(&decoder);
// find task
SStreamTask *pTask = streamMetaAcquireTask(pSnode->pMeta, req.streamId, req.downstreamTaskId);
if (pTask == NULL) {
return -1;
}
// do process request
if (streamProcessScanHistoryFinishReq(pTask, &req, &pMsg->info) < 0) {
streamMetaReleaseTask(pSnode->pMeta, pTask);
return -1;
}
streamMetaReleaseTask(pSnode->pMeta, pTask);
return 0;
}
int32_t sndProcessTaskScanHistoryFinishRsp(SSnode *pSnode, SRpcMsg *pMsg) {
char* msg = POINTER_SHIFT(pMsg->pCont, sizeof(SMsgHead));
int32_t msgLen = pMsg->contLen - sizeof(SMsgHead);
// deserialize
SStreamCompleteHistoryMsg req = {0};
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t*)msg, msgLen);
tDecodeCompleteHistoryDataMsg(&decoder, &req);
tDecoderClear(&decoder);
SStreamTask* pTask = streamMetaAcquireTask(pSnode->pMeta, req.streamId, req.upstreamTaskId);
if (pTask == NULL) {
sndError("vgId:%d process scan history finish rsp, failed to find task:0x%x, it may be destroyed",
pSnode->pMeta->vgId, req.upstreamTaskId);
return -1;
}
int32_t remain = atomic_sub_fetch_32(&pTask->notReadyTasks, 1);
if (remain > 0) {
sndDebug("s-task:%s scan-history finish rsp received from downstream task:0x%x, unfinished remain:%d",
pTask->id.idStr, req.downstreamId, remain);
} else {
sndDebug(
"s-task:%s scan-history finish rsp received from downstream task:0x%x, all downstream tasks rsp scan-history "
"completed msg",
pTask->id.idStr, req.downstreamId);
streamProcessScanHistoryFinishRsp(pTask);
}
streamMetaReleaseTask(pSnode->pMeta, pTask);
return 0;
}
// downstream task has complete the stream task checkpoint procedure, let's start the handle the rsp by execute task
int32_t sndProcessTaskCheckpointReadyMsg(SSnode *pSnode, SRpcMsg* pMsg) {
SStreamMeta* pMeta = pSnode->pMeta;
char* msg = POINTER_SHIFT(pMsg->pCont, sizeof(SMsgHead));
int32_t len = pMsg->contLen - sizeof(SMsgHead);
int32_t code = 0;
SStreamCheckpointReadyMsg req = {0};
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t*)msg, len);
if (tDecodeStreamCheckpointReadyMsg(&decoder, &req) < 0) {
code = TSDB_CODE_MSG_DECODE_ERROR;
tDecoderClear(&decoder);
return code;
}
tDecoderClear(&decoder);
SStreamTask* pTask = streamMetaAcquireTask(pMeta, req.streamId, req.upstreamTaskId);
if (pTask == NULL) {
sndError("vgId:%d failed to find s-task:0x%x, it may have been destroyed already", pMeta->vgId, req.downstreamTaskId);
return code;
}
sndDebug("snode vgId:%d s-task:%s received the checkpoint ready msg from task:0x%x (vgId:%d), handle it", pMeta->vgId,
pTask->id.idStr, req.downstreamTaskId, req.downstreamNodeId);
streamProcessCheckpointReadyMsg(pTask);
streamMetaReleaseTask(pMeta, pTask);
return code;
}
int32_t sndProcessStreamTaskCheckReq(SSnode *pSnode, SRpcMsg *pMsg) {
char *msgStr = pMsg->pCont;
char *msgBody = POINTER_SHIFT(msgStr, sizeof(SMsgHead));
int32_t msgLen = pMsg->contLen - sizeof(SMsgHead);
SStreamTaskCheckReq req;
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t *)msgBody, msgLen);
tDecodeStreamTaskCheckReq(&decoder, &req);
tDecoderClear(&decoder);
int32_t taskId = req.downstreamTaskId;
SStreamTaskCheckRsp rsp = {
.reqId = req.reqId,
.streamId = req.streamId,
.childId = req.childId,
.downstreamNodeId = req.downstreamNodeId,
.downstreamTaskId = req.downstreamTaskId,
.upstreamNodeId = req.upstreamNodeId,
.upstreamTaskId = req.upstreamTaskId,
};
SStreamTask *pTask = streamMetaAcquireTask(pSnode->pMeta, req.streamId, taskId);
if (pTask != NULL) {
rsp.status = streamTaskCheckStatus(pTask, req.upstreamTaskId, req.upstreamNodeId, req.stage, &rsp.oldStage);
streamMetaReleaseTask(pSnode->pMeta, pTask);
char* p = NULL;
streamTaskGetStatus(pTask, &p);
sndDebug("s-task:%s status:%s, recv task check req(reqId:0x%" PRIx64 ") task:0x%x (vgId:%d), ready:%d",
pTask->id.idStr, p, rsp.reqId, rsp.upstreamTaskId, rsp.upstreamNodeId, rsp.status);
} else {
rsp.status = TASK_DOWNSTREAM_NOT_READY;
sndDebug("recv task check(taskId:0x%x not built yet) req(reqId:0x%" PRIx64 ") from task:0x%x (vgId:%d), rsp status %d",
taskId, rsp.reqId, rsp.upstreamTaskId, rsp.upstreamNodeId, rsp.status);
}
SEncoder encoder;
int32_t code;
int32_t len;
tEncodeSize(tEncodeStreamTaskCheckRsp, &rsp, len, code);
if (code < 0) {
sndError("vgId:%d failed to encode task check rsp, task:0x%x", pSnode->pMeta->vgId, taskId);
return -1;
}
void *buf = rpcMallocCont(sizeof(SMsgHead) + len);
((SMsgHead *)buf)->vgId = htonl(req.upstreamNodeId);
void *abuf = POINTER_SHIFT(buf, sizeof(SMsgHead));
tEncoderInit(&encoder, (uint8_t *)abuf, len);
tEncodeStreamTaskCheckRsp(&encoder, &rsp);
tEncoderClear(&encoder);
SRpcMsg rspMsg = {.code = 0, .pCont = buf, .contLen = sizeof(SMsgHead) + len, .info = pMsg->info};
tmsgSendRsp(&rspMsg);
return 0;
}
int32_t sndProcessStreamTaskCheckRsp(SSnode* pSnode, SRpcMsg* pMsg) {
char* pReq = POINTER_SHIFT(pMsg->pCont, sizeof(SMsgHead));
int32_t len = pMsg->contLen - sizeof(SMsgHead);
int32_t code;
SStreamTaskCheckRsp rsp;
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t*)pReq, len);
code = tDecodeStreamTaskCheckRsp(&decoder, &rsp);
if (code < 0) {
tDecoderClear(&decoder);
return -1;
}
tDecoderClear(&decoder);
sndDebug("tq task:0x%x (vgId:%d) recv check rsp(reqId:0x%" PRIx64 ") from 0x%x (vgId:%d) status %d",
rsp.upstreamTaskId, rsp.upstreamNodeId, rsp.reqId, rsp.downstreamTaskId, rsp.downstreamNodeId, rsp.status);
SStreamTask* pTask = streamMetaAcquireTask(pSnode->pMeta, rsp.streamId, rsp.upstreamTaskId);
if (pTask == NULL) {
sndError("tq failed to locate the stream task:0x%x (vgId:%d), it may have been destroyed", rsp.upstreamTaskId,
pSnode->pMeta->vgId);
return -1;
}
code = streamProcessCheckRsp(pTask, &rsp);
streamMetaReleaseTask(pSnode->pMeta, pTask);
return code;
}
int32_t sndProcessTaskUpdateReq(SSnode* pSnode, SRpcMsg* pMsg) {
SStreamMeta* pMeta = pSnode->pMeta;
int32_t vgId = SNODE_HANDLE;
char* msg = POINTER_SHIFT(pMsg->pCont, sizeof(SMsgHead));
int32_t len = pMsg->contLen - sizeof(SMsgHead);
SRpcMsg rsp = {.info = pMsg->info, .code = TSDB_CODE_SUCCESS};
SStreamTaskNodeUpdateMsg req = {0};
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t*)msg, len);
if (tDecodeStreamTaskUpdateMsg(&decoder, &req) < 0) {
rsp.code = TSDB_CODE_MSG_DECODE_ERROR;
sndError("vgId:%d failed to decode task update msg, code:%s", vgId, tstrerror(rsp.code));
tDecoderClear(&decoder);
return rsp.code;
}
tDecoderClear(&decoder);
// update the nodeEpset when it exists
streamMetaWLock(pMeta);
// the task epset may be updated again and again, when replaying the WAL, the task may be in stop status.
STaskId id = {.streamId = req.streamId, .taskId = req.taskId};
SStreamTask** ppTask = (SStreamTask**)taosHashGet(pMeta->pTasksMap, &id, sizeof(id));
if (ppTask == NULL || *ppTask == NULL) {
sndError("vgId:%d failed to acquire task:0x%x when handling update, it may have been dropped already", pMeta->vgId,
req.taskId);
rsp.code = TSDB_CODE_SUCCESS;
streamMetaWUnLock(pMeta);
taosArrayDestroy(req.pNodeList);
return rsp.code;
}
SStreamTask* pTask = *ppTask;
if (pMeta->updateInfo.transId != req.transId) {
pMeta->updateInfo.transId = req.transId;
sndInfo("s-task:%s receive new trans to update nodeEp msg from mnode, transId:%d", pTask->id.idStr, req.transId);
// info needs to be kept till the new trans to update the nodeEp arrived.
taosHashClear(pMeta->updateInfo.pTasks);
} else {
sndDebug("s-task:%s recv trans to update nodeEp from mnode, transId:%d", pTask->id.idStr, req.transId);
}
STaskUpdateEntry entry = {.streamId = req.streamId, .taskId = req.taskId, .transId = req.transId};
void* exist = taosHashGet(pMeta->updateInfo.pTasks, &entry, sizeof(STaskUpdateEntry));
if (exist != NULL) {
sndDebug("s-task:%s (vgId:%d) already update in trans:%d, discard the nodeEp update msg", pTask->id.idStr, vgId,
req.transId);
rsp.code = TSDB_CODE_SUCCESS;
streamMetaWUnLock(pMeta);
taosArrayDestroy(req.pNodeList);
return rsp.code;
}
streamMetaWUnLock(pMeta);
// the following two functions should not be executed within the scope of meta lock to avoid deadlock
streamTaskUpdateEpsetInfo(pTask, req.pNodeList);
streamTaskResetStatus(pTask);
// continue after lock the meta again
streamMetaWLock(pMeta);
SStreamTask** ppHTask = NULL;
if (HAS_RELATED_FILLHISTORY_TASK(pTask)) {
ppHTask = (SStreamTask**)taosHashGet(pMeta->pTasksMap, &pTask->hTaskInfo.id, sizeof(pTask->hTaskInfo.id));
if (ppHTask == NULL || *ppHTask == NULL) {
sndError("vgId:%d failed to acquire fill-history task:0x%x when handling update, it may have been dropped already",
pMeta->vgId, req.taskId);
CLEAR_RELATED_FILLHISTORY_TASK(pTask);
} else {
sndDebug("s-task:%s fill-history task update nodeEp along with stream task", (*ppHTask)->id.idStr);
streamTaskUpdateEpsetInfo(*ppHTask, req.pNodeList);
}
}
{
streamMetaSaveTask(pMeta, pTask);
if (ppHTask != NULL) {
streamMetaSaveTask(pMeta, *ppHTask);
}
if (streamMetaCommit(pMeta) < 0) {
// persist to disk
}
}
streamTaskStop(pTask);
// keep the already handled info
taosHashPut(pMeta->updateInfo.pTasks, &entry, sizeof(entry), NULL, 0);
if (ppHTask != NULL) {
streamTaskStop(*ppHTask);
sndDebug("s-task:%s task nodeEp update completed, streamTask and related fill-history task closed", pTask->id.idStr);
taosHashPut(pMeta->updateInfo.pTasks, &(*ppHTask)->id, sizeof(pTask->id), NULL, 0);
} else {
sndDebug("s-task:%s task nodeEp update completed, streamTask closed", pTask->id.idStr);
}
rsp.code = 0;
// possibly only handle the stream task.
int32_t numOfTasks = streamMetaGetNumOfTasks(pMeta);
int32_t updateTasks = taosHashGetSize(pMeta->updateInfo.pTasks);
pMeta->startInfo.tasksWillRestart = 1;
if (updateTasks < numOfTasks) {
sndDebug("vgId:%d closed tasks:%d, unclosed:%d, all tasks will be started when nodeEp update completed", vgId,
updateTasks, (numOfTasks - updateTasks));
streamMetaWUnLock(pMeta);
} else {
sndDebug("vgId:%d all %d task(s) nodeEp updated and closed", vgId, numOfTasks);
#if 1
sndStartStreamTaskAsync(pSnode, true);
streamMetaWUnLock(pMeta);
#else
streamMetaWUnLock(pMeta);
// For debug purpose.
// the following procedure consume many CPU resource, result in the re-election of leader
// with high probability. So we employ it as a test case for the stream processing framework, with
// checkpoint/restart/nodeUpdate etc.
while(1) {
int32_t startVal = atomic_val_compare_exchange_32(&pMeta->startInfo.taskStarting, 0, 1);
if (startVal == 0) {
break;
}
tqDebug("vgId:%d in start stream tasks procedure, wait for 500ms and recheck", vgId);
taosMsleep(500);
}
while (streamMetaTaskInTimer(pMeta)) {
tqDebug("vgId:%d some tasks in timer, wait for 100ms and recheck", pMeta->vgId);
taosMsleep(100);
}
streamMetaWLock(pMeta);
int32_t code = streamMetaReopen(pMeta);
if (code != 0) {
tqError("vgId:%d failed to reopen stream meta", vgId);
streamMetaWUnLock(pMeta);
taosArrayDestroy(req.pNodeList);
return -1;
}
if (streamMetaLoadAllTasks(pTq->pStreamMeta) < 0) {
tqError("vgId:%d failed to load stream tasks", vgId);
streamMetaWUnLock(pMeta);
taosArrayDestroy(req.pNodeList);
return -1;
}
if (vnodeIsRoleLeader(pTq->pVnode) && !tsDisableStream) {
tqInfo("vgId:%d start all stream tasks after all being updated", vgId);
tqResetStreamTaskStatus(pTq);
tqStartStreamTaskAsync(pTq, false);
} else {
tqInfo("vgId:%d, follower node not start stream tasks", vgId);
}
streamMetaWUnLock(pMeta);
#endif
}
taosArrayDestroy(req.pNodeList);
return rsp.code;
}
int32_t sndProcessStreamMsg(SSnode *pSnode, SRpcMsg *pMsg) {
switch (pMsg->msgType) {
case TDMT_STREAM_TASK_RUN:
return sndProcessTaskRunReq(pSnode, pMsg);
return tqStreamTaskProcessRunReq(pSnode->pMeta, pMsg, true);
case TDMT_STREAM_TASK_DISPATCH:
return sndProcessTaskDispatchReq(pSnode, pMsg, true);
return tqStreamTaskProcessDispatchReq(pSnode->pMeta, pMsg);
case TDMT_STREAM_TASK_DISPATCH_RSP:
return sndProcessTaskDispatchRsp(pSnode, pMsg);
return tqStreamTaskProcessDispatchRsp(pSnode->pMeta, pMsg);
case TDMT_STREAM_RETRIEVE:
return sndProcessTaskRetrieveReq(pSnode, pMsg);
case TDMT_STREAM_RETRIEVE_RSP:
return sndProcessTaskRetrieveRsp(pSnode, pMsg);
return tqStreamTaskProcessRetrieveReq(pSnode->pMeta, pMsg);
case TDMT_STREAM_RETRIEVE_RSP: // 1036
break;
case TDMT_VND_STREAM_SCAN_HISTORY_FINISH:
return sndProcessTaskScanHistoryFinishReq(pSnode, pMsg);
return tqStreamTaskProcessScanHistoryFinishReq(pSnode->pMeta, pMsg);
case TDMT_VND_STREAM_SCAN_HISTORY_FINISH_RSP:
return sndProcessTaskScanHistoryFinishRsp(pSnode, pMsg);
return tqStreamTaskProcessScanHistoryFinishRsp(pSnode->pMeta, pMsg);
case TDMT_VND_STREAM_TASK_CHECK:
return sndProcessStreamTaskCheckReq(pSnode, pMsg);
return tqStreamTaskProcessCheckReq(pSnode->pMeta, pMsg);
case TDMT_VND_STREAM_TASK_CHECK_RSP:
return sndProcessStreamTaskCheckRsp(pSnode, pMsg);
return tqStreamTaskProcessCheckRsp(pSnode->pMeta, pMsg, true);
case TDMT_STREAM_TASK_CHECKPOINT_READY:
return sndProcessTaskCheckpointReadyMsg(pSnode, pMsg);
return tqStreamTaskProcessCheckpointReadyMsg(pSnode->pMeta, pMsg);
default:
sndError("invalid snode msg:%d", pMsg->msgType);
ASSERT(0);
}
return 0;
@ -912,16 +195,15 @@ int32_t sndProcessStreamMsg(SSnode *pSnode, SRpcMsg *pMsg) {
int32_t sndProcessWriteMsg(SSnode *pSnode, SRpcMsg *pMsg, SRpcMsg *pRsp) {
switch (pMsg->msgType) {
case TDMT_STREAM_TASK_DEPLOY: {
void *pReq = POINTER_SHIFT(pMsg->pCont, sizeof(SMsgHead));
void * pReq = POINTER_SHIFT(pMsg->pCont, sizeof(SMsgHead));
int32_t len = pMsg->contLen - sizeof(SMsgHead);
return sndProcessTaskDeployReq(pSnode, pReq, len);
return tqStreamTaskProcessDeployReq(pSnode->pMeta, -1, pReq, len, true, true);
}
case TDMT_STREAM_TASK_DROP:
return sndProcessTaskDropReq(pSnode, pMsg->pCont, pMsg->contLen);
return tqStreamTaskProcessDropReq(pSnode->pMeta, pMsg->pCont, pMsg->contLen);
case TDMT_VND_STREAM_TASK_UPDATE:
sndProcessTaskUpdateReq(pSnode, pMsg);
break;
return tqStreamTaskProcessUpdateReq(pSnode->pMeta, &pSnode->msgCb, pMsg, true);
default:
ASSERT(0);
}

11
source/dnode/vnode/CMakeLists.txt
View File

@ -1,4 +1,5 @@
# vnode
add_subdirectory(src/tqCommon)
add_library(vnode STATIC "")
set(
VNODE_SOURCE_FILES
@ -13,6 +14,8 @@ set(
"src/vnd/vnodeSnapshot.c"
"src/vnd/vnodeRetention.c"
"src/vnd/vnodeInitApi.c"
"src/vnd/vnodeAsync.c"
"src/vnd/vnodeHash.c"
# meta
"src/meta/metaOpen.c"
@ -117,6 +120,7 @@ if (${BUILD_CONTRIB})
PUBLIC "inc"
PUBLIC "src/inc"
PUBLIC "${TD_SOURCE_DIR}/include/libs/scalar"
PUBLIC "${TD_SOURCE_DIR}/include/dnode/vnode"
PUBLIC "${TD_SOURCE_DIR}/contrib/rocksdb/include"
)
else()
@ -125,6 +129,7 @@ else()
PUBLIC "inc"
PUBLIC "src/inc"
PUBLIC "${TD_SOURCE_DIR}/include/libs/scalar"
PUBLIC "${TD_SOURCE_DIR}/include/dnode/vnode"
)
if (${TD_LINUX})
target_include_directories(
@ -138,11 +143,6 @@ else()
endif()
endif()
target_include_directories(
vnode
PUBLIC "${TD_SOURCE_DIR}/include/dnode/vnode"
)
target_link_libraries(
vnode
PUBLIC os
@ -163,6 +163,7 @@ target_link_libraries(
PUBLIC transport
PUBLIC stream
PUBLIC index
PUBLIC tqCommon
)
IF (TD_GRANT)

2
source/dnode/vnode/inc/vnode.h
View File

@ -253,8 +253,6 @@ bool tqNextDataBlockFilterOut(STqReader *pReader, SHashObj *filterOutUids);
int32_t tqRetrieveDataBlock(STqReader *pReader, SSDataBlock **pRes, const char *idstr);
int32_t tqRetrieveTaosxBlock(STqReader *pReader, SArray *blocks, SArray *schemas, SSubmitTbData **pSubmitTbDataRet);
int32_t vnodeEnqueueStreamMsg(SVnode *pVnode, SRpcMsg *pMsg);
// sma
int32_t smaGetTSmaDays(SVnodeCfg *pCfg, void *pCont, uint32_t contLen, int32_t *days);

1
source/dnode/vnode/src/inc/meta.h
View File

@ -79,6 +79,7 @@ struct SMeta {
char* path;
SVnode* pVnode;
bool changed;
TDB* pEnv;
TXN* txn;
TTB* pTbDb;

1
source/dnode/vnode/src/inc/tq.h
View File

@ -153,7 +153,6 @@ char* tqOffsetBuildFName(const char* path, int32_t fVer);
int32_t tqOffsetRestoreFromFile(STqOffsetStore* pStore, const char* fname);
// tqStream
int32_t tqResetStreamTaskStatus(STQ* pTq);
int32_t tqStopStreamTasks(STQ* pTq);
// tq util

7
source/dnode/vnode/src/inc/tsdb.h
View File

@ -309,7 +309,12 @@ int32_t tsdbTakeReadSnap2(STsdbReader *pReader, _query_reseek_func_t reseek, STs
void tsdbUntakeReadSnap2(STsdbReader *pReader, STsdbReadSnap *pSnap, bool proactive);
// tsdbMerge.c ==============================================================================================
int32_t tsdbSchedMerge(STsdb *tsdb, int32_t fid);
typedef struct {
STsdb *tsdb;
int32_t fid;
} SMergeArg;
int32_t tsdbMerge(void *arg);
// tsdbDiskData ==============================================================================================
int32_t tDiskDataBuilderCreate(SDiskDataBuilder **ppBuilder);

29
source/dnode/vnode/src/inc/vnd.h
View File

@ -48,9 +48,32 @@ int32_t vnodeCheckCfg(const SVnodeCfg*);
int32_t vnodeEncodeConfig(const void* pObj, SJson* pJson);
int32_t vnodeDecodeConfig(const SJson* pJson, void* pObj);
// vnodeAsync.c
typedef struct SVAsync SVAsync;
typedef enum {
EVA_PRIORITY_HIGH = 0,
EVA_PRIORITY_NORMAL,
EVA_PRIORITY_LOW,
} EVAPriority;
#define VNODE_ASYNC_VALID_CHANNEL_ID(channelId) ((channelId) > 0)
#define VNODE_ASYNC_VALID_TASK_ID(taskId) ((taskId) > 0)
int32_t vnodeAsyncInit(SVAsync** async, char* label);
int32_t vnodeAsyncDestroy(SVAsync** async);
int32_t vnodeAChannelInit(SVAsync* async, int64_t* channelId);
int32_t vnodeAChannelDestroy(SVAsync* async, int64_t channelId, bool waitRunning);
int32_t vnodeAsync(SVAsync* async, EVAPriority priority, int32_t (*execute)(void*), void (*complete)(void*), void* arg,
int64_t* taskId);
int32_t vnodeAsyncC(SVAsync* async, int64_t channelId, EVAPriority priority, int32_t (*execute)(void*),
void (*complete)(void*), void* arg, int64_t* taskId);
int32_t vnodeAWait(SVAsync* async, int64_t taskId);
int32_t vnodeACancel(SVAsync* async, int64_t taskId);
int32_t vnodeAsyncSetWorkers(SVAsync* async, int32_t numWorkers);
// vnodeModule.c
int vnodeScheduleTask(int (*execute)(void*), void* arg);
int vnodeScheduleTaskEx(int tpid, int (*execute)(void*), void* arg);
extern SVAsync* vnodeAsyncHandle[2];
// vnodeBufPool.c
typedef struct SVBufPoolNode SVBufPoolNode;
@ -110,7 +133,7 @@ int32_t vnodeAsyncCommit(SVnode* pVnode);
bool vnodeShouldRollback(SVnode* pVnode);
// vnodeSync.c
int32_t vnodeSyncOpen(SVnode *pVnode, char *path, int32_t vnodeVersion);
int32_t vnodeSyncOpen(SVnode* pVnode, char* path, int32_t vnodeVersion);
int32_t vnodeSyncStart(SVnode* pVnode);
void vnodeSyncPreClose(SVnode* pVnode);
void vnodeSyncPostClose(SVnode* pVnode);

34
source/dnode/vnode/src/inc/vnodeInt.h
View File

@ -213,7 +213,7 @@ int32_t tsdbBegin(STsdb* pTsdb);
// int32_t tsdbCommit(STsdb* pTsdb, SCommitInfo* pInfo);
int32_t tsdbCacheCommit(STsdb* pTsdb);
int32_t tsdbCompact(STsdb* pTsdb, SCompactInfo* pInfo);
int32_t tsdbRetention(STsdb *tsdb, int64_t now, int32_t sync);
int32_t tsdbRetention(STsdb* tsdb, int64_t now, int32_t sync);
// int32_t tsdbFinishCommit(STsdb* pTsdb);
// int32_t tsdbRollbackCommit(STsdb* pTsdb);
int tsdbScanAndConvertSubmitMsg(STsdb* pTsdb, SSubmitReq2* pMsg);
@ -238,7 +238,6 @@ int32_t tqProcessTaskUpdateReq(STQ* pTq, SRpcMsg* pMsg);
int32_t tqProcessTaskResetReq(STQ* pTq, SRpcMsg* pMsg);
int32_t tqProcessTaskDropHTask(STQ* pTq, SRpcMsg* pMsg);
int32_t tqStartStreamTaskAsync(STQ* pTq, bool restart);
int32_t tqRestartStreamTasks(STQ* pTq);
int32_t tqExpandTask(STQ* pTq, SStreamTask* pTask, int64_t ver);
int32_t tqScanWal(STQ* pTq);
@ -267,7 +266,7 @@ int32_t tqProcessTaskResumeReq(STQ* pTq, int64_t version, char* msg, int32_t msg
int32_t tqProcessTaskCheckReq(STQ* pTq, SRpcMsg* pMsg);
int32_t tqProcessTaskCheckRsp(STQ* pTq, SRpcMsg* pMsg);
int32_t tqProcessTaskRunReq(STQ* pTq, SRpcMsg* pMsg);
int32_t tqProcessTaskDispatchReq(STQ* pTq, SRpcMsg* pMsg, bool exec);
int32_t tqProcessTaskDispatchReq(STQ* pTq, SRpcMsg* pMsg);
int32_t tqProcessTaskDispatchRsp(STQ* pTq, SRpcMsg* pMsg);
int32_t tqProcessTaskRetrieveReq(STQ* pTq, SRpcMsg* pMsg);
int32_t tqProcessTaskRetrieveRsp(STQ* pTq, SRpcMsg* pMsg);
@ -453,13 +452,16 @@ struct SVnode {
SVBufPool* recycleTail;
SVBufPool* onRecycle;
// commit variables
int64_t commitChannel;
int64_t commitTask;
SMeta* pMeta;
SSma* pSma;
STsdb* pTsdb;
SWal* pWal;
STQ* pTq;
SSink* pSink;
tsem_t canCommit;
int64_t sync;
TdThreadMutex lock;
bool blocked;
@ -499,18 +501,18 @@ struct SSma {
void* pRSmaEnv;
};
#define SMA_CFG(s) (&(s)->pVnode->config)
#define SMA_TSDB_CFG(s) (&(s)->pVnode->config.tsdbCfg)
#define SMA_RETENTION(s) ((SRetention*)&(s)->pVnode->config.tsdbCfg.retentions)
#define SMA_LOCKED(s) ((s)->locked)
#define SMA_META(s) ((s)->pVnode->pMeta)
#define SMA_VID(s) TD_VID((s)->pVnode)
#define SMA_TFS(s) ((s)->pVnode->pTfs)
#define SMA_TSMA_ENV(s) ((s)->pTSmaEnv)
#define SMA_RSMA_ENV(s) ((s)->pRSmaEnv)
#define SMA_RSMA_TSDB0(s) ((s)->pVnode->pTsdb)
#define SMA_RSMA_TSDB1(s) ((s)->pRSmaTsdb[TSDB_RETENTION_L0])
#define SMA_RSMA_TSDB2(s) ((s)->pRSmaTsdb[TSDB_RETENTION_L1])
#define SMA_CFG(s) (&(s)->pVnode->config)
#define SMA_TSDB_CFG(s) (&(s)->pVnode->config.tsdbCfg)
#define SMA_RETENTION(s) ((SRetention*)&(s)->pVnode->config.tsdbCfg.retentions)
#define SMA_LOCKED(s) ((s)->locked)
#define SMA_META(s) ((s)->pVnode->pMeta)
#define SMA_VID(s) TD_VID((s)->pVnode)
#define SMA_TFS(s) ((s)->pVnode->pTfs)
#define SMA_TSMA_ENV(s) ((s)->pTSmaEnv)
#define SMA_RSMA_ENV(s) ((s)->pRSmaEnv)
#define SMA_RSMA_TSDB0(s) ((s)->pVnode->pTsdb)
#define SMA_RSMA_TSDB1(s) ((s)->pRSmaTsdb[TSDB_RETENTION_L0])
#define SMA_RSMA_TSDB2(s) ((s)->pRSmaTsdb[TSDB_RETENTION_L1])
#define SMA_RSMA_GET_TSDB(pVnode, level) ((level == 0) ? pVnode->pTsdb : pVnode->pSma->pRSmaTsdb[level - 1])
// sma

2
source/dnode/vnode/src/meta/metaCommit.c
View File

@ -56,7 +56,7 @@ int metaPrepareAsyncCommit(SMeta *pMeta) {
code = ttlMgrFlush(pMeta->pTtlMgr, pMeta->txn);
metaULock(pMeta);
code = tdbCommit(pMeta->pEnv, pMeta->txn);
pMeta->changed = false;
return code;
}

48
source/dnode/vnode/src/meta/metaTable.c
View File

@ -251,6 +251,7 @@ int metaCreateSTable(SMeta *pMeta, int64_t version, SVCreateStbReq *pReq) {
++pMeta->pVnode->config.vndStats.numOfSTables;
pMeta->changed = true;
metaDebug("vgId:%d, stb:%s is created, suid:%" PRId64, TD_VID(pMeta->pVnode), pReq->name, pReq->suid);
return 0;
@ -325,6 +326,8 @@ _drop_super_table:
metaUpdTimeSeriesNum(pMeta);
pMeta->changed = true;
_exit:
tdbFree(pKey);
tdbFree(pData);
@ -424,6 +427,8 @@ int metaAlterSTable(SMeta *pMeta, int64_t version, SVCreateStbReq *pReq) {
metaTimeSeriesNotifyCheck(pMeta);
}
pMeta->changed = true;
_exit:
if (oStbEntry.pBuf) taosMemoryFree(oStbEntry.pBuf);
tDecoderClear(&dc);
@ -847,6 +852,7 @@ int metaCreateTable(SMeta *pMeta, int64_t ver, SVCreateTbReq *pReq, STableMetaRs
}
}
pMeta->changed = true;
metaDebug("vgId:%d, table:%s uid %" PRId64 " is created, type:%" PRId8, TD_VID(pMeta->pVnode), pReq->name, pReq->uid,
pReq->type);
return 0;
@ -895,6 +901,7 @@ int metaDropTable(SMeta *pMeta, int64_t version, SVDropTbReq *pReq, SArray *tbUi
*tbUid = uid;
}
pMeta->changed = true;
_exit:
tdbFree(pData);
return rc;
@ -938,6 +945,8 @@ void metaDropTables(SMeta *pMeta, SArray *tbUids) {
}
}
tSimpleHashCleanup(suidHash);
pMeta->changed = true;
}
static int32_t metaFilterTableByHash(SMeta *pMeta, SArray *uidList) {
@ -1233,6 +1242,7 @@ static int metaAlterTableColumn(SMeta *pMeta, int64_t version, SVAlterTbReq *pAl
if (pAlterTbReq->colName == NULL) {
terrno = TSDB_CODE_INVALID_MSG;
metaError("meta/table: null pAlterTbReq->colName");
return -1;
}
@ -1300,20 +1310,27 @@ static int metaAlterTableColumn(SMeta *pMeta, int64_t version, SVAlterTbReq *pAl
SMetaEntry oldEntry = {.type = TSDB_NORMAL_TABLE, .uid = entry.uid};
oldEntry.ntbEntry.schemaRow.nCols = pSchema->nCols;
int32_t iCol = 0;
int32_t rowLen = -1;
if (pAlterTbReq->action == TSDB_ALTER_TABLE_ADD_COLUMN ||
pAlterTbReq->action == TSDB_ALTER_TABLE_UPDATE_COLUMN_BYTES) {
rowLen = 0;
}
int32_t iCol = 0, jCol = 0;
SSchema *qColumn = NULL;
for (;;) {
pColumn = NULL;
qColumn = NULL;
if (iCol >= pSchema->nCols) break;
pColumn = &pSchema->pSchema[iCol];
if (jCol >= pSchema->nCols) break;
qColumn = &pSchema->pSchema[jCol];
if (NULL == pAlterTbReq->colName) {
metaError("meta/table: null pAlterTbReq->colName");
return -1;
if (!pColumn && (strcmp(qColumn->name, pAlterTbReq->colName) == 0)) {
pColumn = qColumn;
iCol = jCol;
if (rowLen < 0) break;
}
if (strcmp(pColumn->name, pAlterTbReq->colName) == 0) break;
iCol++;
rowLen += qColumn->bytes;
++jCol;
}
entry.version = version;
@ -1328,6 +1345,10 @@ static int metaAlterTableColumn(SMeta *pMeta, int64_t version, SVAlterTbReq *pAl
if ((terrno = grantCheck(TSDB_GRANT_TIMESERIES)) < 0) {
goto _err;
}
if (rowLen + pAlterTbReq->bytes > TSDB_MAX_BYTES_PER_ROW) {
terrno = TSDB_CODE_PAR_INVALID_ROW_LENGTH;
goto _err;
}
pSchema->version++;
pSchema->nCols++;
pNewSchema = taosMemoryMalloc(sizeof(SSchema) * pSchema->nCols);
@ -1369,10 +1390,14 @@ static int metaAlterTableColumn(SMeta *pMeta, int64_t version, SVAlterTbReq *pAl
terrno = TSDB_CODE_VND_COL_NOT_EXISTS;
goto _err;
}
if (!IS_VAR_DATA_TYPE(pColumn->type) || pColumn->bytes > pAlterTbReq->colModBytes) {
if (!IS_VAR_DATA_TYPE(pColumn->type) || pColumn->bytes >= pAlterTbReq->colModBytes) {
terrno = TSDB_CODE_VND_INVALID_TABLE_ACTION;
goto _err;
}
if (rowLen + pAlterTbReq->colModBytes - pColumn->bytes > TSDB_MAX_BYTES_PER_ROW) {
terrno = TSDB_CODE_PAR_INVALID_ROW_LENGTH;
goto _err;
}
if (tqCheckColModifiable(pMeta->pVnode->pTq, uid, pColumn->colId) != 0) {
terrno = TSDB_CODE_VND_COL_SUBSCRIBED;
goto _err;
@ -1970,6 +1995,7 @@ _err:
}
int metaAlterTable(SMeta *pMeta, int64_t version, SVAlterTbReq *pReq, STableMetaRsp *pMetaRsp) {
pMeta->changed = true;
switch (pReq->action) {
case TSDB_ALTER_TABLE_ADD_COLUMN:
case TSDB_ALTER_TABLE_DROP_COLUMN:

80
source/dnode/vnode/src/sma/smaRollup.c
View File

@ -61,7 +61,7 @@ struct SRSmaQTaskInfoItem {
int32_t len;
int8_t type;
int64_t suid;
void *qTaskInfo;
void * qTaskInfo;
};
static void tdRSmaQTaskInfoFree(qTaskInfo_t *taskHandle, int32_t vgId, int32_t level) {
@ -185,7 +185,7 @@ int32_t tdUpdateTbUidList(SSma *pSma, STbUidStore *pStore, bool isAdd) {
void *pIter = NULL;
while ((pIter = taosHashIterate(pStore->uidHash, pIter))) {
tb_uid_t *pTbSuid = (tb_uid_t *)taosHashGetKey(pIter, NULL);
SArray *pTbUids = *(SArray **)pIter;
SArray * pTbUids = *(SArray **)pIter;
if (tdUpdateTbUidListImpl(pSma, pTbSuid, pTbUids, isAdd) != TSDB_CODE_SUCCESS) {
taosHashCancelIterate(pStore->uidHash, pIter);
@ -213,7 +213,7 @@ int32_t tdFetchTbUidList(SSma *pSma, STbUidStore **ppStore, tb_uid_t suid, tb_ui
}
SRSmaStat *pStat = (SRSmaStat *)SMA_ENV_STAT(pEnv);
SHashObj *infoHash = NULL;
SHashObj * infoHash = NULL;
if (!pStat || !(infoHash = RSMA_INFO_HASH(pStat))) {
terrno = TSDB_CODE_RSMA_INVALID_STAT;
return TSDB_CODE_FAILED;
@ -264,11 +264,11 @@ static int32_t tdSetRSmaInfoItemParams(SSma *pSma, SRSmaParam *param, SRSmaStat
int8_t idx) {
if ((param->qmsgLen > 0) && param->qmsg[idx]) {
SRSmaInfoItem *pItem = &(pRSmaInfo->items[idx]);
SRetention *pRetention = SMA_RETENTION(pSma);
STsdbCfg *pTsdbCfg = SMA_TSDB_CFG(pSma);
SVnode *pVnode = pSma->pVnode;
SRetention * pRetention = SMA_RETENTION(pSma);
STsdbCfg * pTsdbCfg = SMA_TSDB_CFG(pSma);
SVnode * pVnode = pSma->pVnode;
char taskInfDir[TSDB_FILENAME_LEN] = {0};
void *pStreamState = NULL;
void * pStreamState = NULL;
// set the backend of stream state
tdRSmaQTaskInfoGetFullPath(pVnode, pRSmaInfo->suid, idx + 1, pVnode->pTfs, taskInfDir);
@ -297,6 +297,8 @@ static int32_t tdSetRSmaInfoItemParams(SSma *pSma, SRSmaParam *param, SRSmaStat
sprintf(pStreamTask->exec.qmsg, "%s", RSMA_EXEC_TASK_FLAG);
pStreamTask->chkInfo.checkpointId = streamMetaGetLatestCheckpointId(pStreamTask->pMeta);
tdRSmaTaskInit(pStreamTask->pMeta, pItem, &pStreamTask->id);
pStreamTask->status.pSM = streamCreateStateMachine(pStreamTask);
pStreamState = streamStateOpen(taskInfDir, pStreamTask, true, -1, -1);
if (!pStreamState) {
terrno = TSDB_CODE_RSMA_STREAM_STATE_OPEN;
@ -372,7 +374,7 @@ int32_t tdRSmaProcessCreateImpl(SSma *pSma, SRSmaParam *param, int64_t suid, con
}
#endif
SSmaEnv *pEnv = SMA_RSMA_ENV(pSma);
SSmaEnv * pEnv = SMA_RSMA_ENV(pSma);
SRSmaStat *pStat = (SRSmaStat *)SMA_ENV_STAT(pEnv);
SRSmaInfo *pRSmaInfo = NULL;
@ -651,9 +653,7 @@ static int32_t tdRSmaProcessDelReq(SSma *pSma, int64_t suid, int8_t level, SBatc
((SMsgHead *)pBuf)->vgId = TD_VID(pSma->pVnode);
SRpcMsg delMsg = {.msgType = TDMT_VND_BATCH_DEL,
.pCont = pBuf,
.contLen = len + sizeof(SMsgHead)};
SRpcMsg delMsg = {.msgType = TDMT_VND_BATCH_DEL, .pCont = pBuf, .contLen = len + sizeof(SMsgHead)};
code = tmsgPutToQueue(&pSma->pVnode->msgCb, WRITE_QUEUE, &delMsg);
TSDB_CHECK_CODE(code, lino, _exit);
}
@ -673,8 +673,8 @@ static int32_t tdRSmaExecAndSubmitResult(SSma *pSma, qTaskInfo_t taskInfo, SRSma
int32_t code = 0;
int32_t lino = 0;
SSDataBlock *output = NULL;
SArray *pResList = pItem->pResList;
STSchema *pTSchema = pInfo->pTSchema;
SArray * pResList = pItem->pResList;
STSchema * pTSchema = pInfo->pTSchema;
int64_t suid = pInfo->suid;
while (1) {
@ -733,7 +733,7 @@ static int32_t tdRSmaExecAndSubmitResult(SSma *pSma, qTaskInfo_t taskInfo, SRSma
}
}
STsdb *sinkTsdb = (pItem->level == TSDB_RETENTION_L1 ? pSma->pRSmaTsdb[0] : pSma->pRSmaTsdb[1]);
STsdb * sinkTsdb = (pItem->level == TSDB_RETENTION_L1 ? pSma->pRSmaTsdb[0] : pSma->pRSmaTsdb[1]);
SSubmitReq2 *pReq = NULL;
if (buildSubmitReqFromDataBlock(&pReq, output, pTSchema, output->info.id.groupId, SMA_VID(pSma), suid) < 0) {
@ -795,7 +795,7 @@ _exit:
static int32_t tdExecuteRSmaImplAsync(SSma *pSma, int64_t version, const void *pMsg, int32_t len, int32_t inputType,
SRSmaInfo *pInfo, tb_uid_t suid) {
int32_t size = RSMA_EXEC_MSG_HLEN + len; // header + payload
void *qItem = taosAllocateQitem(size, DEF_QITEM, 0);
void * qItem = taosAllocateQitem(size, DEF_QITEM, 0);
if (!qItem) {
return TSDB_CODE_FAILED;
@ -870,10 +870,10 @@ static int32_t tdRsmaPrintSubmitReq(SSma *pSma, SSubmitReq *pReq) {
* @param level
* @return int32_t
*/
static int32_t tdExecuteRSmaImpl(SSma *pSma, const void *pMsg, int32_t msgSize, int64_t version, int32_t inputType, SRSmaInfo *pInfo,
ERsmaExecType type, int8_t level) {
static int32_t tdExecuteRSmaImpl(SSma *pSma, const void *pMsg, int32_t msgSize, int64_t version, int32_t inputType,
SRSmaInfo *pInfo, ERsmaExecType type, int8_t level) {
int32_t idx = level - 1;
void *qTaskInfo = RSMA_INFO_QTASK(pInfo, idx);
void * qTaskInfo = RSMA_INFO_QTASK(pInfo, idx);
SRSmaInfoItem *pItem = RSMA_INFO_ITEM(pInfo, idx);
if (!qTaskInfo) {
@ -887,8 +887,9 @@ static int32_t tdExecuteRSmaImpl(SSma *pSma, const void *pMsg, int32_t msgSize,
return TSDB_CODE_FAILED;
}
smaDebug("vgId:%d, execute rsma %" PRIi8 " task for qTaskInfo:%p, suid:%" PRIu64 ", nMsg:%d, submitReqVer:%" PRIi64 ", inputType:%d", SMA_VID(pSma), level,
RSMA_INFO_QTASK(pInfo, idx), pInfo->suid, msgSize, version, inputType);
smaDebug("vgId:%d, execute rsma %" PRIi8 " task for qTaskInfo:%p, suid:%" PRIu64 ", nMsg:%d, submitReqVer:%" PRIi64
", inputType:%d",
SMA_VID(pSma), level, RSMA_INFO_QTASK(pInfo, idx), pInfo->suid, msgSize, version, inputType);
if ((terrno = qSetSMAInput(qTaskInfo, pMsg, msgSize, inputType)) < 0) {
smaError("vgId:%d, rsma %" PRIi8 " qSetStreamInput failed since %s", SMA_VID(pSma), level, tstrerror(terrno));
@ -912,7 +913,7 @@ static int32_t tdExecuteRSmaImpl(SSma *pSma, const void *pMsg, int32_t msgSize,
static SRSmaInfo *tdAcquireRSmaInfoBySuid(SSma *pSma, int64_t suid) {
int32_t code = 0;
int32_t lino = 0;
SSmaEnv *pEnv = SMA_RSMA_ENV(pSma);
SSmaEnv * pEnv = SMA_RSMA_ENV(pSma);
SRSmaStat *pStat = NULL;
SRSmaInfo *pRSmaInfo = NULL;
@ -1067,8 +1068,8 @@ _err:
static int32_t tdRSmaRestoreQTaskInfoInit(SSma *pSma, int64_t *nTables) {
int32_t code = 0;
int32_t lino = 0;
SVnode *pVnode = pSma->pVnode;
SArray *suidList = NULL;
SVnode * pVnode = pSma->pVnode;
SArray * suidList = NULL;
STbUidStore uidStore = {0};
SMetaReader mr = {0};
tb_uid_t suid = 0;
@ -1196,7 +1197,7 @@ int32_t tdRSmaPersistExecImpl(SRSmaStat *pRSmaStat, SHashObj *pInfoHash) {
int32_t code = 0;
int32_t lino = 0;
int32_t nTaskInfo = 0;
SSma *pSma = pRSmaStat->pSma;
SSma * pSma = pRSmaStat->pSma;
SVnode *pVnode = pSma->pVnode;
if (taosHashGetSize(pInfoHash) <= 0) {
@ -1229,7 +1230,7 @@ int32_t tdRSmaPersistExecImpl(SRSmaStat *pRSmaStat, SHashObj *pInfoHash) {
do {
int32_t nStreamFlushed = 0;
int32_t nSleep = 0;
void *infoHash = NULL;
void * infoHash = NULL;
while (true) {
while ((infoHash = taosHashIterate(pInfoHash, infoHash))) {
SRSmaInfo *pRSmaInfo = *(SRSmaInfo **)infoHash;
@ -1271,7 +1272,7 @@ _checkpoint:
SStreamMeta *pMeta = NULL;
int64_t checkpointId = taosGetTimestampNs();
bool checkpointBuilt = false;
void *infoHash = NULL;
void * infoHash = NULL;
while ((infoHash = taosHashIterate(pInfoHash, infoHash))) {
SRSmaInfo *pRSmaInfo = *(SRSmaInfo **)infoHash;
if (RSMA_INFO_IS_DEL(pRSmaInfo)) {
@ -1282,11 +1283,12 @@ _checkpoint:
SRSmaInfoItem *pItem = RSMA_INFO_ITEM(pRSmaInfo, i);
if (pItem && pItem->pStreamTask) {
SStreamTask *pTask = pItem->pStreamTask;
atomic_store_32(&pTask->pMeta->chkptNotReadyTasks, 1);
pTask->checkpointingId = checkpointId;
pTask->chkInfo.checkpointId = pTask->checkpointingId;
// atomic_store_32(&pTask->pMeta->chkptNotReadyTasks, 1);
pTask->chkInfo.checkpointingId = checkpointId;
pTask->chkInfo.checkpointId = checkpointId; // 1pTask->checkpointingId;
pTask->chkInfo.checkpointVer = pItem->submitReqVer;
pTask->info.triggerParam = pItem->fetchResultVer;
pTask->info.taskLevel = TASK_LEVEL_SMA;
if (!checkpointBuilt) {
// the stream states share one checkpoint
@ -1342,10 +1344,10 @@ _exit:
* @param tmrId
*/
static void tdRSmaFetchTrigger(void *param, void *tmrId) {
SRSmaRef *pRSmaRef = NULL;
SSma *pSma = NULL;
SRSmaStat *pStat = NULL;
SRSmaInfo *pRSmaInfo = NULL;
SRSmaRef * pRSmaRef = NULL;
SSma * pSma = NULL;
SRSmaStat * pStat = NULL;
SRSmaInfo * pRSmaInfo = NULL;
SRSmaInfoItem *pItem = NULL;
if (!(pRSmaRef = taosHashGet(smaMgmt.refHash, &param, POINTER_BYTES))) {
@ -1513,7 +1515,7 @@ _err:
}
static int32_t tdRSmaBatchExec(SSma *pSma, SRSmaInfo *pInfo, STaosQall *qall, SArray *pSubmitArr, ERsmaExecType type) {
void *msg = NULL;
void * msg = NULL;
int8_t resume = 0;
int32_t nSubmit = 0;
int32_t nDelete = 0;
@ -1628,11 +1630,11 @@ _err:
int32_t tdRSmaProcessExecImpl(SSma *pSma, ERsmaExecType type) {
int32_t code = 0;
int32_t lino = 0;
SVnode *pVnode = pSma->pVnode;
SSmaEnv *pEnv = SMA_RSMA_ENV(pSma);
SVnode * pVnode = pSma->pVnode;
SSmaEnv * pEnv = SMA_RSMA_ENV(pSma);
SRSmaStat *pRSmaStat = (SRSmaStat *)SMA_ENV_STAT(pEnv);
SHashObj *infoHash = NULL;
SArray *pSubmitArr = NULL;
SHashObj * infoHash = NULL;
SArray * pSubmitArr = NULL;
bool isFetchAll = false;
if (!pRSmaStat || !(infoHash = RSMA_INFO_HASH(pRSmaStat))) {
@ -1731,4 +1733,4 @@ _exit:
smaError("vgId:%d, %s failed at line %d since %s", TD_VID(pVnode), __func__, lino, tstrerror(code));
}
return code;
}
}

714
source/dnode/vnode/src/tq/tq.c
View File

@ -14,8 +14,8 @@
*/
#include "tq.h"
#include "stream.h"
#include "vnd.h"
#include "tqCommon.h"
typedef struct {
int8_t inited;
@ -750,21 +750,27 @@ int32_t tqExpandTask(STQ* pTq, SStreamTask* pTask, int64_t nextProcessVer) {
if (pTask->info.taskLevel == TASK_LEVEL__SOURCE) {
SStreamTask* pStateTask = pTask;
SStreamTask task = {0};
STaskId taskId = {.streamId = 0, .taskId = 0};
if (pTask->info.fillHistory) {
task.id.streamId = pTask->streamTaskId.streamId;
task.id.taskId = pTask->streamTaskId.taskId;
task.pMeta = pTask->pMeta;
pStateTask = &task;
taskId.streamId = pTask->id.streamId;
taskId.taskId = pTask->id.taskId;
pTask->id.streamId = pTask->streamTaskId.streamId;
pTask->id.taskId = pTask->streamTaskId.taskId;
}
pTask->pState = streamStateOpen(pTq->pStreamMeta->path, pStateTask, false, -1, -1);
pTask->pState = streamStateOpen(pTq->pStreamMeta->path, pTask, false, -1, -1);
if (pTask->pState == NULL) {
tqError("s-task:%s (vgId:%d) failed to open state for task", pTask->id.idStr, vgId);
return -1;
} else {
tqDebug("s-task:%s state:%p", pTask->id.idStr, pTask->pState);
}
if (pTask->info.fillHistory) {
pTask->id.streamId = taskId.streamId;
pTask->id.taskId = taskId.taskId;
}
SReadHandle handle = {
.checkpointId = pTask->chkInfo.checkpointId,
@ -785,15 +791,17 @@ int32_t tqExpandTask(STQ* pTq, SStreamTask* pTask, int64_t nextProcessVer) {
qSetTaskId(pTask->exec.pExecutor, pTask->id.taskId, pTask->id.streamId);
} else if (pTask->info.taskLevel == TASK_LEVEL__AGG) {
SStreamTask* pSateTask = pTask;
SStreamTask task = {0};
// SStreamTask task = {0};
STaskId taskId = {.streamId = 0, .taskId = 0};
if (pTask->info.fillHistory) {
task.id.streamId = pTask->streamTaskId.streamId;
task.id.taskId = pTask->streamTaskId.taskId;
task.pMeta = pTask->pMeta;
pSateTask = &task;
taskId.streamId = pTask->id.streamId;
taskId.taskId = pTask->id.taskId;
pTask->id.streamId = pTask->streamTaskId.streamId;
pTask->id.taskId = pTask->streamTaskId.taskId;
}
pTask->pState = streamStateOpen(pTq->pStreamMeta->path, pSateTask, false, -1, -1);
pTask->pState = streamStateOpen(pTq->pStreamMeta->path, pTask, false, -1, -1);
if (pTask->pState == NULL) {
tqError("s-task:%s (vgId:%d) failed to open state for task", pTask->id.idStr, vgId);
return -1;
@ -801,6 +809,11 @@ int32_t tqExpandTask(STQ* pTq, SStreamTask* pTask, int64_t nextProcessVer) {
tqDebug("s-task:%s state:%p", pTask->id.idStr, pTask->pState);
}
if (pTask->info.fillHistory) {
pTask->id.streamId = taskId.streamId;
pTask->id.taskId = taskId.taskId;
}
int32_t numOfVgroups = (int32_t)taosArrayGetSize(pTask->upstreamInfo.pList);
SReadHandle handle = {
.checkpointId = pTask->chkInfo.checkpointId,
@ -891,175 +904,15 @@ int32_t tqExpandTask(STQ* pTq, SStreamTask* pTask, int64_t nextProcessVer) {
}
int32_t tqProcessTaskCheckReq(STQ* pTq, SRpcMsg* pMsg) {
char* msgStr = pMsg->pCont;
char* msgBody = POINTER_SHIFT(msgStr, sizeof(SMsgHead));
int32_t msgLen = pMsg->contLen - sizeof(SMsgHead);
SStreamMeta* pMeta = pTq->pStreamMeta;
SStreamTaskCheckReq req;
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t*)msgBody, msgLen);
tDecodeStreamTaskCheckReq(&decoder, &req);
tDecoderClear(&decoder);
int32_t taskId = req.downstreamTaskId;
SStreamTaskCheckRsp rsp = {
.reqId = req.reqId,
.streamId = req.streamId,
.childId = req.childId,
.downstreamNodeId = req.downstreamNodeId,
.downstreamTaskId = req.downstreamTaskId,
.upstreamNodeId = req.upstreamNodeId,
.upstreamTaskId = req.upstreamTaskId,
};
// only the leader node handle the check request
if (pMeta->role == NODE_ROLE_FOLLOWER) {
tqError(
"s-task:0x%x invalid check msg from upstream:0x%x(vgId:%d), vgId:%d is follower, not handle check status msg",
taskId, req.upstreamTaskId, req.upstreamNodeId, pMeta->vgId);
rsp.status = TASK_DOWNSTREAM_NOT_LEADER;
} else {
SStreamTask* pTask = streamMetaAcquireTask(pMeta, req.streamId, taskId);
if (pTask != NULL) {
rsp.status = streamTaskCheckStatus(pTask, req.upstreamTaskId, req.upstreamNodeId, req.stage, &rsp.oldStage);
streamMetaReleaseTask(pMeta, pTask);
char* p = NULL;
streamTaskGetStatus(pTask, &p);
tqDebug("s-task:%s status:%s, stage:%" PRId64 " recv task check req(reqId:0x%" PRIx64
") task:0x%x (vgId:%d), check_status:%d",
pTask->id.idStr, p, rsp.oldStage, rsp.reqId, rsp.upstreamTaskId, rsp.upstreamNodeId, rsp.status);
} else {
rsp.status = TASK_DOWNSTREAM_NOT_READY;
tqDebug("tq recv task check(taskId:0x%" PRIx64 "-0x%x not built yet) req(reqId:0x%" PRIx64
") from task:0x%x (vgId:%d), rsp check_status %d",
req.streamId, taskId, rsp.reqId, rsp.upstreamTaskId, rsp.upstreamNodeId, rsp.status);
}
}
return streamSendCheckRsp(pMeta, &req, &rsp, &pMsg->info, taskId);
return tqStreamTaskProcessCheckReq(pTq->pStreamMeta, pMsg);
}
int32_t tqProcessTaskCheckRsp(STQ* pTq, SRpcMsg* pMsg) {
char* pReq = POINTER_SHIFT(pMsg->pCont, sizeof(SMsgHead));
int32_t len = pMsg->contLen - sizeof(SMsgHead);
int32_t vgId = pTq->pStreamMeta->vgId;
int32_t code;
SStreamTaskCheckRsp rsp;
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t*)pReq, len);
code = tDecodeStreamTaskCheckRsp(&decoder, &rsp);
if (code < 0) {
terrno = TSDB_CODE_INVALID_MSG;
tDecoderClear(&decoder);
tqError("vgId:%d failed to parse check rsp msg, code:%s", vgId, tstrerror(terrno));
return -1;
}
tDecoderClear(&decoder);
tqDebug("tq task:0x%x (vgId:%d) recv check rsp(reqId:0x%" PRIx64 ") from 0x%x (vgId:%d) status %d",
rsp.upstreamTaskId, rsp.upstreamNodeId, rsp.reqId, rsp.downstreamTaskId, rsp.downstreamNodeId, rsp.status);
if (!vnodeIsRoleLeader(pTq->pVnode)) {
tqError("vgId:%d not leader, task:0x%x not handle the check rsp, downstream:0x%x (vgId:%d)", vgId,
rsp.upstreamTaskId, rsp.downstreamTaskId, rsp.downstreamNodeId);
return code;
}
SStreamTask* pTask = streamMetaAcquireTask(pTq->pStreamMeta, rsp.streamId, rsp.upstreamTaskId);
if (pTask == NULL) {
tqError("tq failed to locate the stream task:0x%" PRIx64 "-0x%x (vgId:%d), it may have been destroyed or stopped",
rsp.streamId, rsp.upstreamTaskId, pTq->pStreamMeta->vgId);
terrno = TSDB_CODE_STREAM_TASK_NOT_EXIST;
return -1;
}
code = streamProcessCheckRsp(pTask, &rsp);
streamMetaReleaseTask(pTq->pStreamMeta, pTask);
return code;
return tqStreamTaskProcessCheckRsp(pTq->pStreamMeta, pMsg, vnodeIsRoleLeader(pTq->pVnode));
}
int32_t tqProcessTaskDeployReq(STQ* pTq, int64_t sversion, char* msg, int32_t msgLen) {
int32_t code = 0;
int32_t vgId = TD_VID(pTq->pVnode);
if (tsDisableStream) {
tqInfo("vgId:%d stream disabled, not deploy stream tasks", vgId);
return code;
}
tqDebug("vgId:%d receive new stream task deploy msg, start to build stream task", vgId);
// 1.deserialize msg and build task
int32_t size = sizeof(SStreamTask);
SStreamTask* pTask = taosMemoryCalloc(1, size);
if (pTask == NULL) {
tqError("vgId:%d failed to create stream task due to out of memory, alloc size:%d", vgId, size);
return TSDB_CODE_OUT_OF_MEMORY;
}
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t*)msg, msgLen);
code = tDecodeStreamTask(&decoder, pTask);
tDecoderClear(&decoder);
if (code != TSDB_CODE_SUCCESS) {
taosMemoryFree(pTask);
return TSDB_CODE_INVALID_MSG;
}
SStreamMeta* pStreamMeta = pTq->pStreamMeta;
// 2.save task, use the latest commit version as the initial start version of stream task.
int32_t taskId = pTask->id.taskId;
int64_t streamId = pTask->id.streamId;
bool added = false;
streamMetaWLock(pStreamMeta);
code = streamMetaRegisterTask(pStreamMeta, sversion, pTask, &added);
int32_t numOfTasks = streamMetaGetNumOfTasks(pStreamMeta);
streamMetaWUnLock(pStreamMeta);
if (code < 0) {
tqError("failed to add s-task:0x%x into vgId:%d meta, total:%d, code:%s", vgId, taskId, numOfTasks,
tstrerror(code));
tFreeStreamTask(pTask);
return code;
}
// added into meta store, pTask cannot be reference since it may have been destroyed by other threads already now if
// it is added into the meta store
if (added) {
// only handled in the leader node
if (vnodeIsRoleLeader(pTq->pVnode)) {
tqDebug("vgId:%d s-task:0x%x is deployed and add into meta, numOfTasks:%d", vgId, taskId, numOfTasks);
SStreamTask* p = streamMetaAcquireTask(pStreamMeta, streamId, taskId);
bool restored = pTq->pVnode->restored;
if (p != NULL && restored && p->info.fillHistory == 0) {
EStreamTaskEvent event = (HAS_RELATED_FILLHISTORY_TASK(p)) ? TASK_EVENT_INIT_STREAM_SCANHIST : TASK_EVENT_INIT;
streamTaskHandleEvent(p->status.pSM, event);
} else if (!restored) {
tqWarn("s-task:%s not launched since vnode(vgId:%d) not ready", p->id.idStr, vgId);
}
if (p != NULL) {
streamMetaReleaseTask(pStreamMeta, p);
}
} else {
tqDebug("vgId:%d not leader, not launch stream task s-task:0x%x", vgId, taskId);
}
} else {
tqWarn("vgId:%d failed to add s-task:0x%x, since already exists in meta store", vgId, taskId);
tFreeStreamTask(pTask);
}
return code;
return tqStreamTaskProcessDeployReq(pTq->pStreamMeta, sversion, msg, msgLen, vnodeIsRoleLeader(pTq->pVnode), pTq->pVnode->restored);
}
static void doStartFillhistoryStep2(SStreamTask* pTask, SStreamTask* pStreamTask, STQ* pTq) {
@ -1241,190 +1094,39 @@ int32_t tqProcessTaskScanHistory(STQ* pTq, SRpcMsg* pMsg) {
// only the agg tasks and the sink tasks will receive this message from upstream tasks
int32_t tqProcessTaskScanHistoryFinishReq(STQ* pTq, SRpcMsg* pMsg) {
char* msg = POINTER_SHIFT(pMsg->pCont, sizeof(SMsgHead));
int32_t msgLen = pMsg->contLen - sizeof(SMsgHead);
// deserialize
SStreamScanHistoryFinishReq req = {0};
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t*)msg, msgLen);
tDecodeStreamScanHistoryFinishReq(&decoder, &req);
tDecoderClear(&decoder);
SStreamTask* pTask = streamMetaAcquireTask(pTq->pStreamMeta, req.streamId, req.downstreamTaskId);
if (pTask == NULL) {
tqError("vgId:%d process scan history finish msg, failed to find task:0x%x, it may be destroyed",
pTq->pStreamMeta->vgId, req.downstreamTaskId);
return -1;
}
tqDebug("s-task:%s receive scan-history finish msg from task:0x%x", pTask->id.idStr, req.upstreamTaskId);
int32_t code = streamProcessScanHistoryFinishReq(pTask, &req, &pMsg->info);
streamMetaReleaseTask(pTq->pStreamMeta, pTask);
return code;
return tqStreamTaskProcessScanHistoryFinishReq(pTq->pStreamMeta, pMsg);
}
int32_t tqProcessTaskScanHistoryFinishRsp(STQ* pTq, SRpcMsg* pMsg) {
char* msg = POINTER_SHIFT(pMsg->pCont, sizeof(SMsgHead));
int32_t msgLen = pMsg->contLen - sizeof(SMsgHead);
// deserialize
SStreamCompleteHistoryMsg req = {0};
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t*)msg, msgLen);
tDecodeCompleteHistoryDataMsg(&decoder, &req);
tDecoderClear(&decoder);
SStreamTask* pTask = streamMetaAcquireTask(pTq->pStreamMeta, req.streamId, req.upstreamTaskId);
if (pTask == NULL) {
tqError("vgId:%d process scan history finish rsp, failed to find task:0x%x, it may be destroyed",
pTq->pStreamMeta->vgId, req.upstreamTaskId);
return -1;
}
int32_t remain = atomic_sub_fetch_32(&pTask->notReadyTasks, 1);
if (remain > 0) {
tqDebug("s-task:%s scan-history finish rsp received from downstream task:0x%x, unfinished remain:%d",
pTask->id.idStr, req.downstreamId, remain);
} else {
tqDebug(
"s-task:%s scan-history finish rsp received from downstream task:0x%x, all downstream tasks rsp scan-history "
"completed msg",
pTask->id.idStr, req.downstreamId);
streamProcessScanHistoryFinishRsp(pTask);
}
streamMetaReleaseTask(pTq->pStreamMeta, pTask);
return 0;
return tqStreamTaskProcessScanHistoryFinishRsp(pTq->pStreamMeta, pMsg);
}
int32_t tqProcessTaskRunReq(STQ* pTq, SRpcMsg* pMsg) {
SStreamTaskRunReq* pReq = pMsg->pCont;
int32_t taskId = pReq->taskId;
int32_t vgId = TD_VID(pTq->pVnode);
if (taskId == STREAM_EXEC_EXTRACT_DATA_IN_WAL_ID) { // all tasks are extracted submit data from the wal
tqScanWal(pTq);
return 0;
} else if (taskId == STREAM_EXEC_START_ALL_TASKS_ID) {
tqStartStreamTasks(pTq);
return 0;
} else if (taskId == STREAM_EXEC_RESTART_ALL_TASKS_ID) {
tqRestartStreamTasks(pTq);
return 0;
}
SStreamTask* pTask = streamMetaAcquireTask(pTq->pStreamMeta, pReq->streamId, taskId);
if (pTask != NULL) { // even in halt status, the data in inputQ must be processed
char* p = NULL;
if (streamTaskReadyToRun(pTask, &p)) {
tqDebug("vgId:%d s-task:%s start to process block from inputQ, next checked ver:%" PRId64, vgId, pTask->id.idStr,
pTask->chkInfo.nextProcessVer);
streamExecTask(pTask);
} else {
int8_t status = streamTaskSetSchedStatusInactive(pTask);
tqDebug("vgId:%d s-task:%s ignore run req since not in ready state, status:%s, sched-status:%d", vgId,
pTask->id.idStr, p, status);
}
streamMetaReleaseTask(pTq->pStreamMeta, pTask);
int32_t code = tqStreamTaskProcessRunReq(pTq->pStreamMeta, pMsg, vnodeIsRoleLeader(pTq->pVnode));
if(code == 0 && taskId > 0){
tqScanWalAsync(pTq, false);
return 0;
} else { // NOTE: pTask->status.schedStatus is not updated since it is not be handled by the run exec.
// todo add one function to handle this
tqError("vgId:%d failed to found s-task, taskId:0x%x may have been dropped", vgId, taskId);
return -1;
}
return code;
}
int32_t tqProcessTaskDispatchReq(STQ* pTq, SRpcMsg* pMsg, bool exec) {
char* msgStr = pMsg->pCont;
char* msgBody = POINTER_SHIFT(msgStr, sizeof(SMsgHead));
int32_t msgLen = pMsg->contLen - sizeof(SMsgHead);
SStreamDispatchReq req = {0};
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t*)msgBody, msgLen);
tDecodeStreamDispatchReq(&decoder, &req);
tDecoderClear(&decoder);
tqDebug("s-task:0x%x recv dispatch msg from 0x%x(vgId:%d)", req.taskId, req.upstreamTaskId, req.upstreamNodeId);
SStreamTask* pTask = streamMetaAcquireTask(pTq->pStreamMeta, req.streamId, req.taskId);
if (pTask) {
SRpcMsg rsp = {.info = pMsg->info, .code = 0};
streamProcessDispatchMsg(pTask, &req, &rsp);
tDeleteStreamDispatchReq(&req);
streamMetaReleaseTask(pTq->pStreamMeta, pTask);
return 0;
} else {
tqError("vgId:%d failed to find task:0x%x to handle the dispatch req, it may have been destroyed already",
pTq->pStreamMeta->vgId, req.taskId);
tDeleteStreamDispatchReq(&req);
return -1;
}
int32_t tqProcessTaskDispatchReq(STQ* pTq, SRpcMsg* pMsg) {
return tqStreamTaskProcessDispatchReq(pTq->pStreamMeta, pMsg);
}
int32_t tqProcessTaskDispatchRsp(STQ* pTq, SRpcMsg* pMsg) {
SStreamDispatchRsp* pRsp = POINTER_SHIFT(pMsg->pCont, sizeof(SMsgHead));
int32_t vgId = pTq->pStreamMeta->vgId;
pRsp->upstreamTaskId = htonl(pRsp->upstreamTaskId);
pRsp->streamId = htobe64(pRsp->streamId);
pRsp->downstreamTaskId = htonl(pRsp->downstreamTaskId);
pRsp->downstreamNodeId = htonl(pRsp->downstreamNodeId);
pRsp->stage = htobe64(pRsp->stage);
pRsp->msgId = htonl(pRsp->msgId);
SStreamTask* pTask = streamMetaAcquireTask(pTq->pStreamMeta, pRsp->streamId, pRsp->upstreamTaskId);
if (pTask) {
streamProcessDispatchRsp(pTask, pRsp, pMsg->code);
streamMetaReleaseTask(pTq->pStreamMeta, pTask);
return TSDB_CODE_SUCCESS;
} else {
tqDebug("vgId:%d failed to handle the dispatch rsp, since find task:0x%x failed", vgId, pRsp->upstreamTaskId);
terrno = TSDB_CODE_STREAM_TASK_NOT_EXIST;
return terrno;
}
return tqStreamTaskProcessDispatchRsp(pTq->pStreamMeta, pMsg);
}
int32_t tqProcessTaskDropReq(STQ* pTq, char* msg, int32_t msgLen) {
SVDropStreamTaskReq* pReq = (SVDropStreamTaskReq*)msg;
int32_t vgId = TD_VID(pTq->pVnode);
SStreamMeta* pMeta = pTq->pStreamMeta;
tqDebug("vgId:%d receive msg to drop s-task:0x%x", vgId, pReq->taskId);
SStreamTask* pTask = streamMetaAcquireTask(pMeta, pReq->streamId, pReq->taskId);
if (pTask != NULL) {
// drop the related fill-history task firstly
if (HAS_RELATED_FILLHISTORY_TASK(pTask)) {
STaskId* pHTaskId = &pTask->hTaskInfo.id;
streamMetaUnregisterTask(pMeta, pHTaskId->streamId, pHTaskId->taskId);
tqDebug("vgId:%d drop fill-history task:0x%x dropped firstly", vgId, (int32_t)pHTaskId->taskId);
}
streamMetaReleaseTask(pMeta, pTask);
}
// drop the stream task now
streamMetaUnregisterTask(pMeta, pReq->streamId, pReq->taskId);
// commit the update
streamMetaWLock(pMeta);
int32_t numOfTasks = streamMetaGetNumOfTasks(pMeta);
tqDebug("vgId:%d task:0x%x dropped, remain tasks:%d", vgId, pReq->taskId, numOfTasks);
if (streamMetaCommit(pMeta) < 0) {
// persist to disk
}
streamMetaWUnLock(pMeta);
return 0;
return tqStreamTaskProcessDropReq(pTq->pStreamMeta, msg, msgLen);
}
int32_t tqProcessTaskPauseReq(STQ* pTq, int64_t sversion, char* msg, int32_t msgLen) {
@ -1533,30 +1235,7 @@ int32_t tqProcessTaskResumeReq(STQ* pTq, int64_t sversion, char* msg, int32_t ms
}
int32_t tqProcessTaskRetrieveReq(STQ* pTq, SRpcMsg* pMsg) {
char* msgStr = pMsg->pCont;
char* msgBody = POINTER_SHIFT(msgStr, sizeof(SMsgHead));
int32_t msgLen = pMsg->contLen - sizeof(SMsgHead);
SDecoder decoder;
SStreamRetrieveReq req;
tDecoderInit(&decoder, (uint8_t*)msgBody, msgLen);
tDecodeStreamRetrieveReq(&decoder, &req);
tDecoderClear(&decoder);
int32_t vgId = pTq->pStreamMeta->vgId;
SStreamTask* pTask = streamMetaAcquireTask(pTq->pStreamMeta, req.streamId, req.dstTaskId);
if (pTask == NULL) {
tqError("vgId:%d process retrieve req, failed to acquire task:0x%x, it may have been dropped already", vgId,
req.dstTaskId);
return -1;
}
SRpcMsg rsp = {.info = pMsg->info, .code = 0};
streamProcessRetrieveReq(pTask, &req, &rsp);
streamMetaReleaseTask(pTq->pStreamMeta, pTask);
tDeleteStreamRetrieveReq(&req);
return 0;
return tqStreamTaskProcessRetrieveReq(pTq->pStreamMeta, pMsg);
}
int32_t tqProcessTaskRetrieveRsp(STQ* pTq, SRpcMsg* pMsg) {
@ -1564,89 +1243,6 @@ int32_t tqProcessTaskRetrieveRsp(STQ* pTq, SRpcMsg* pMsg) {
return 0;
}
// todo refactor.
int32_t vnodeEnqueueStreamMsg(SVnode* pVnode, SRpcMsg* pMsg) {
STQ* pTq = pVnode->pTq;
int32_t vgId = pVnode->config.vgId;
SMsgHead* msgStr = pMsg->pCont;
char* msgBody = POINTER_SHIFT(msgStr, sizeof(SMsgHead));
int32_t msgLen = pMsg->contLen - sizeof(SMsgHead);
int32_t code = 0;
SStreamDispatchReq req;
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t*)msgBody, msgLen);
if (tDecodeStreamDispatchReq(&decoder, &req) < 0) {
code = TSDB_CODE_MSG_DECODE_ERROR;
tDecoderClear(&decoder);
goto FAIL;
}
tDecoderClear(&decoder);
int32_t taskId = req.taskId;
tqDebug("vgId:%d receive dispatch msg to s-task:0x%" PRIx64 "-0x%x", vgId, req.streamId, taskId);
// for test purpose
// if (req.type == STREAM_INPUT__CHECKPOINT_TRIGGER) {
// code = TSDB_CODE_STREAM_TASK_NOT_EXIST;
// goto FAIL;
// }
SStreamTask* pTask = streamMetaAcquireTask(pTq->pStreamMeta, req.streamId, taskId);
if (pTask != NULL) {
SRpcMsg rsp = {.info = pMsg->info, .code = 0};
streamProcessDispatchMsg(pTask, &req, &rsp);
tDeleteStreamDispatchReq(&req);
streamMetaReleaseTask(pTq->pStreamMeta, pTask);
rpcFreeCont(pMsg->pCont);
taosFreeQitem(pMsg);
return 0;
} else {
tDeleteStreamDispatchReq(&req);
}
code = TSDB_CODE_STREAM_TASK_NOT_EXIST;
FAIL:
if (pMsg->info.handle == NULL) {
tqError("s-task:0x%x vgId:%d msg handle is null, abort enqueue dispatch msg", vgId, taskId);
return -1;
}
SMsgHead* pRspHead = rpcMallocCont(sizeof(SMsgHead) + sizeof(SStreamDispatchRsp));
if (pRspHead == NULL) {
SRpcMsg rsp = {.code = TSDB_CODE_OUT_OF_MEMORY, .info = pMsg->info};
tqError("s-task:0x%x send dispatch error rsp, code:%s", taskId, tstrerror(code));
tmsgSendRsp(&rsp);
rpcFreeCont(pMsg->pCont);
taosFreeQitem(pMsg);
return -1;
}
pRspHead->vgId = htonl(req.upstreamNodeId);
ASSERT(pRspHead->vgId != 0);
SStreamDispatchRsp* pRsp = POINTER_SHIFT(pRspHead, sizeof(SMsgHead));
pRsp->streamId = htobe64(req.streamId);
pRsp->upstreamTaskId = htonl(req.upstreamTaskId);
pRsp->upstreamNodeId = htonl(req.upstreamNodeId);
pRsp->downstreamNodeId = htonl(pVnode->config.vgId);
pRsp->downstreamTaskId = htonl(req.taskId);
pRsp->msgId = htonl(req.msgId);
pRsp->stage = htobe64(req.stage);
pRsp->inputStatus = TASK_OUTPUT_STATUS__NORMAL;
int32_t len = sizeof(SMsgHead) + sizeof(SStreamDispatchRsp);
SRpcMsg rsp = {.code = code, .info = pMsg->info, .contLen = len, .pCont = pRspHead};
tqError("s-task:0x%x send dispatch error rsp, code:%s", taskId, tstrerror(code));
tmsgSendRsp(&rsp);
rpcFreeCont(pMsg->pCont);
taosFreeQitem(pMsg);
return -1;
}
int32_t tqProcessTaskCheckPointSourceReq(STQ* pTq, SRpcMsg* pMsg, SRpcMsg* pRsp) {
int32_t vgId = TD_VID(pTq->pVnode);
SStreamMeta* pMeta = pTq->pStreamMeta;
@ -1697,14 +1293,13 @@ int32_t tqProcessTaskCheckPointSourceReq(STQ* pTq, SRpcMsg* pMsg, SRpcMsg* pRsp)
return TSDB_CODE_SUCCESS;
}
// downstream not ready, current the stream tasks are not all ready. Ignore this checkpoint req.
if (pTask->status.downstreamReady != 1) {
pTask->chkInfo.failedId = req.checkpointId; // record the latest failed checkpoint id
pTask->checkpointingId = req.checkpointId;
pTask->chkInfo.checkpointingId = req.checkpointId;
qError("s-task:%s not ready for checkpoint, since downstream not ready, ignore this checkpoint:%" PRId64
", set it failure",
pTask->id.idStr, req.checkpointId);
tqError("s-task:%s not ready for checkpoint, since downstream not ready, ignore this checkpoint:%" PRId64
", set it failure",
pTask->id.idStr, req.checkpointId);
streamMetaReleaseTask(pMeta, pTask);
SRpcMsg rsp = {0};
@ -1733,10 +1328,10 @@ int32_t tqProcessTaskCheckPointSourceReq(STQ* pTq, SRpcMsg* pMsg, SRpcMsg* pRsp)
// check if the checkpoint msg already sent or not.
if (status == TASK_STATUS__CK) {
ASSERT(pTask->checkpointingId == req.checkpointId);
ASSERT(pTask->chkInfo.checkpointingId == req.checkpointId);
tqWarn("s-task:%s recv checkpoint-source msg again checkpointId:%" PRId64
" already received, ignore this msg and continue process checkpoint",
pTask->id.idStr, pTask->checkpointingId);
pTask->id.idStr, pTask->chkInfo.checkpointingId);
taosThreadMutexUnlock(&pTask->lock);
streamMetaReleaseTask(pMeta, pTask);
@ -1752,10 +1347,6 @@ int32_t tqProcessTaskCheckPointSourceReq(STQ* pTq, SRpcMsg* pMsg, SRpcMsg* pRsp)
// set the initial value for generating check point
// set the mgmt epset info according to the checkout source msg from mnode, todo update mgmt epset if needed
if (pMeta->chkptNotReadyTasks == 0) {
pMeta->chkptNotReadyTasks = pMeta->numOfStreamTasks;
}
total = pMeta->numOfStreamTasks;
streamMetaWUnLock(pMeta);
@ -1776,220 +1367,11 @@ int32_t tqProcessTaskCheckPointSourceReq(STQ* pTq, SRpcMsg* pMsg, SRpcMsg* pRsp)
// downstream task has complete the stream task checkpoint procedure, let's start the handle the rsp by execute task
int32_t tqProcessTaskCheckpointReadyMsg(STQ* pTq, SRpcMsg* pMsg) {
int32_t vgId = TD_VID(pTq->pVnode);
SStreamMeta* pMeta = pTq->pStreamMeta;
char* msg = POINTER_SHIFT(pMsg->pCont, sizeof(SMsgHead));
int32_t len = pMsg->contLen - sizeof(SMsgHead);
int32_t code = 0;
SStreamCheckpointReadyMsg req = {0};
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t*)msg, len);
if (tDecodeStreamCheckpointReadyMsg(&decoder, &req) < 0) {
code = TSDB_CODE_MSG_DECODE_ERROR;
tDecoderClear(&decoder);
return code;
}
tDecoderClear(&decoder);
SStreamTask* pTask = streamMetaAcquireTask(pMeta, req.streamId, req.upstreamTaskId);
if (pTask == NULL) {
tqError("vgId:%d failed to find s-task:0x%x, it may have been destroyed already", vgId, req.downstreamTaskId);
return code;
}
tqDebug("vgId:%d s-task:%s received the checkpoint ready msg from task:0x%x (vgId:%d), handle it", vgId,
pTask->id.idStr, req.downstreamTaskId, req.downstreamNodeId);
streamProcessCheckpointReadyMsg(pTask);
streamMetaReleaseTask(pMeta, pTask);
return code;
return tqStreamTaskProcessCheckpointReadyMsg(pTq->pStreamMeta, pMsg);
}
int32_t tqProcessTaskUpdateReq(STQ* pTq, SRpcMsg* pMsg) {
SStreamMeta* pMeta = pTq->pStreamMeta;
int32_t vgId = TD_VID(pTq->pVnode);
char* msg = POINTER_SHIFT(pMsg->pCont, sizeof(SMsgHead));
int32_t len = pMsg->contLen - sizeof(SMsgHead);
SRpcMsg rsp = {.info = pMsg->info, .code = TSDB_CODE_SUCCESS};
SStreamTaskNodeUpdateMsg req = {0};
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t*)msg, len);
if (tDecodeStreamTaskUpdateMsg(&decoder, &req) < 0) {
rsp.code = TSDB_CODE_MSG_DECODE_ERROR;
tqError("vgId:%d failed to decode task update msg, code:%s", vgId, tstrerror(rsp.code));
tDecoderClear(&decoder);
return rsp.code;
}
tDecoderClear(&decoder);
// update the nodeEpset when it exists
streamMetaWLock(pMeta);
// the task epset may be updated again and again, when replaying the WAL, the task may be in stop status.
STaskId id = {.streamId = req.streamId, .taskId = req.taskId};
SStreamTask** ppTask = (SStreamTask**)taosHashGet(pMeta->pTasksMap, &id, sizeof(id));
if (ppTask == NULL || *ppTask == NULL) {
tqError("vgId:%d failed to acquire task:0x%x when handling update, it may have been dropped already", pMeta->vgId,
req.taskId);
rsp.code = TSDB_CODE_SUCCESS;
streamMetaWUnLock(pMeta);
taosArrayDestroy(req.pNodeList);
return rsp.code;
}
SStreamTask* pTask = *ppTask;
if (pMeta->updateInfo.transId != req.transId) {
pMeta->updateInfo.transId = req.transId;
tqInfo("s-task:%s receive new trans to update nodeEp msg from mnode, transId:%d", pTask->id.idStr, req.transId);
// info needs to be kept till the new trans to update the nodeEp arrived.
taosHashClear(pMeta->updateInfo.pTasks);
} else {
tqDebug("s-task:%s recv trans to update nodeEp from mnode, transId:%d", pTask->id.idStr, req.transId);
}
STaskUpdateEntry entry = {.streamId = req.streamId, .taskId = req.taskId, .transId = req.transId};
void* exist = taosHashGet(pMeta->updateInfo.pTasks, &entry, sizeof(STaskUpdateEntry));
if (exist != NULL) {
tqDebug("s-task:%s (vgId:%d) already update in trans:%d, discard the nodeEp update msg", pTask->id.idStr, vgId,
req.transId);
rsp.code = TSDB_CODE_SUCCESS;
streamMetaWUnLock(pMeta);
taosArrayDestroy(req.pNodeList);
return rsp.code;
}
streamMetaWUnLock(pMeta);
// the following two functions should not be executed within the scope of meta lock to avoid deadlock
streamTaskUpdateEpsetInfo(pTask, req.pNodeList);
streamTaskResetStatus(pTask);
// continue after lock the meta again
streamMetaWLock(pMeta);
SStreamTask** ppHTask = NULL;
if (HAS_RELATED_FILLHISTORY_TASK(pTask)) {
ppHTask = (SStreamTask**)taosHashGet(pMeta->pTasksMap, &pTask->hTaskInfo.id, sizeof(pTask->hTaskInfo.id));
if (ppHTask == NULL || *ppHTask == NULL) {
tqError("vgId:%d failed to acquire fill-history task:0x%x when handling update, it may have been dropped already",
pMeta->vgId, req.taskId);
CLEAR_RELATED_FILLHISTORY_TASK(pTask);
} else {
tqDebug("s-task:%s fill-history task update nodeEp along with stream task", (*ppHTask)->id.idStr);
streamTaskUpdateEpsetInfo(*ppHTask, req.pNodeList);
}
}
{
streamMetaSaveTask(pMeta, pTask);
if (ppHTask != NULL) {
streamMetaSaveTask(pMeta, *ppHTask);
}
if (streamMetaCommit(pMeta) < 0) {
// persist to disk
}
}
streamTaskStop(pTask);
// keep the already handled info
taosHashPut(pMeta->updateInfo.pTasks, &entry, sizeof(entry), NULL, 0);
if (ppHTask != NULL) {
streamTaskStop(*ppHTask);
tqDebug("s-task:%s task nodeEp update completed, streamTask and related fill-history task closed", pTask->id.idStr);
taosHashPut(pMeta->updateInfo.pTasks, &(*ppHTask)->id, sizeof(pTask->id), NULL, 0);
} else {
tqDebug("s-task:%s task nodeEp update completed, streamTask closed", pTask->id.idStr);
}
rsp.code = 0;
// possibly only handle the stream task.
int32_t numOfTasks = streamMetaGetNumOfTasks(pMeta);
int32_t updateTasks = taosHashGetSize(pMeta->updateInfo.pTasks);
pMeta->startInfo.tasksWillRestart = 1;
if (updateTasks < numOfTasks) {
tqDebug("vgId:%d closed tasks:%d, unclosed:%d, all tasks will be started when nodeEp update completed", vgId,
updateTasks, (numOfTasks - updateTasks));
streamMetaWUnLock(pMeta);
} else {
if (!pTq->pVnode->restored) {
tqDebug("vgId:%d vnode restore not completed, not restart the tasks, clear the start after nodeUpdate flag",
vgId);
pMeta->startInfo.tasksWillRestart = 0;
streamMetaWUnLock(pMeta);
} else {
tqDebug("vgId:%d all %d task(s) nodeEp updated and closed", vgId, numOfTasks);
#if 1
tqStartStreamTaskAsync(pTq, true);
streamMetaWUnLock(pMeta);
#else
streamMetaWUnLock(pMeta);
// For debug purpose.
// the following procedure consume many CPU resource, result in the re-election of leader
// with high probability. So we employ it as a test case for the stream processing framework, with
// checkpoint/restart/nodeUpdate etc.
while (1) {
int32_t startVal = atomic_val_compare_exchange_32(&pMeta->startInfo.taskStarting, 0, 1);
if (startVal == 0) {
break;
}
tqDebug("vgId:%d in start stream tasks procedure, wait for 500ms and recheck", vgId);
taosMsleep(500);
}
while (streamMetaTaskInTimer(pMeta)) {
tqDebug("vgId:%d some tasks in timer, wait for 100ms and recheck", pMeta->vgId);
taosMsleep(100);
}
streamMetaWLock(pMeta);
int32_t code = streamMetaReopen(pMeta);
if (code != 0) {
tqError("vgId:%d failed to reopen stream meta", vgId);
streamMetaWUnLock(pMeta);
taosArrayDestroy(req.pNodeList);
return -1;
}
streamMetaInitBackend(pMeta);
if (streamMetaLoadAllTasks(pTq->pStreamMeta) < 0) {
tqError("vgId:%d failed to load stream tasks", vgId);
streamMetaWUnLock(pMeta);
taosArrayDestroy(req.pNodeList);
return -1;
}
if (vnodeIsRoleLeader(pTq->pVnode) && !tsDisableStream) {
tqInfo("vgId:%d start all stream tasks after all being updated", vgId);
tqResetStreamTaskStatus(pTq);
tqStartStreamTaskAsync(pTq, false);
} else {
tqInfo("vgId:%d, follower node not start stream tasks", vgId);
}
streamMetaWUnLock(pMeta);
#endif
}
}
taosArrayDestroy(req.pNodeList);
return rsp.code;
return tqStreamTaskProcessUpdateReq(pTq->pStreamMeta, &pTq->pVnode->msgCb, pMsg, pTq->pVnode->restored);
}
int32_t tqProcessTaskResetReq(STQ* pTq, SRpcMsg* pMsg) {
@ -2016,7 +1398,7 @@ int32_t tqProcessTaskResetReq(STQ* pTq, SRpcMsg* pMsg) {
}
int32_t tqProcessTaskDropHTask(STQ* pTq, SRpcMsg* pMsg) {
SVDropHTaskReq* pReq = (SVDropHTaskReq*) pMsg->pCont;
SVDropHTaskReq* pReq = (SVDropHTaskReq*)pMsg->pCont;
SStreamMeta* pMeta = pTq->pStreamMeta;
SStreamTask* pTask = streamMetaAcquireTask(pMeta, pReq->streamId, pReq->taskId);

32
source/dnode/vnode/src/tq/tqStreamStateSnap.c
View File

@ -104,8 +104,8 @@ int32_t streamStateSnapRead(SStreamStateReader* pReader, uint8_t** ppData) {
pHdr->type = SNAP_DATA_STREAM_STATE_BACKEND;
pHdr->size = len;
memcpy(pHdr->data, rowData, len);
tqDebug("vgId:%d, vnode stream-state snapshot read data success", TD_VID(pReader->pTq->pVnode));
taosMemoryFree(rowData);
tqDebug("vgId:%d, vnode stream-state snapshot read data success", TD_VID(pReader->pTq->pVnode));
return code;
_err:
@ -139,7 +139,7 @@ int32_t streamStateSnapWriterOpen(STQ* pTq, int64_t sver, int64_t ever, SStreamS
pWriter->sver = sver;
pWriter->ever = ever;
sprintf(tdir, "%s%s%s%s%s", pTq->path, TD_DIRSEP, VNODE_TQ_STREAM, TD_DIRSEP, "received");
sprintf(tdir, "%s%s%s", pTq->path, TD_DIRSEP, VNODE_TQ_STREAM);
taosMkDir(tdir);
SStreamSnapWriter* pSnapWriter = NULL;
@ -167,25 +167,19 @@ int32_t streamStateSnapWriterClose(SStreamStateWriter* pWriter, int8_t rollback)
return code;
}
int32_t streamStateRebuildFromSnap(SStreamStateWriter* pWriter, int64_t chkpId) {
tqDebug("vgId:%d, vnode %s start to rebuild stream-state", TD_VID(pWriter->pTq->pVnode), STREAM_STATE_TRANSFER);
streamMetaWLock(pWriter->pTq->pStreamMeta);
int32_t code = streamMetaReopen(pWriter->pTq->pStreamMeta);
if (code == 0) {
streamMetaInitBackend(pWriter->pTq->pStreamMeta);
code = streamStateLoadTasks(pWriter);
}
streamMetaWUnLock(pWriter->pTq->pStreamMeta);
tqDebug("vgId:%d, vnode %s succ to rebuild stream-state", TD_VID(pWriter->pTq->pVnode), STREAM_STATE_TRANSFER);
taosMemoryFree(pWriter);
return code;
}
int32_t streamStateLoadTasks(SStreamStateWriter* pWriter) { return streamMetaLoadAllTasks(pWriter->pTq->pStreamMeta); }
int32_t streamStateSnapWrite(SStreamStateWriter* pWriter, uint8_t* pData, uint32_t nData) {
tqDebug("vgId:%d, vnode %s snapshot write data", TD_VID(pWriter->pTq->pVnode), STREAM_STATE_TRANSFER);
return streamSnapWrite(pWriter->pWriterImpl, pData + sizeof(SSnapDataHdr), nData - sizeof(SSnapDataHdr));
}
int32_t streamStateRebuildFromSnap(SStreamStateWriter* pWriter, int64_t chkpId) {
tqDebug("vgId:%d, vnode %s start to rebuild stream-state", TD_VID(pWriter->pTq->pVnode), STREAM_STATE_TRANSFER);
int32_t code = streamStateLoadTasks(pWriter);
tqDebug("vgId:%d, vnode %s succ to rebuild stream-state", TD_VID(pWriter->pTq->pVnode), STREAM_STATE_TRANSFER);
taosMemoryFree(pWriter);
return code;
}
int32_t streamStateLoadTasks(SStreamStateWriter* pWriter) {
return streamMetaReloadAllTasks(pWriter->pTq->pStreamMeta);
}

170
source/dnode/vnode/src/tq/tqStreamTask.c
View File

@ -15,7 +15,6 @@
#include "tq.h"
#include "vnd.h"
#include "stream.h"
#define MAX_REPEAT_SCAN_THRESHOLD 3
#define SCAN_WAL_IDLE_DURATION 100
@ -61,154 +60,6 @@ int32_t tqScanWal(STQ* pTq) {
return 0;
}
int32_t tqStartStreamTasks(STQ* pTq) {
int32_t code = TSDB_CODE_SUCCESS;
int32_t vgId = TD_VID(pTq->pVnode);
SStreamMeta* pMeta = pTq->pStreamMeta;
int32_t numOfTasks = taosArrayGetSize(pMeta->pTaskList);
tqDebug("vgId:%d start to check all %d stream task(s) downstream status", vgId, numOfTasks);
if (numOfTasks == 0) {
return TSDB_CODE_SUCCESS;
}
SArray* pTaskList = NULL;
streamMetaWLock(pMeta);
pTaskList = taosArrayDup(pMeta->pTaskList, NULL);
taosHashClear(pMeta->startInfo.pReadyTaskSet);
taosHashClear(pMeta->startInfo.pFailedTaskSet);
pMeta->startInfo.startTs = taosGetTimestampMs();
streamMetaWUnLock(pMeta);
// broadcast the check downstream tasks msg
for (int32_t i = 0; i < numOfTasks; ++i) {
SStreamTaskId* pTaskId = taosArrayGet(pTaskList, i);
SStreamTask* pTask = streamMetaAcquireTask(pMeta, pTaskId->streamId, pTaskId->taskId);
if (pTask == NULL) {
continue;
}
// fill-history task can only be launched by related stream tasks.
if (pTask->info.fillHistory == 1) {
streamMetaReleaseTask(pMeta, pTask);
continue;
}
if (pTask->status.downstreamReady == 1) {
if (HAS_RELATED_FILLHISTORY_TASK(pTask)) {
tqDebug("s-task:%s downstream ready, no need to check downstream, check only related fill-history task",
pTask->id.idStr);
streamLaunchFillHistoryTask(pTask);
}
streamMetaUpdateTaskDownstreamStatus(pTask, pTask->execInfo.init, pTask->execInfo.start, true);
streamMetaReleaseTask(pMeta, pTask);
continue;
}
EStreamTaskEvent event = (HAS_RELATED_FILLHISTORY_TASK(pTask)) ? TASK_EVENT_INIT_STREAM_SCANHIST : TASK_EVENT_INIT;
int32_t ret = streamTaskHandleEvent(pTask->status.pSM, event);
if (ret != TSDB_CODE_SUCCESS) {
code = ret;
}
streamMetaReleaseTask(pMeta, pTask);
}
taosArrayDestroy(pTaskList);
return code;
}
int32_t tqRestartStreamTasks(STQ* pTq) {
SStreamMeta* pMeta = pTq->pStreamMeta;
int32_t vgId = pMeta->vgId;
int32_t code = 0;
int64_t st = taosGetTimestampMs();
while(1) {
int32_t startVal = atomic_val_compare_exchange_32(&pMeta->startInfo.taskStarting, 0, 1);
if (startVal == 0) {
break;
}
tqDebug("vgId:%d in start stream tasks procedure, wait for 500ms and recheck", vgId);
taosMsleep(500);
}
terrno = 0;
tqInfo("vgId:%d tasks are all updated and stopped, restart all tasks, triggered by transId:%d", vgId,
pMeta->updateInfo.transId);
while (streamMetaTaskInTimer(pMeta)) {
tqDebug("vgId:%d some tasks in timer, wait for 100ms and recheck", pMeta->vgId);
taosMsleep(100);
}
streamMetaWLock(pMeta);
code = streamMetaReopen(pMeta);
if (code != TSDB_CODE_SUCCESS) {
tqError("vgId:%d failed to reopen stream meta", vgId);
streamMetaWUnLock(pMeta);
code = terrno;
return code;
}
streamMetaInitBackend(pMeta);
int64_t el = taosGetTimestampMs() - st;
tqInfo("vgId:%d close&reload state elapsed time:%.3fs", vgId, el/1000.);
code = streamMetaLoadAllTasks(pTq->pStreamMeta);
if (code != TSDB_CODE_SUCCESS) {
tqError("vgId:%d failed to load stream tasks, code:%s", vgId, tstrerror(terrno));
streamMetaWUnLock(pMeta);
code = terrno;
return code;
}
if (vnodeIsRoleLeader(pTq->pVnode) && !tsDisableStream) {
tqInfo("vgId:%d restart all stream tasks after all tasks being updated", vgId);
tqResetStreamTaskStatus(pTq);
streamMetaWUnLock(pMeta);
tqStartStreamTasks(pTq);
} else {
streamMetaResetStartInfo(&pMeta->startInfo);
streamMetaWUnLock(pMeta);
tqInfo("vgId:%d, follower node not start stream tasks", vgId);
}
code = terrno;
return code;
}
int32_t tqStartStreamTaskAsync(STQ* pTq, bool restart) {
SStreamMeta* pMeta = pTq->pStreamMeta;
int32_t vgId = pMeta->vgId;
int32_t numOfTasks = taosArrayGetSize(pMeta->pTaskList);
if (numOfTasks == 0) {
tqDebug("vgId:%d no stream tasks existed to run", vgId);
return 0;
}
SStreamTaskRunReq* pRunReq = rpcMallocCont(sizeof(SStreamTaskRunReq));
if (pRunReq == NULL) {
terrno = TSDB_CODE_OUT_OF_MEMORY;
tqError("vgId:%d failed to create msg to start wal scanning to launch stream tasks, code:%s", vgId, terrstr());
return -1;
}
tqDebug("vgId:%d start all %d stream task(s) async", vgId, numOfTasks);
pRunReq->head.vgId = vgId;
pRunReq->streamId = 0;
pRunReq->taskId = restart? STREAM_EXEC_RESTART_ALL_TASKS_ID:STREAM_EXEC_START_ALL_TASKS_ID;
SRpcMsg msg = {.msgType = TDMT_STREAM_TASK_RUN, .pCont = pRunReq, .contLen = sizeof(SStreamTaskRunReq)};
tmsgPutToQueue(&pTq->pVnode->msgCb, STREAM_QUEUE, &msg);
return 0;
}
int32_t tqScanWalAsync(STQ* pTq, bool ckPause) {
int32_t vgId = TD_VID(pTq->pVnode);
SStreamMeta* pMeta = pTq->pStreamMeta;
@ -302,27 +153,6 @@ int32_t tqStopStreamTasks(STQ* pTq) {
return 0;
}
int32_t tqResetStreamTaskStatus(STQ* pTq) {
SStreamMeta* pMeta = pTq->pStreamMeta;
int32_t vgId = TD_VID(pTq->pVnode);
int32_t numOfTasks = taosArrayGetSize(pMeta->pTaskList);
tqDebug("vgId:%d reset all %d stream task(s) status to be uninit", vgId, numOfTasks);
if (numOfTasks == 0) {
return TSDB_CODE_SUCCESS;
}
for (int32_t i = 0; i < numOfTasks; ++i) {
SStreamTaskId* pTaskId = taosArrayGet(pMeta->pTaskList, i);
STaskId id = {.streamId = pTaskId->streamId, .taskId = pTaskId->taskId};
SStreamTask** pTask = taosHashGet(pMeta->pTasksMap, &id, sizeof(id));
streamTaskResetStatus(*pTask);
}
return 0;
}
int32_t setWalReaderStartOffset(SStreamTask* pTask, int32_t vgId) {
// seek the stored version and extract data from WAL
int64_t firstVer = walReaderGetValidFirstVer(pTask->exec.pWalReader);

1
source/dnode/vnode/src/tq/tqStreamTaskSnap.c
View File

@ -238,7 +238,6 @@ int32_t streamTaskSnapWrite(SStreamTaskWriter* pWriter, uint8_t* pData, uint32_t
goto _err;
}
tDecoderClear(&decoder);
// tdbTbInsert(TTB *pTb, const void *pKey, int keyLen, const void *pVal, int valLen, TXN *pTxn)
int64_t key[2] = {taskId.streamId, taskId.taskId};
taosWLockLatch(&pTq->pStreamMeta->lock);

20
source/dnode/vnode/src/tqCommon/CMakeLists.txt Normal file
View File

@ -0,0 +1,20 @@
aux_source_directory(. TQ_SOURCE_FILES)
add_library(tqCommon STATIC ${TQ_SOURCE_FILES})
target_include_directories(
tqCommon
PUBLIC "../inc"
PUBLIC "../../inc"
)
target_link_libraries(
tqCommon
PRIVATE stream
PRIVATE common
PRIVATE transport
PRIVATE executor
PRIVATE index
PRIVATE qcom
PRIVATE qworker
PRIVATE sync
PRIVATE tfs
)

809
source/dnode/vnode/src/tqCommon/tqCommon.c Normal file
View File

@ -0,0 +1,809 @@
/*
* Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
*
* This program is free software: you can use, redistribute, and/or modify
* it under the terms of the GNU Affero General Public License, version 3
* or later ("AGPL"), as published by the Free Software Foundation.
*
* This program is distributed in the hope that it will be useful, but WITHOUT
* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
* FITNESS FOR A PARTICULAR PURPOSE.
*
* You should have received a copy of the GNU Affero General Public License
* along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#include "tstream.h"
#include "tmsgcb.h"
#include "tq.h"
typedef struct STaskUpdateEntry {
int64_t streamId;
int32_t taskId;
int32_t transId;
} STaskUpdateEntry;
int32_t tqStreamTaskStartAsync(SStreamMeta* pMeta, SMsgCb* cb, bool restart) {
int32_t vgId = pMeta->vgId;
int32_t numOfTasks = taosArrayGetSize(pMeta->pTaskList);
if (numOfTasks == 0) {
tqDebug("vgId:%d no stream tasks existed to run", vgId);
return 0;
}
SStreamTaskRunReq* pRunReq = rpcMallocCont(sizeof(SStreamTaskRunReq));
if (pRunReq == NULL) {
terrno = TSDB_CODE_OUT_OF_MEMORY;
tqError("vgId:%d failed to create msg to start wal scanning to launch stream tasks, code:%s", vgId, terrstr());
return -1;
}
tqDebug("vgId:%d start all %d stream task(s) async", vgId, numOfTasks);
pRunReq->head.vgId = vgId;
pRunReq->streamId = 0;
pRunReq->taskId = restart? STREAM_EXEC_RESTART_ALL_TASKS_ID:STREAM_EXEC_START_ALL_TASKS_ID;
SRpcMsg msg = {.msgType = TDMT_STREAM_TASK_RUN, .pCont = pRunReq, .contLen = sizeof(SStreamTaskRunReq)};
tmsgPutToQueue(cb, STREAM_QUEUE, &msg);
return 0;
}
int32_t tqStreamTaskProcessUpdateReq(SStreamMeta* pMeta, SMsgCb* cb, SRpcMsg* pMsg, bool restored) {
int32_t vgId = pMeta->vgId;
char* msg = POINTER_SHIFT(pMsg->pCont, sizeof(SMsgHead));
int32_t len = pMsg->contLen - sizeof(SMsgHead);
SRpcMsg rsp = {.info = pMsg->info, .code = TSDB_CODE_SUCCESS};
SStreamTaskNodeUpdateMsg req = {0};
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t*)msg, len);
if (tDecodeStreamTaskUpdateMsg(&decoder, &req) < 0) {
rsp.code = TSDB_CODE_MSG_DECODE_ERROR;
tqError("vgId:%d failed to decode task update msg, code:%s", vgId, tstrerror(rsp.code));
tDecoderClear(&decoder);
return rsp.code;
}
tDecoderClear(&decoder);
// update the nodeEpset when it exists
streamMetaWLock(pMeta);
// the task epset may be updated again and again, when replaying the WAL, the task may be in stop status.
STaskId id = {.streamId = req.streamId, .taskId = req.taskId};
SStreamTask** ppTask = (SStreamTask**)taosHashGet(pMeta->pTasksMap, &id, sizeof(id));
if (ppTask == NULL || *ppTask == NULL) {
tqError("vgId:%d failed to acquire task:0x%x when handling update, it may have been dropped already", pMeta->vgId,
req.taskId);
rsp.code = TSDB_CODE_SUCCESS;
streamMetaWUnLock(pMeta);
taosArrayDestroy(req.pNodeList);
return rsp.code;
}
SStreamTask* pTask = *ppTask;
if (pMeta->updateInfo.transId != req.transId) {
pMeta->updateInfo.transId = req.transId;
tqInfo("s-task:%s receive new trans to update nodeEp msg from mnode, transId:%d", pTask->id.idStr, req.transId);
// info needs to be kept till the new trans to update the nodeEp arrived.
taosHashClear(pMeta->updateInfo.pTasks);
} else {
tqDebug("s-task:%s recv trans to update nodeEp from mnode, transId:%d", pTask->id.idStr, req.transId);
}
STaskUpdateEntry entry = {.streamId = req.streamId, .taskId = req.taskId, .transId = req.transId};
void* exist = taosHashGet(pMeta->updateInfo.pTasks, &entry, sizeof(STaskUpdateEntry));
if (exist != NULL) {
tqDebug("s-task:%s (vgId:%d) already update in trans:%d, discard the nodeEp update msg", pTask->id.idStr, vgId,
req.transId);
rsp.code = TSDB_CODE_SUCCESS;
streamMetaWUnLock(pMeta);
taosArrayDestroy(req.pNodeList);
return rsp.code;
}
streamMetaWUnLock(pMeta);
// the following two functions should not be executed within the scope of meta lock to avoid deadlock
streamTaskUpdateEpsetInfo(pTask, req.pNodeList);
streamTaskResetStatus(pTask);
// continue after lock the meta again
streamMetaWLock(pMeta);
SStreamTask** ppHTask = NULL;
if (HAS_RELATED_FILLHISTORY_TASK(pTask)) {
ppHTask = (SStreamTask**)taosHashGet(pMeta->pTasksMap, &pTask->hTaskInfo.id, sizeof(pTask->hTaskInfo.id));
if (ppHTask == NULL || *ppHTask == NULL) {
tqError("vgId:%d failed to acquire fill-history task:0x%x when handling update, it may have been dropped already",
pMeta->vgId, req.taskId);
CLEAR_RELATED_FILLHISTORY_TASK(pTask);
} else {
tqDebug("s-task:%s fill-history task update nodeEp along with stream task", (*ppHTask)->id.idStr);
streamTaskUpdateEpsetInfo(*ppHTask, req.pNodeList);
}
}
{
streamMetaSaveTask(pMeta, pTask);
if (ppHTask != NULL) {
streamMetaSaveTask(pMeta, *ppHTask);
}
if (streamMetaCommit(pMeta) < 0) {
// persist to disk
}
}
streamTaskStop(pTask);
// keep the already handled info
taosHashPut(pMeta->updateInfo.pTasks, &entry, sizeof(entry), NULL, 0);
if (ppHTask != NULL) {
streamTaskStop(*ppHTask);
tqDebug("s-task:%s task nodeEp update completed, streamTask and related fill-history task closed", pTask->id.idStr);
taosHashPut(pMeta->updateInfo.pTasks, &(*ppHTask)->id, sizeof(pTask->id), NULL, 0);
} else {
tqDebug("s-task:%s task nodeEp update completed, streamTask closed", pTask->id.idStr);
}
rsp.code = 0;
// possibly only handle the stream task.
int32_t numOfTasks = streamMetaGetNumOfTasks(pMeta);
int32_t updateTasks = taosHashGetSize(pMeta->updateInfo.pTasks);
pMeta->startInfo.tasksWillRestart = 1;
if (updateTasks < numOfTasks) {
tqDebug("vgId:%d closed tasks:%d, unclosed:%d, all tasks will be started when nodeEp update completed", vgId,
updateTasks, (numOfTasks - updateTasks));
streamMetaWUnLock(pMeta);
} else {
if (!restored) {
tqDebug("vgId:%d vnode restore not completed, not restart the tasks, clear the start after nodeUpdate flag", vgId);
pMeta->startInfo.tasksWillRestart = 0;
streamMetaWUnLock(pMeta);
} else {
tqDebug("vgId:%d all %d task(s) nodeEp updated and closed", vgId, numOfTasks);
#if 1
tqStreamTaskStartAsync(pMeta, cb, true);
streamMetaWUnLock(pMeta);
#else
streamMetaWUnLock(pMeta);
// For debug purpose.
// the following procedure consume many CPU resource, result in the re-election of leader
// with high probability. So we employ it as a test case for the stream processing framework, with
// checkpoint/restart/nodeUpdate etc.
while (1) {
int32_t startVal = atomic_val_compare_exchange_32(&pMeta->startInfo.taskStarting, 0, 1);
if (startVal == 0) {
break;
}
tqDebug("vgId:%d in start stream tasks procedure, wait for 500ms and recheck", vgId);
taosMsleep(500);
}
while (streamMetaTaskInTimer(pMeta)) {
tqDebug("vgId:%d some tasks in timer, wait for 100ms and recheck", pMeta->vgId);
taosMsleep(100);
}
streamMetaWLock(pMeta);
int32_t code = streamMetaReopen(pMeta);
if (code != 0) {
tqError("vgId:%d failed to reopen stream meta", vgId);
streamMetaWUnLock(pMeta);
taosArrayDestroy(req.pNodeList);
return -1;
}
streamMetaInitBackend(pMeta);
if (streamMetaLoadAllTasks(pTq->pStreamMeta) < 0) {
tqError("vgId:%d failed to load stream tasks", vgId);
streamMetaWUnLock(pMeta);
taosArrayDestroy(req.pNodeList);
return -1;
}
if (vnodeIsRoleLeader(pTq->pVnode) && !tsDisableStream) {
tqInfo("vgId:%d start all stream tasks after all being updated", vgId);
resetStreamTaskStatus(pTq->pStreamMeta);
tqStartStreamTaskAsync(pTq, false);
} else {
tqInfo("vgId:%d, follower node not start stream tasks", vgId);
}
streamMetaWUnLock(pMeta);
#endif
}
}
taosArrayDestroy(req.pNodeList);
return rsp.code;
}
int32_t tqStreamTaskProcessDispatchReq(SStreamMeta* pMeta, SRpcMsg* pMsg) {
char* msgStr = pMsg->pCont;
char* msgBody = POINTER_SHIFT(msgStr, sizeof(SMsgHead));
int32_t msgLen = pMsg->contLen - sizeof(SMsgHead);
SStreamDispatchReq req = {0};
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t*)msgBody, msgLen);
if (tDecodeStreamDispatchReq(&decoder, &req) < 0) {
tDecoderClear(&decoder);
return TSDB_CODE_MSG_DECODE_ERROR;
}
tDecoderClear(&decoder);
tqDebug("s-task:0x%x recv dispatch msg from 0x%x(vgId:%d)", req.taskId, req.upstreamTaskId, req.upstreamNodeId);
SStreamTask* pTask = streamMetaAcquireTask(pMeta, req.streamId, req.taskId);
if (pTask) {
SRpcMsg rsp = {.info = pMsg->info, .code = 0};
if (streamProcessDispatchMsg(pTask, &req, &rsp) != 0){
return -1;
}
tDeleteStreamDispatchReq(&req);
streamMetaReleaseTask(pMeta, pTask);
return 0;
} else {
tqError("vgId:%d failed to find task:0x%x to handle the dispatch req, it may have been destroyed already",
pMeta->vgId, req.taskId);
SMsgHead* pRspHead = rpcMallocCont(sizeof(SMsgHead) + sizeof(SStreamDispatchRsp));
if (pRspHead == NULL) {
terrno = TSDB_CODE_OUT_OF_MEMORY;
tqError("s-task:0x%x send dispatch error rsp, out of memory", req.taskId);
return -1;
}
pRspHead->vgId = htonl(req.upstreamNodeId);
ASSERT(pRspHead->vgId != 0);
SStreamDispatchRsp* pRsp = POINTER_SHIFT(pRspHead, sizeof(SMsgHead));
pRsp->streamId = htobe64(req.streamId);
pRsp->upstreamTaskId = htonl(req.upstreamTaskId);
pRsp->upstreamNodeId = htonl(req.upstreamNodeId);
pRsp->downstreamNodeId = htonl(pMeta->vgId);
pRsp->downstreamTaskId = htonl(req.taskId);
pRsp->msgId = htonl(req.msgId);
pRsp->stage = htobe64(req.stage);
pRsp->inputStatus = TASK_OUTPUT_STATUS__NORMAL;
int32_t len = sizeof(SMsgHead) + sizeof(SStreamDispatchRsp);
SRpcMsg rsp = {.code = TSDB_CODE_STREAM_TASK_NOT_EXIST, .info = pMsg->info, .contLen = len, .pCont = pRspHead};
tqError("s-task:0x%x send dispatch error rsp, no task", req.taskId);
tmsgSendRsp(&rsp);
tDeleteStreamDispatchReq(&req);
return 0;
}
}
int32_t tqStreamTaskProcessDispatchRsp(SStreamMeta* pMeta, SRpcMsg* pMsg) {
SStreamDispatchRsp* pRsp = POINTER_SHIFT(pMsg->pCont, sizeof(SMsgHead));
int32_t vgId = pMeta->vgId;
pRsp->upstreamTaskId = htonl(pRsp->upstreamTaskId);
pRsp->streamId = htobe64(pRsp->streamId);
pRsp->downstreamTaskId = htonl(pRsp->downstreamTaskId);
pRsp->downstreamNodeId = htonl(pRsp->downstreamNodeId);
pRsp->stage = htobe64(pRsp->stage);
pRsp->msgId = htonl(pRsp->msgId);
SStreamTask* pTask = streamMetaAcquireTask(pMeta, pRsp->streamId, pRsp->upstreamTaskId);
if (pTask) {
streamProcessDispatchRsp(pTask, pRsp, pMsg->code);
streamMetaReleaseTask(pMeta, pTask);
return TSDB_CODE_SUCCESS;
} else {
tqDebug("vgId:%d failed to handle the dispatch rsp, since find task:0x%x failed", vgId, pRsp->upstreamTaskId);
terrno = TSDB_CODE_STREAM_TASK_NOT_EXIST;
return terrno;
}
}
int32_t tqStreamTaskProcessRetrieveReq(SStreamMeta* pMeta, SRpcMsg* pMsg) {
char* msgStr = pMsg->pCont;
char* msgBody = POINTER_SHIFT(msgStr, sizeof(SMsgHead));
int32_t msgLen = pMsg->contLen - sizeof(SMsgHead);
SDecoder decoder;
SStreamRetrieveReq req;
tDecoderInit(&decoder, (uint8_t*)msgBody, msgLen);
tDecodeStreamRetrieveReq(&decoder, &req);
tDecoderClear(&decoder);
SStreamTask* pTask = streamMetaAcquireTask(pMeta, req.streamId, req.dstTaskId);
if (pTask == NULL) {
tqError("vgId:%d process retrieve req, failed to acquire task:0x%x, it may have been dropped already", pMeta->vgId,
req.dstTaskId);
return -1;
}
SRpcMsg rsp = {.info = pMsg->info, .code = 0};
streamProcessRetrieveReq(pTask, &req, &rsp);
streamMetaReleaseTask(pMeta, pTask);
tDeleteStreamRetrieveReq(&req);
return 0;
}
int32_t tqStreamTaskProcessScanHistoryFinishReq(SStreamMeta* pMeta, SRpcMsg* pMsg) {
char* msg = POINTER_SHIFT(pMsg->pCont, sizeof(SMsgHead));
int32_t msgLen = pMsg->contLen - sizeof(SMsgHead);
// deserialize
SStreamScanHistoryFinishReq req = {0};
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t*)msg, msgLen);
tDecodeStreamScanHistoryFinishReq(&decoder, &req);
tDecoderClear(&decoder);
SStreamTask* pTask = streamMetaAcquireTask(pMeta, req.streamId, req.downstreamTaskId);
if (pTask == NULL) {
tqError("vgId:%d process scan history finish msg, failed to find task:0x%x, it may be destroyed",
pMeta->vgId, req.downstreamTaskId);
return -1;
}
tqDebug("s-task:%s receive scan-history finish msg from task:0x%x", pTask->id.idStr, req.upstreamTaskId);
int32_t code = streamProcessScanHistoryFinishReq(pTask, &req, &pMsg->info);
streamMetaReleaseTask(pMeta, pTask);
return code;
}
int32_t tqStreamTaskProcessScanHistoryFinishRsp(SStreamMeta* pMeta, SRpcMsg* pMsg) {
char* msg = POINTER_SHIFT(pMsg->pCont, sizeof(SMsgHead));
int32_t msgLen = pMsg->contLen - sizeof(SMsgHead);
// deserialize
SStreamCompleteHistoryMsg req = {0};
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t*)msg, msgLen);
tDecodeCompleteHistoryDataMsg(&decoder, &req);
tDecoderClear(&decoder);
SStreamTask* pTask = streamMetaAcquireTask(pMeta, req.streamId, req.upstreamTaskId);
if (pTask == NULL) {
tqError("vgId:%d process scan history finish rsp, failed to find task:0x%x, it may be destroyed",
pMeta->vgId, req.upstreamTaskId);
return -1;
}
int32_t remain = atomic_sub_fetch_32(&pTask->notReadyTasks, 1);
if (remain > 0) {
tqDebug("s-task:%s scan-history finish rsp received from downstream task:0x%x, unfinished remain:%d",
pTask->id.idStr, req.downstreamId, remain);
} else {
tqDebug(
"s-task:%s scan-history finish rsp received from downstream task:0x%x, all downstream tasks rsp scan-history "
"completed msg",
pTask->id.idStr, req.downstreamId);
streamProcessScanHistoryFinishRsp(pTask);
}
streamMetaReleaseTask(pMeta, pTask);
return 0;
}
int32_t tqStreamTaskProcessCheckReq(SStreamMeta* pMeta, SRpcMsg* pMsg) {
char* msgStr = pMsg->pCont;
char* msgBody = POINTER_SHIFT(msgStr, sizeof(SMsgHead));
int32_t msgLen = pMsg->contLen - sizeof(SMsgHead);
SStreamTaskCheckReq req;
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t*)msgBody, msgLen);
tDecodeStreamTaskCheckReq(&decoder, &req);
tDecoderClear(&decoder);
int32_t taskId = req.downstreamTaskId;
SStreamTaskCheckRsp rsp = {
.reqId = req.reqId,
.streamId = req.streamId,
.childId = req.childId,
.downstreamNodeId = req.downstreamNodeId,
.downstreamTaskId = req.downstreamTaskId,
.upstreamNodeId = req.upstreamNodeId,
.upstreamTaskId = req.upstreamTaskId,
};
// only the leader node handle the check request
if (pMeta->role == NODE_ROLE_FOLLOWER) {
tqError("s-task:0x%x invalid check msg from upstream:0x%x(vgId:%d), vgId:%d is follower, not handle check status msg",
taskId, req.upstreamTaskId, req.upstreamNodeId, pMeta->vgId);
rsp.status = TASK_DOWNSTREAM_NOT_LEADER;
} else {
SStreamTask* pTask = streamMetaAcquireTask(pMeta, req.streamId, taskId);
if (pTask != NULL) {
rsp.status = streamTaskCheckStatus(pTask, req.upstreamTaskId, req.upstreamNodeId, req.stage, &rsp.oldStage);
streamMetaReleaseTask(pMeta, pTask);
char* p = NULL;
streamTaskGetStatus(pTask, &p);
tqDebug("s-task:%s status:%s, stage:%"PRId64" recv task check req(reqId:0x%" PRIx64 ") task:0x%x (vgId:%d), check_status:%d",
pTask->id.idStr, p, rsp.oldStage, rsp.reqId, rsp.upstreamTaskId, rsp.upstreamNodeId, rsp.status);
} else {
rsp.status = TASK_DOWNSTREAM_NOT_READY;
tqDebug("tq recv task check(taskId:0x%" PRIx64 "-0x%x not built yet) req(reqId:0x%" PRIx64
") from task:0x%x (vgId:%d), rsp check_status %d",
req.streamId, taskId, rsp.reqId, rsp.upstreamTaskId, rsp.upstreamNodeId, rsp.status);
}
}
return streamSendCheckRsp(pMeta, &req, &rsp, &pMsg->info, taskId);
}
int32_t tqStreamTaskProcessCheckRsp(SStreamMeta* pMeta, SRpcMsg* pMsg, bool isLeader) {
char* pReq = POINTER_SHIFT(pMsg->pCont, sizeof(SMsgHead));
int32_t len = pMsg->contLen - sizeof(SMsgHead);
int32_t vgId = pMeta->vgId;
int32_t code;
SStreamTaskCheckRsp rsp;
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t*)pReq, len);
code = tDecodeStreamTaskCheckRsp(&decoder, &rsp);
if (code < 0) {
terrno = TSDB_CODE_INVALID_MSG;
tDecoderClear(&decoder);
tqError("vgId:%d failed to parse check rsp msg, code:%s", vgId, tstrerror(terrno));
return -1;
}
tDecoderClear(&decoder);
tqDebug("tq task:0x%x (vgId:%d) recv check rsp(reqId:0x%" PRIx64 ") from 0x%x (vgId:%d) status %d",
rsp.upstreamTaskId, rsp.upstreamNodeId, rsp.reqId, rsp.downstreamTaskId, rsp.downstreamNodeId, rsp.status);
if (!isLeader) {
tqError("vgId:%d not leader, task:0x%x not handle the check rsp, downstream:0x%x (vgId:%d)", vgId,
rsp.upstreamTaskId, rsp.downstreamTaskId, rsp.downstreamNodeId);
return code;
}
SStreamTask* pTask = streamMetaAcquireTask(pMeta, rsp.streamId, rsp.upstreamTaskId);
if (pTask == NULL) {
tqError("tq failed to locate the stream task:0x%" PRIx64 "-0x%x (vgId:%d), it may have been destroyed or stopped",
rsp.streamId, rsp.upstreamTaskId, vgId);
terrno = TSDB_CODE_STREAM_TASK_NOT_EXIST;
return -1;
}
code = streamProcessCheckRsp(pTask, &rsp);
streamMetaReleaseTask(pMeta, pTask);
return code;
}
int32_t tqStreamTaskProcessCheckpointReadyMsg(SStreamMeta* pMeta, SRpcMsg* pMsg) {
int32_t vgId = pMeta->vgId;
char* msg = POINTER_SHIFT(pMsg->pCont, sizeof(SMsgHead));
int32_t len = pMsg->contLen - sizeof(SMsgHead);
int32_t code = 0;
SStreamCheckpointReadyMsg req = {0};
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t*)msg, len);
if (tDecodeStreamCheckpointReadyMsg(&decoder, &req) < 0) {
code = TSDB_CODE_MSG_DECODE_ERROR;
tDecoderClear(&decoder);
return code;
}
tDecoderClear(&decoder);
SStreamTask* pTask = streamMetaAcquireTask(pMeta, req.streamId, req.upstreamTaskId);
if (pTask == NULL) {
tqError("vgId:%d failed to find s-task:0x%x, it may have been destroyed already", vgId, req.downstreamTaskId);
return code;
}
tqDebug("vgId:%d s-task:%s received the checkpoint ready msg from task:0x%x (vgId:%d), handle it", vgId,
pTask->id.idStr, req.downstreamTaskId, req.downstreamNodeId);
streamProcessCheckpointReadyMsg(pTask);
streamMetaReleaseTask(pMeta, pTask);
return code;
}
int32_t tqStreamTaskProcessDeployReq(SStreamMeta* pMeta, int64_t sversion, char* msg, int32_t msgLen, bool isLeader, bool restored) {
int32_t code = 0;
int32_t vgId = pMeta->vgId;
if (tsDisableStream) {
tqInfo("vgId:%d stream disabled, not deploy stream tasks", vgId);
return code;
}
tqDebug("vgId:%d receive new stream task deploy msg, start to build stream task", vgId);
// 1.deserialize msg and build task
int32_t size = sizeof(SStreamTask);
SStreamTask* pTask = taosMemoryCalloc(1, size);
if (pTask == NULL) {
tqError("vgId:%d failed to create stream task due to out of memory, alloc size:%d", vgId, size);
return TSDB_CODE_OUT_OF_MEMORY;
}
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t*)msg, msgLen);
code = tDecodeStreamTask(&decoder, pTask);
tDecoderClear(&decoder);
if (code != TSDB_CODE_SUCCESS) {
taosMemoryFree(pTask);
return TSDB_CODE_INVALID_MSG;
}
// 2.save task, use the latest commit version as the initial start version of stream task.
int32_t taskId = pTask->id.taskId;
int64_t streamId = pTask->id.streamId;
bool added = false;
streamMetaWLock(pMeta);
code = streamMetaRegisterTask(pMeta, sversion, pTask, &added);
int32_t numOfTasks = streamMetaGetNumOfTasks(pMeta);
streamMetaWUnLock(pMeta);
if (code < 0) {
tqError("failed to add s-task:0x%x into vgId:%d meta, total:%d, code:%s", vgId, taskId, numOfTasks, tstrerror(code));
tFreeStreamTask(pTask);
return code;
}
// added into meta store, pTask cannot be reference since it may have been destroyed by other threads already now if
// it is added into the meta store
if (added) {
// only handled in the leader node
if (isLeader) {
tqDebug("vgId:%d s-task:0x%x is deployed and add into meta, numOfTasks:%d", vgId, taskId, numOfTasks);
SStreamTask* p = streamMetaAcquireTask(pMeta, streamId, taskId);
if (p != NULL && restored && p->info.fillHistory == 0) {
EStreamTaskEvent event = (HAS_RELATED_FILLHISTORY_TASK(p)) ? TASK_EVENT_INIT_STREAM_SCANHIST : TASK_EVENT_INIT;
streamTaskHandleEvent(p->status.pSM, event);
} else if (!restored) {
tqWarn("s-task:%s not launched since vnode(vgId:%d) not ready", p->id.idStr, vgId);
}
if (p != NULL) {
streamMetaReleaseTask(pMeta, p);
}
} else {
tqDebug("vgId:%d not leader, not launch stream task s-task:0x%x", vgId, taskId);
}
} else {
tqWarn("vgId:%d failed to add s-task:0x%x, since already exists in meta store", vgId, taskId);
tFreeStreamTask(pTask);
}
return code;
}
int32_t tqStreamTaskProcessDropReq(SStreamMeta* pMeta, char* msg, int32_t msgLen) {
SVDropStreamTaskReq* pReq = (SVDropStreamTaskReq*)msg;
int32_t vgId = pMeta->vgId;
tqDebug("vgId:%d receive msg to drop s-task:0x%x", vgId, pReq->taskId);
SStreamTask* pTask = streamMetaAcquireTask(pMeta, pReq->streamId, pReq->taskId);
if (pTask != NULL) {
// drop the related fill-history task firstly
if (HAS_RELATED_FILLHISTORY_TASK(pTask)) {
STaskId* pHTaskId = &pTask->hTaskInfo.id;
streamMetaUnregisterTask(pMeta, pHTaskId->streamId, pHTaskId->taskId);
tqDebug("vgId:%d drop fill-history task:0x%x dropped firstly", vgId, (int32_t)pHTaskId->taskId);
}
streamMetaReleaseTask(pMeta, pTask);
}
// drop the stream task now
streamMetaUnregisterTask(pMeta, pReq->streamId, pReq->taskId);
// commit the update
streamMetaWLock(pMeta);
int32_t numOfTasks = streamMetaGetNumOfTasks(pMeta);
tqDebug("vgId:%d task:0x%x dropped, remain tasks:%d", vgId, pReq->taskId, numOfTasks);
if (streamMetaCommit(pMeta) < 0) {
// persist to disk
}
streamMetaWUnLock(pMeta);
return 0;
}
int32_t startStreamTasks(SStreamMeta* pMeta) {
int32_t code = TSDB_CODE_SUCCESS;
int32_t vgId = pMeta->vgId;
int32_t numOfTasks = taosArrayGetSize(pMeta->pTaskList);
tqDebug("vgId:%d start to check all %d stream task(s) downstream status", vgId, numOfTasks);
if (numOfTasks == 0) {
return TSDB_CODE_SUCCESS;
}
SArray* pTaskList = NULL;
streamMetaWLock(pMeta);
pTaskList = taosArrayDup(pMeta->pTaskList, NULL);
taosHashClear(pMeta->startInfo.pReadyTaskSet);
taosHashClear(pMeta->startInfo.pFailedTaskSet);
pMeta->startInfo.startTs = taosGetTimestampMs();
streamMetaWUnLock(pMeta);
// broadcast the check downstream tasks msg
for (int32_t i = 0; i < numOfTasks; ++i) {
SStreamTaskId* pTaskId = taosArrayGet(pTaskList, i);
SStreamTask* pTask = streamMetaAcquireTask(pMeta, pTaskId->streamId, pTaskId->taskId);
if (pTask == NULL) {
continue;
}
// fill-history task can only be launched by related stream tasks.
if (pTask->info.fillHistory == 1) {
streamMetaReleaseTask(pMeta, pTask);
continue;
}
if (pTask->status.downstreamReady == 1) {
if (HAS_RELATED_FILLHISTORY_TASK(pTask)) {
tqDebug("s-task:%s downstream ready, no need to check downstream, check only related fill-history task",
pTask->id.idStr);
streamLaunchFillHistoryTask(pTask);
}
streamMetaUpdateTaskDownstreamStatus(pTask, pTask->execInfo.init, pTask->execInfo.start, true);
streamMetaReleaseTask(pMeta, pTask);
continue;
}
EStreamTaskEvent event = (HAS_RELATED_FILLHISTORY_TASK(pTask)) ? TASK_EVENT_INIT_STREAM_SCANHIST : TASK_EVENT_INIT;
int32_t ret = streamTaskHandleEvent(pTask->status.pSM, event);
if (ret != TSDB_CODE_SUCCESS) {
code = ret;
}
streamMetaReleaseTask(pMeta, pTask);
}
taosArrayDestroy(pTaskList);
return code;
}
int32_t resetStreamTaskStatus(SStreamMeta* pMeta) {
int32_t vgId = pMeta->vgId;
int32_t numOfTasks = taosArrayGetSize(pMeta->pTaskList);
tqDebug("vgId:%d reset all %d stream task(s) status to be uninit", vgId, numOfTasks);
if (numOfTasks == 0) {
return TSDB_CODE_SUCCESS;
}
for (int32_t i = 0; i < numOfTasks; ++i) {
SStreamTaskId* pTaskId = taosArrayGet(pMeta->pTaskList, i);
STaskId id = {.streamId = pTaskId->streamId, .taskId = pTaskId->taskId};
SStreamTask** pTask = taosHashGet(pMeta->pTasksMap, &id, sizeof(id));
streamTaskResetStatus(*pTask);
}
return 0;
}
static int32_t restartStreamTasks(SStreamMeta* pMeta, bool isLeader) {
int32_t vgId = pMeta->vgId;
int32_t code = 0;
int64_t st = taosGetTimestampMs();
while(1) {
int32_t startVal = atomic_val_compare_exchange_32(&pMeta->startInfo.taskStarting, 0, 1);
if (startVal == 0) {
break;
}
tqDebug("vgId:%d in start stream tasks procedure, wait for 500ms and recheck", vgId);
taosMsleep(500);
}
terrno = 0;
tqInfo("vgId:%d tasks are all updated and stopped, restart all tasks, triggered by transId:%d", vgId,
pMeta->updateInfo.transId);
while (streamMetaTaskInTimer(pMeta)) {
tqDebug("vgId:%d some tasks in timer, wait for 100ms and recheck", pMeta->vgId);
taosMsleep(100);
}
streamMetaWLock(pMeta);
code = streamMetaReopen(pMeta);
if (code != TSDB_CODE_SUCCESS) {
tqError("vgId:%d failed to reopen stream meta", vgId);
streamMetaWUnLock(pMeta);
code = terrno;
return code;
}
streamMetaInitBackend(pMeta);
int64_t el = taosGetTimestampMs() - st;
tqInfo("vgId:%d close&reload state elapsed time:%.3fs", vgId, el/1000.);
code = streamMetaLoadAllTasks(pMeta);
if (code != TSDB_CODE_SUCCESS) {
tqError("vgId:%d failed to load stream tasks, code:%s", vgId, tstrerror(terrno));
streamMetaWUnLock(pMeta);
code = terrno;
return code;
}
if (isLeader && !tsDisableStream) {
tqInfo("vgId:%d restart all stream tasks after all tasks being updated", vgId);
resetStreamTaskStatus(pMeta);
streamMetaWUnLock(pMeta);
startStreamTasks(pMeta);
} else {
streamMetaResetStartInfo(&pMeta->startInfo);
streamMetaWUnLock(pMeta);
tqInfo("vgId:%d, follower node not start stream tasks", vgId);
}
code = terrno;
return code;
}
int32_t tqStreamTaskProcessRunReq(SStreamMeta* pMeta, SRpcMsg* pMsg, bool isLeader) {
SStreamTaskRunReq* pReq = pMsg->pCont;
int32_t taskId = pReq->taskId;
int32_t vgId = pMeta->vgId;
if (taskId == STREAM_EXEC_START_ALL_TASKS_ID) {
startStreamTasks(pMeta);
return 0;
} else if (taskId == STREAM_EXEC_RESTART_ALL_TASKS_ID) {
restartStreamTasks(pMeta, isLeader);
return 0;
}
SStreamTask* pTask = streamMetaAcquireTask(pMeta, pReq->streamId, taskId);
if (pTask != NULL) { // even in halt status, the data in inputQ must be processed
char* p = NULL;
if (streamTaskReadyToRun(pTask, &p)) {
tqDebug("vgId:%d s-task:%s start to process block from inputQ, next checked ver:%" PRId64, vgId, pTask->id.idStr,
pTask->chkInfo.nextProcessVer);
streamExecTask(pTask);
} else {
int8_t status = streamTaskSetSchedStatusInactive(pTask);
tqDebug("vgId:%d s-task:%s ignore run req since not in ready state, status:%s, sched-status:%d", vgId,
pTask->id.idStr, p, status);
}
streamMetaReleaseTask(pMeta, pTask);
return 0;
} else { // NOTE: pTask->status.schedStatus is not updated since it is not be handled by the run exec.
// todo add one function to handle this
tqError("vgId:%d failed to found s-task, taskId:0x%x may have been dropped", vgId, taskId);
return -1;
}
}

6
source/dnode/vnode/src/tsdb/tsdbCache.c
View File

@ -1131,9 +1131,13 @@ int32_t tsdbCacheDel(STsdb *pTsdb, tb_uid_t suid, tb_uid_t uid, TSKEY sKey, TSKE
char **values_list = taosMemoryCalloc(num_keys * 2, sizeof(char *));
size_t *values_list_sizes = taosMemoryCalloc(num_keys * 2, sizeof(size_t));
char **errs = taosMemoryCalloc(num_keys * 2, sizeof(char *));
(void)tsdbCacheCommit(pTsdb);
taosThreadMutexLock(&pTsdb->lruMutex);
taosThreadMutexLock(&pTsdb->rCache.rMutex);
rocksMayWrite(pTsdb, true, false, false);
// rocksMayWrite(pTsdb, true, false, false);
rocksdb_multi_get(pTsdb->rCache.db, pTsdb->rCache.readoptions, num_keys * 2, (const char *const *)keys_list,
keys_list_sizes, values_list, values_list_sizes, errs);
taosThreadMutexUnlock(&pTsdb->rCache.rMutex);

2
source/dnode/vnode/src/tsdb/tsdbCommit2.c
View File

@ -421,7 +421,7 @@ static int32_t tsdbCommitFileSetBegin(SCommitter2 *committer) {
if (mtime < committer->ctx->now - tsS3UploadDelaySec) {
committer->ctx->skipTsRow = true;
}
} else if (s3Size(object_name) > 0) {
} else /*if (s3Size(object_name) > 0) */ {
committer->ctx->skipTsRow = true;
}
}

67
source/dnode/vnode/src/tsdb/tsdbDataFileRW.c
View File

@ -45,7 +45,7 @@ static int32_t tsdbDataFileReadHeadFooter(SDataFileReader *reader) {
int32_t ftype = TSDB_FTYPE_HEAD;
if (reader->fd[ftype]) {
code = tsdbReadFile(reader->fd[ftype], reader->config->files[ftype].file.size - sizeof(SHeadFooter),
(uint8_t *)reader->headFooter, sizeof(SHeadFooter));
(uint8_t *)reader->headFooter, sizeof(SHeadFooter), 0);
TSDB_CHECK_CODE(code, lino, _exit);
}
@ -67,7 +67,7 @@ static int32_t tsdbDataFileReadTombFooter(SDataFileReader *reader) {
int32_t ftype = TSDB_FTYPE_TOMB;
if (reader->fd[ftype]) {
code = tsdbReadFile(reader->fd[ftype], reader->config->files[ftype].file.size - sizeof(STombFooter),
(uint8_t *)reader->tombFooter, sizeof(STombFooter));
(uint8_t *)reader->tombFooter, sizeof(STombFooter), 0);
TSDB_CHECK_CODE(code, lino, _exit);
}
reader->ctx->tombFooterLoaded = true;
@ -161,7 +161,7 @@ int32_t tsdbDataFileReadBrinBlk(SDataFileReader *reader, const TBrinBlkArray **b
}
code = tsdbReadFile(reader->fd[TSDB_FTYPE_HEAD], reader->headFooter->brinBlkPtr->offset, data,
reader->headFooter->brinBlkPtr->size);
reader->headFooter->brinBlkPtr->size, 0);
if (code) {
taosMemoryFree(data);
TSDB_CHECK_CODE(code, lino, _exit);
@ -191,7 +191,8 @@ int32_t tsdbDataFileReadBrinBlock(SDataFileReader *reader, const SBrinBlk *brinB
code = tRealloc(&reader->config->bufArr[0], brinBlk->dp->size);
TSDB_CHECK_CODE(code, lino, _exit);
code = tsdbReadFile(reader->fd[TSDB_FTYPE_HEAD], brinBlk->dp->offset, reader->config->bufArr[0], brinBlk->dp->size);
code =
tsdbReadFile(reader->fd[TSDB_FTYPE_HEAD], brinBlk->dp->offset, reader->config->bufArr[0], brinBlk->dp->size, 0);
TSDB_CHECK_CODE(code, lino, _exit);
int32_t size = 0;
@ -232,7 +233,8 @@ int32_t tsdbDataFileReadBlockData(SDataFileReader *reader, const SBrinRecord *re
code = tRealloc(&reader->config->bufArr[0], record->blockSize);
TSDB_CHECK_CODE(code, lino, _exit);
code = tsdbReadFile(reader->fd[TSDB_FTYPE_DATA], record->blockOffset, reader->config->bufArr[0], record->blockSize);
code =
tsdbReadFile(reader->fd[TSDB_FTYPE_DATA], record->blockOffset, reader->config->bufArr[0], record->blockSize, 0);
TSDB_CHECK_CODE(code, lino, _exit);
code = tDecmprBlockData(reader->config->bufArr[0], record->blockSize, bData, &reader->config->bufArr[1]);
@ -257,8 +259,8 @@ int32_t tsdbDataFileReadBlockDataByColumn(SDataFileReader *reader, const SBrinRe
code = tRealloc(&reader->config->bufArr[0], record->blockKeySize);
TSDB_CHECK_CODE(code, lino, _exit);
code =
tsdbReadFile(reader->fd[TSDB_FTYPE_DATA], record->blockOffset, reader->config->bufArr[0], record->blockKeySize);
code = tsdbReadFile(reader->fd[TSDB_FTYPE_DATA], record->blockOffset, reader->config->bufArr[0], record->blockKeySize,
0);
TSDB_CHECK_CODE(code, lino, _exit);
// hdr
@ -296,10 +298,46 @@ int32_t tsdbDataFileReadBlockDataByColumn(SDataFileReader *reader, const SBrinRe
TSDB_CHECK_CODE(code, lino, _exit);
code = tsdbReadFile(reader->fd[TSDB_FTYPE_DATA], record->blockOffset + record->blockKeySize,
reader->config->bufArr[0], hdr->szBlkCol);
reader->config->bufArr[0], hdr->szBlkCol, 0);
TSDB_CHECK_CODE(code, lino, _exit);
}
int64_t szHint = 0;
if (bData->nColData > 3) {
int64_t offset = 0;
SBlockCol bc = {.cid = 0};
SBlockCol *blockCol = &bc;
size = 0;
SColData *colData = tBlockDataGetColDataByIdx(bData, 0);
while (blockCol && blockCol->cid < colData->cid) {
if (size < hdr->szBlkCol) {
size += tGetBlockCol(reader->config->bufArr[0] + size, blockCol);
} else {
ASSERT(size == hdr->szBlkCol);
blockCol = NULL;
}
}
if (blockCol && blockCol->flag == HAS_VALUE) {
offset = blockCol->offset;
SColData *colDataEnd = tBlockDataGetColDataByIdx(bData, bData->nColData - 1);
while (blockCol && blockCol->cid < colDataEnd->cid) {
if (size < hdr->szBlkCol) {
size += tGetBlockCol(reader->config->bufArr[0] + size, blockCol);
} else {
ASSERT(size == hdr->szBlkCol);
blockCol = NULL;
}
}
if (blockCol && blockCol->flag == HAS_VALUE) {
szHint = blockCol->offset + blockCol->szBitmap + blockCol->szOffset + blockCol->szValue - offset;
}
}
}
SBlockCol bc[1] = {{.cid = 0}};
SBlockCol *blockCol = bc;
@ -338,7 +376,7 @@ int32_t tsdbDataFileReadBlockDataByColumn(SDataFileReader *reader, const SBrinRe
code = tsdbReadFile(reader->fd[TSDB_FTYPE_DATA],
record->blockOffset + record->blockKeySize + hdr->szBlkCol + blockCol->offset,
reader->config->bufArr[1], size1);
reader->config->bufArr[1], size1, i > 0 ? 0 : szHint);
TSDB_CHECK_CODE(code, lino, _exit);
code = tsdbDecmprColData(reader->config->bufArr[1], blockCol, hdr->cmprAlg, hdr->nRow, colData,
@ -366,7 +404,7 @@ int32_t tsdbDataFileReadBlockSma(SDataFileReader *reader, const SBrinRecord *rec
code = tRealloc(&reader->config->bufArr[0], record->smaSize);
TSDB_CHECK_CODE(code, lino, _exit);
code = tsdbReadFile(reader->fd[TSDB_FTYPE_SMA], record->smaOffset, reader->config->bufArr[0], record->smaSize);
code = tsdbReadFile(reader->fd[TSDB_FTYPE_SMA], record->smaOffset, reader->config->bufArr[0], record->smaSize, 0);
TSDB_CHECK_CODE(code, lino, _exit);
// decode sma data
@ -405,7 +443,7 @@ int32_t tsdbDataFileReadTombBlk(SDataFileReader *reader, const TTombBlkArray **t
}
code = tsdbReadFile(reader->fd[TSDB_FTYPE_TOMB], reader->tombFooter->tombBlkPtr->offset, data,
reader->tombFooter->tombBlkPtr->size);
reader->tombFooter->tombBlkPtr->size, 0);
if (code) {
taosMemoryFree(data);
TSDB_CHECK_CODE(code, lino, _exit);
@ -435,7 +473,8 @@ int32_t tsdbDataFileReadTombBlock(SDataFileReader *reader, const STombBlk *tombB
code = tRealloc(&reader->config->bufArr[0], tombBlk->dp->size);
TSDB_CHECK_CODE(code, lino, _exit);
code = tsdbReadFile(reader->fd[TSDB_FTYPE_TOMB], tombBlk->dp->offset, reader->config->bufArr[0], tombBlk->dp->size);
code =
tsdbReadFile(reader->fd[TSDB_FTYPE_TOMB], tombBlk->dp->offset, reader->config->bufArr[0], tombBlk->dp->size, 0);
TSDB_CHECK_CODE(code, lino, _exit);
int32_t size = 0;
@ -488,8 +527,8 @@ struct SDataFileWriter {
STombBlock tombBlock[1];
int32_t tombBlockIdx;
// range
SVersionRange range;
SVersionRange tombRange;
SVersionRange range;
SVersionRange tombRange;
} ctx[1];
STFile files[TSDB_FTYPE_MAX];

2
source/dnode/vnode/src/tsdb/tsdbDef.h
View File

@ -34,7 +34,7 @@ typedef struct SFDataPtr {
extern int32_t tsdbOpenFile(const char *path, STsdb *pTsdb, int32_t flag, STsdbFD **ppFD);
extern void tsdbCloseFile(STsdbFD **ppFD);
extern int32_t tsdbWriteFile(STsdbFD *pFD, int64_t offset, const uint8_t *pBuf, int64_t size);
extern int32_t tsdbReadFile(STsdbFD *pFD, int64_t offset, uint8_t *pBuf, int64_t size);
extern int32_t tsdbReadFile(STsdbFD *pFD, int64_t offset, uint8_t *pBuf, int64_t size, int64_t szHint);
extern int32_t tsdbFsyncFile(STsdbFD *pFD);
#ifdef __cplusplus

197
source/dnode/vnode/src/tsdb/tsdbFS2.c
View File

@ -20,8 +20,6 @@
#define BLOCK_COMMIT_FACTOR 3
extern int vnodeScheduleTask(int (*execute)(void *), void *arg);
extern int vnodeScheduleTaskEx(int tpid, int (*execute)(void *), void *arg);
extern void remove_file(const char *fname, bool last_level);
#define TSDB_FS_EDIT_MIN TSDB_FEDIT_COMMIT
@ -651,7 +649,6 @@ _exit:
static int32_t close_file_system(STFileSystem *fs) {
TARRAY2_CLEAR(fs->fSetArr, tsdbTFileSetClear);
TARRAY2_CLEAR(fs->fSetArrTmp, tsdbTFileSetClear);
// TODO
return 0;
}
@ -748,36 +745,31 @@ _exit:
return code;
}
static void tsdbDoWaitBgTask(STFileSystem *fs, STFSBgTask *task) {
task->numWait++;
taosThreadCondWait(task->done, &fs->tsdb->mutex);
task->numWait--;
int32_t tsdbFSCancelAllBgTask(STFileSystem *fs) {
TARRAY2(int64_t) channelArr = {0};
if (task->numWait == 0) {
taosThreadCondDestroy(task->done);
if (task->destroy) {
task->destroy(task->arg);
// collect all open channels
taosThreadMutexLock(&fs->tsdb->mutex);
STFileSet *fset;
TARRAY2_FOREACH(fs->fSetArr, fset) {
if (VNODE_ASYNC_VALID_CHANNEL_ID(fset->bgTaskChannel)) {
TARRAY2_APPEND(&channelArr, fset->bgTaskChannel);
fset->bgTaskChannel = 0;
}
taosMemoryFree(task);
}
}
taosThreadMutexUnlock(&fs->tsdb->mutex);
static void tsdbDoDoneBgTask(STFileSystem *fs, STFSBgTask *task) {
if (task->numWait > 0) {
taosThreadCondBroadcast(task->done);
} else {
taosThreadCondDestroy(task->done);
if (task->destroy) {
task->destroy(task->arg);
}
taosMemoryFree(task);
}
// destroy all channels
int64_t channel;
TARRAY2_FOREACH(&channelArr, channel) { vnodeAChannelDestroy(vnodeAsyncHandle[1], channel, true); }
TARRAY2_DESTROY(&channelArr, NULL);
return 0;
}
int32_t tsdbCloseFS(STFileSystem **fs) {
if (fs[0] == NULL) return 0;
tsdbFSDisableBgTask(fs[0]);
tsdbFSCancelAllBgTask(*fs);
close_file_system(fs[0]);
destroy_fs(fs);
return 0;
@ -901,7 +893,7 @@ int32_t tsdbFSEditCommit(STFileSystem *fs) {
if (mtime < now - tsS3UploadDelaySec) {
skipMerge = true;
}
} else if (s3Size(object_name) > 0) {
} else /* if (s3Size(object_name) > 0) */ {
skipMerge = true;
}
}
@ -910,7 +902,20 @@ int32_t tsdbFSEditCommit(STFileSystem *fs) {
}
if (!skipMerge) {
code = tsdbSchedMerge(fs->tsdb, fset->fid);
code = tsdbTFileSetOpenChannel(fset);
TSDB_CHECK_CODE(code, lino, _exit);
SMergeArg *arg = taosMemoryMalloc(sizeof(*arg));
if (arg == NULL) {
code = TSDB_CODE_OUT_OF_MEMORY;
TSDB_CHECK_CODE(code, lino, _exit);
}
arg->tsdb = fs->tsdb;
arg->fid = fset->fid;
code = vnodeAsyncC(vnodeAsyncHandle[1], fset->bgTaskChannel, EVA_PRIORITY_HIGH, tsdbMerge, taosMemoryFree,
arg, NULL);
TSDB_CHECK_CODE(code, lino, _exit);
}
}
@ -939,7 +944,11 @@ int32_t tsdbFSEditCommit(STFileSystem *fs) {
}
}
if (tsdbTFileSetIsEmpty(fset) && fset->bgTaskRunning == NULL) {
if (tsdbTFileSetIsEmpty(fset)) {
if (VNODE_ASYNC_VALID_CHANNEL_ID(fset->bgTaskChannel)) {
vnodeAChannelDestroy(vnodeAsyncHandle[1], fset->bgTaskChannel, false);
fset->bgTaskChannel = 0;
}
TARRAY2_REMOVE(fs->fSetArr, i, tsdbTFileSetClear);
} else {
i++;
@ -1179,136 +1188,4 @@ _out:
pHash = NULL;
}
return code;
}
const char *gFSBgTaskName[] = {NULL, "MERGE", "RETENTION", "COMPACT"};
static int32_t tsdbFSRunBgTask(void *arg) {
STFSBgTask *task = (STFSBgTask *)arg;
STFileSystem *fs = task->fs;
task->launchTime = taosGetTimestampMs();
task->run(task->arg);
task->finishTime = taosGetTimestampMs();
tsdbDebug("vgId:%d bg task:%s task id:%" PRId64 " finished, schedule time:%" PRId64 " launch time:%" PRId64
" finish time:%" PRId64,
TD_VID(fs->tsdb->pVnode), gFSBgTaskName[task->type], task->taskid, task->scheduleTime, task->launchTime,
task->finishTime);
taosThreadMutexLock(&fs->tsdb->mutex);
STFileSet *fset = NULL;
tsdbFSGetFSet(fs, task->fid, &fset);
ASSERT(fset != NULL && fset->bgTaskRunning == task);
// free last
tsdbDoDoneBgTask(fs, task);
fset->bgTaskRunning = NULL;
// schedule next
if (fset->bgTaskNum > 0) {
if (fs->stop) {
while (fset->bgTaskNum > 0) {
STFSBgTask *nextTask = fset->bgTaskQueue->next;
nextTask->prev->next = nextTask->next;
nextTask->next->prev = nextTask->prev;
fset->bgTaskNum--;
tsdbDoDoneBgTask(fs, nextTask);
}
} else {
// pop task from head
fset->bgTaskRunning = fset->bgTaskQueue->next;
fset->bgTaskRunning->prev->next = fset->bgTaskRunning->next;
fset->bgTaskRunning->next->prev = fset->bgTaskRunning->prev;
fset->bgTaskNum--;
vnodeScheduleTaskEx(1, tsdbFSRunBgTask, fset->bgTaskRunning);
}
}
taosThreadMutexUnlock(&fs->tsdb->mutex);
return 0;
}
// IMPORTANT: the caller must hold the fs->tsdb->mutex
int32_t tsdbFSScheduleBgTask(STFileSystem *fs, int32_t fid, EFSBgTaskT type, int32_t (*run)(void *),
void (*destroy)(void *), void *arg, int64_t *taskid) {
if (fs->stop) {
if (destroy) {
destroy(arg);
}
return 0;
}
STFileSet *fset;
tsdbFSGetFSet(fs, fid, &fset);
ASSERT(fset != NULL);
for (STFSBgTask *task = fset->bgTaskQueue->next; task != fset->bgTaskQueue; task = task->next) {
if (task->type == type) {
if (destroy) {
destroy(arg);
}
return 0;
}
}
// do schedule task
STFSBgTask *task = taosMemoryCalloc(1, sizeof(STFSBgTask));
if (task == NULL) return TSDB_CODE_OUT_OF_MEMORY;
taosThreadCondInit(task->done, NULL);
task->fs = fs;
task->fid = fid;
task->type = type;
task->run = run;
task->destroy = destroy;
task->arg = arg;
task->scheduleTime = taosGetTimestampMs();
task->taskid = ++fs->taskid;
if (fset->bgTaskRunning == NULL && fset->bgTaskNum == 0) {
// launch task directly
fset->bgTaskRunning = task;
vnodeScheduleTaskEx(1, tsdbFSRunBgTask, task);
} else {
// add to the queue tail
fset->bgTaskNum++;
task->next = fset->bgTaskQueue;
task->prev = fset->bgTaskQueue->prev;
task->prev->next = task;
task->next->prev = task;
}
if (taskid) *taskid = task->taskid;
return 0;
}
int32_t tsdbFSDisableBgTask(STFileSystem *fs) {
taosThreadMutexLock(&fs->tsdb->mutex);
for (;;) {
fs->stop = true;
bool done = true;
STFileSet *fset;
TARRAY2_FOREACH(fs->fSetArr, fset) {
if (fset->bgTaskRunning) {
tsdbDoWaitBgTask(fs, fset->bgTaskRunning);
done = false;
break;
}
}
if (done) break;
}
taosThreadMutexUnlock(&fs->tsdb->mutex);
return 0;
}
int32_t tsdbFSEnableBgTask(STFileSystem *fs) {
taosThreadMutexLock(&fs->tsdb->mutex);
fs->stop = false;
taosThreadMutexUnlock(&fs->tsdb->mutex);
return 0;
}
}

5
source/dnode/vnode/src/tsdb/tsdbFS2.h
View File

@ -55,11 +55,6 @@ int64_t tsdbFSAllocEid(STFileSystem *fs);
int32_t tsdbFSEditBegin(STFileSystem *fs, const TFileOpArray *opArray, EFEditT etype);
int32_t tsdbFSEditCommit(STFileSystem *fs);
int32_t tsdbFSEditAbort(STFileSystem *fs);
// background task
int32_t tsdbFSScheduleBgTask(STFileSystem *fs, int32_t fid, EFSBgTaskT type, int32_t (*run)(void *),
void (*destroy)(void *), void *arg, int64_t *taskid);
int32_t tsdbFSDisableBgTask(STFileSystem *fs);
int32_t tsdbFSEnableBgTask(STFileSystem *fs);
// other
int32_t tsdbFSGetFSet(STFileSystem *fs, int32_t fid, STFileSet **fset);
int32_t tsdbFSCheckCommit(STsdb *tsdb, int32_t fid);

11
source/dnode/vnode/src/tsdb/tsdbFSet2.c
View File

@ -14,6 +14,7 @@
*/
#include "tsdbFSet2.h"
#include "vnd.h"
int32_t tsdbSttLvlInit(int32_t level, SSttLvl **lvl) {
if (!(lvl[0] = taosMemoryMalloc(sizeof(SSttLvl)))) return TSDB_CODE_OUT_OF_MEMORY;
@ -451,10 +452,7 @@ int32_t tsdbTFileSetInit(int32_t fid, STFileSet **fset) {
TARRAY2_INIT(fset[0]->lvlArr);
// background task queue
fset[0]->bgTaskNum = 0;
fset[0]->bgTaskQueue->next = fset[0]->bgTaskQueue;
fset[0]->bgTaskQueue->prev = fset[0]->bgTaskQueue;
fset[0]->bgTaskRunning = NULL;
fset[0]->bgTaskChannel = 0;
// block commit variables
taosThreadCondInit(&fset[0]->canCommit, NULL);
@ -650,3 +648,8 @@ bool tsdbTFileSetIsEmpty(const STFileSet *fset) {
}
return TARRAY2_SIZE(fset->lvlArr) == 0;
}
int32_t tsdbTFileSetOpenChannel(STFileSet *fset) {
if (VNODE_ASYNC_VALID_CHANNEL_ID(fset->bgTaskChannel)) return 0;
return vnodeAChannelInit(vnodeAsyncHandle[1], &fset->bgTaskChannel);
}

41
source/dnode/vnode/src/tsdb/tsdbFSet2.h
View File

@ -22,14 +22,12 @@
extern "C" {
#endif
typedef struct STFileSet STFileSet;
typedef struct STFileOp STFileOp;
typedef struct SSttLvl SSttLvl;
typedef struct STFileOp STFileOp;
typedef struct SSttLvl SSttLvl;
typedef TARRAY2(STFileObj *) TFileObjArray;
typedef TARRAY2(SSttLvl *) TSttLvlArray;
typedef TARRAY2(STFileOp) TFileOpArray;
typedef struct STFileSystem STFileSystem;
typedef struct STFSBgTask STFSBgTask;
typedef enum {
TSDB_FOP_NONE = 0,
@ -72,33 +70,8 @@ bool tsdbTFileSetIsEmpty(const STFileSet *fset);
// stt
int32_t tsdbSttLvlInit(int32_t level, SSttLvl **lvl);
int32_t tsdbSttLvlClear(SSttLvl **lvl);
typedef enum {
TSDB_BG_TASK_MERGER = 1,
TSDB_BG_TASK_RETENTION,
TSDB_BG_TASK_COMPACT,
} EFSBgTaskT;
struct STFSBgTask {
STFileSystem *fs;
int32_t fid;
EFSBgTaskT type;
int32_t (*run)(void *arg);
void (*destroy)(void *arg);
void *arg;
TdThreadCond done[1];
int32_t numWait;
int64_t taskid;
int64_t scheduleTime;
int64_t launchTime;
int64_t finishTime;
struct STFSBgTask *prev;
struct STFSBgTask *next;
};
// open channel
int32_t tsdbTFileSetOpenChannel(STFileSet *fset);
struct STFileOp {
tsdb_fop_t optype;
@ -118,10 +91,8 @@ struct STFileSet {
STFileObj *farr[TSDB_FTYPE_MAX]; // file array
TSttLvlArray lvlArr[1]; // level array
// background task queue
int32_t bgTaskNum;
STFSBgTask bgTaskQueue[1];
STFSBgTask *bgTaskRunning;
// background task channel
int64_t bgTaskChannel;
// block commit variables
TdThreadCond canCommit;

24
source/dnode/vnode/src/tsdb/tsdbMerge.c
View File

@ -17,11 +17,6 @@
#define TSDB_MAX_LEVEL 2 // means max level is 3
typedef struct {
STsdb *tsdb;
int32_t fid;
} SMergeArg;
typedef struct {
STsdb *tsdb;
int32_t fid;
@ -528,7 +523,7 @@ static int32_t tsdbMergeGetFSet(SMerger *merger) {
return 0;
}
static int32_t tsdbMerge(void *arg) {
int32_t tsdbMerge(void *arg) {
int32_t code = 0;
int32_t lino = 0;
SMergeArg *mergeArg = (SMergeArg *)arg;
@ -568,7 +563,7 @@ static int32_t tsdbMerge(void *arg) {
if (mtime < now - tsS3UploadDelaySec) {
skipMerge = true;
}
} else if (s3Size(object_name) > 0) {
} else /* if (s3Size(object_name) > 0) */ {
skipMerge = true;
}
}
@ -597,18 +592,3 @@ _exit:
tsdbTFileSetClear(&merger->fset);
return code;
}
int32_t tsdbSchedMerge(STsdb *tsdb, int32_t fid) {
SMergeArg *arg = taosMemoryMalloc(sizeof(*arg));
if (arg == NULL) {
return TSDB_CODE_OUT_OF_MEMORY;
}
arg->tsdb = tsdb;
arg->fid = fid;
int32_t code = tsdbFSScheduleBgTask(tsdb->pFS, fid, TSDB_BG_TASK_MERGER, tsdbMerge, taosMemoryFree, arg, NULL);
if (code) taosMemoryFree(arg);
return code;
}

56
source/dnode/vnode/src/tsdb/tsdbReaderWriter.c
View File

@ -26,7 +26,17 @@ static int32_t tsdbOpenFileImpl(STsdbFD *pFD) {
if (pFD->pFD == NULL) {
int errsv = errno;
const char *object_name = taosDirEntryBaseName((char *)path);
long s3_size = tsS3Enabled ? s3Size(object_name) : 0;
long s3_size = 0;
if (tsS3Enabled) {
long size = s3Size(object_name);
if (size < 0) {
code = terrno = TSDB_CODE_FAILED_TO_CONNECT_S3;
goto _exit;
}
s3_size = size;
}
if (tsS3Enabled && !strncmp(path + strlen(path) - 5, ".data", 5) && s3_size > 0) {
#ifndef S3_BLOCK_CACHE
s3EvictCache(path, s3_size);
@ -48,6 +58,7 @@ static int32_t tsdbOpenFileImpl(STsdbFD *pFD) {
// pFD->szFile = s3_size;
#endif
} else {
tsdbInfo("no file: %s", path);
code = TAOS_SYSTEM_ERROR(errsv);
// taosMemoryFree(pFD);
goto _exit;
@ -283,7 +294,7 @@ _exit:
return code;
}
static int32_t tsdbReadFileS3(STsdbFD *pFD, int64_t offset, uint8_t *pBuf, int64_t size) {
static int32_t tsdbReadFileS3(STsdbFD *pFD, int64_t offset, uint8_t *pBuf, int64_t size, int64_t szHint) {
int32_t code = 0;
int64_t n = 0;
int32_t szPgCont = PAGE_CONTENT_SIZE(pFD->szPage);
@ -330,7 +341,7 @@ static int32_t tsdbReadFileS3(STsdbFD *pFD, int64_t offset, uint8_t *pBuf, int64
memcpy(pBuf + n, pFD->pBuf + bOffset, nRead);
n += nRead;
pgno++;
++pgno;
bOffset = 0;
}
@ -339,7 +350,12 @@ static int32_t tsdbReadFileS3(STsdbFD *pFD, int64_t offset, uint8_t *pBuf, int64
uint8_t *pBlock = NULL;
int64_t retrieve_offset = PAGE_OFFSET(pgno, pFD->szPage);
int64_t pgnoEnd = pgno - 1 + (bOffset + size - n + szPgCont - 1) / szPgCont;
int64_t retrieve_size = (pgnoEnd - pgno + 1) * pFD->szPage;
if (szHint > 0) {
pgnoEnd = pgno - 1 + (bOffset + szHint - n + szPgCont - 1) / szPgCont;
}
int64_t retrieve_size = (pgnoEnd - pgno + 1) * pFD->szPage;
code = s3GetObjectBlock(pFD->objName, retrieve_offset, retrieve_size, 1, &pBlock);
if (code != TSDB_CODE_SUCCESS) {
goto _exit;
@ -350,6 +366,10 @@ static int32_t tsdbReadFileS3(STsdbFD *pFD, int64_t offset, uint8_t *pBuf, int64
for (int i = 0; i < nPage; ++i) {
tsdbCacheSetPageS3(pFD->pTsdb->pgCache, pFD, pgno, pBlock + i * pFD->szPage);
if (szHint > 0 && n >= size) {
++pgno;
continue;
}
memcpy(pFD->pBuf, pBlock + i * pFD->szPage, pFD->szPage);
// check
@ -364,7 +384,7 @@ static int32_t tsdbReadFileS3(STsdbFD *pFD, int64_t offset, uint8_t *pBuf, int64
memcpy(pBuf + n, pFD->pBuf + bOffset, nRead);
n += nRead;
pgno++;
++pgno;
bOffset = 0;
}
@ -375,7 +395,7 @@ _exit:
return code;
}
int32_t tsdbReadFile(STsdbFD *pFD, int64_t offset, uint8_t *pBuf, int64_t size) {
int32_t tsdbReadFile(STsdbFD *pFD, int64_t offset, uint8_t *pBuf, int64_t size, int64_t szHint) {
int32_t code = 0;
if (!pFD->pFD) {
code = tsdbOpenFileImpl(pFD);
@ -385,7 +405,7 @@ int32_t tsdbReadFile(STsdbFD *pFD, int64_t offset, uint8_t *pBuf, int64_t size)
}
if (pFD->s3File && tsS3BlockSize < 0) {
return tsdbReadFileS3(pFD, offset, pBuf, size);
return tsdbReadFileS3(pFD, offset, pBuf, size, szHint);
} else {
return tsdbReadFileImp(pFD, offset, pBuf, size);
}
@ -1141,7 +1161,7 @@ int32_t tsdbReadBlockIdx(SDataFReader *pReader, SArray *aBlockIdx) {
if (code) goto _err;
// read
code = tsdbReadFile(pReader->pHeadFD, offset, pReader->aBuf[0], size);
code = tsdbReadFile(pReader->pHeadFD, offset, pReader->aBuf[0], size, 0);
if (code) goto _err;
// decode
@ -1178,7 +1198,7 @@ int32_t tsdbReadSttBlk(SDataFReader *pReader, int32_t iStt, SArray *aSttBlk) {
if (code) goto _err;
// read
code = tsdbReadFile(pReader->aSttFD[iStt], offset, pReader->aBuf[0], size);
code = tsdbReadFile(pReader->aSttFD[iStt], offset, pReader->aBuf[0], size, 0);
if (code) goto _err;
// decode
@ -1211,7 +1231,7 @@ int32_t tsdbReadDataBlk(SDataFReader *pReader, SBlockIdx *pBlockIdx, SMapData *m
if (code) goto _err;
// read
code = tsdbReadFile(pReader->pHeadFD, offset, pReader->aBuf[0], size);
code = tsdbReadFile(pReader->pHeadFD, offset, pReader->aBuf[0], size, 0);
if (code) goto _err;
// decode
@ -1242,7 +1262,7 @@ int32_t tsdbReadBlockSma(SDataFReader *pReader, SDataBlk *pDataBlk, SArray *aCol
if (code) goto _err;
// read
code = tsdbReadFile(pReader->pSmaFD, pSmaInfo->offset, pReader->aBuf[0], pSmaInfo->size);
code = tsdbReadFile(pReader->pSmaFD, pSmaInfo->offset, pReader->aBuf[0], pSmaInfo->size, 0);
if (code) goto _err;
// decode
@ -1276,7 +1296,7 @@ static int32_t tsdbReadBlockDataImpl(SDataFReader *pReader, SBlockInfo *pBlkInfo
code = tRealloc(&pReader->aBuf[0], pBlkInfo->szKey);
if (code) goto _err;
code = tsdbReadFile(pFD, pBlkInfo->offset, pReader->aBuf[0], pBlkInfo->szKey);
code = tsdbReadFile(pFD, pBlkInfo->offset, pReader->aBuf[0], pBlkInfo->szKey, 0);
if (code) goto _err;
SDiskDataHdr hdr;
@ -1322,7 +1342,7 @@ static int32_t tsdbReadBlockDataImpl(SDataFReader *pReader, SBlockInfo *pBlkInfo
code = tRealloc(&pReader->aBuf[0], hdr.szBlkCol);
if (code) goto _err;
code = tsdbReadFile(pFD, offset, pReader->aBuf[0], hdr.szBlkCol);
code = tsdbReadFile(pFD, offset, pReader->aBuf[0], hdr.szBlkCol, 0);
if (code) goto _err;
}
@ -1366,7 +1386,7 @@ static int32_t tsdbReadBlockDataImpl(SDataFReader *pReader, SBlockInfo *pBlkInfo
code = tRealloc(&pReader->aBuf[1], size);
if (code) goto _err;
code = tsdbReadFile(pFD, offset, pReader->aBuf[1], size);
code = tsdbReadFile(pFD, offset, pReader->aBuf[1], size, 0);
if (code) goto _err;
code = tsdbDecmprColData(pReader->aBuf[1], pBlockCol, hdr.cmprAlg, hdr.nRow, pColData, &pReader->aBuf[2]);
@ -1392,7 +1412,7 @@ int32_t tsdbReadDataBlockEx(SDataFReader *pReader, SDataBlk *pDataBlk, SBlockDat
if (code) goto _err;
// read
code = tsdbReadFile(pReader->pDataFD, pBlockInfo->offset, pReader->aBuf[0], pBlockInfo->szBlock);
code = tsdbReadFile(pReader->pDataFD, pBlockInfo->offset, pReader->aBuf[0], pBlockInfo->szBlock, 0);
if (code) goto _err;
// decmpr
@ -1444,7 +1464,7 @@ int32_t tsdbReadSttBlockEx(SDataFReader *pReader, int32_t iStt, SSttBlk *pSttBlk
TSDB_CHECK_CODE(code, lino, _exit);
// read
code = tsdbReadFile(pReader->aSttFD[iStt], pSttBlk->bInfo.offset, pReader->aBuf[0], pSttBlk->bInfo.szBlock);
code = tsdbReadFile(pReader->aSttFD[iStt], pSttBlk->bInfo.offset, pReader->aBuf[0], pSttBlk->bInfo.szBlock, 0);
TSDB_CHECK_CODE(code, lino, _exit);
// decmpr
@ -1700,7 +1720,7 @@ int32_t tsdbReadDelDatav1(SDelFReader *pReader, SDelIdx *pDelIdx, SArray *aDelDa
if (code) goto _err;
// read
code = tsdbReadFile(pReader->pReadH, offset, pReader->aBuf[0], size);
code = tsdbReadFile(pReader->pReadH, offset, pReader->aBuf[0], size, 0);
if (code) goto _err;
// // decode
@ -1740,7 +1760,7 @@ int32_t tsdbReadDelIdx(SDelFReader *pReader, SArray *aDelIdx) {
if (code) goto _err;
// read
code = tsdbReadFile(pReader->pReadH, offset, pReader->aBuf[0], size);
code = tsdbReadFile(pReader->pReadH, offset, pReader->aBuf[0], size, 0);
if (code) goto _err;
// decode

Some files were not shown because too many files have changed in this diff Show More