Merge pull request #23019 from taosdata/case/TD-26408-3.0

fix: add case for lost data after split 3.0

source/libs/stream/src/streamBackendRocksdb.c

@@ -779,7 +779,10 @@ int32_t chkpGetAllDbCfHandle(SStreamMeta* pMeta, rocksdb_column_family_handle_t*
     int64_t id = *(int64_t*)pIter;
 
     SBackendCfWrapper* wrapper = taosAcquireRef(streamBackendCfWrapperId, id);
-    if (wrapper == NULL) continue;
+    if (wrapper == NULL) {
+      pIter = taosHashIterate(pMeta->pTaskBackendUnique, pIter);
+      continue;
+    }
 
     taosThreadRwlockRdlock(&wrapper->rwLock);
     for (int i = 0; i < sizeof(ginitDict) / sizeof(ginitDict[0]); i++) {
@@ -795,6 +798,10 @@ int32_t chkpGetAllDbCfHandle(SStreamMeta* pMeta, rocksdb_column_family_handle_t*
   }
 
   int32_t nCf = taosArrayGetSize(pHandle);
+  if (nCf == 0) {
+    taosArrayDestroy(pHandle);
+    return nCf;
+  }
 
   rocksdb_column_family_handle_t** ppCf = taosMemoryCalloc(nCf, sizeof(rocksdb_column_family_handle_t*));
   for (int i = 0; i < nCf; i++) {
@@ -827,6 +834,7 @@ _ERROR:
   return code;
 }
 int32_t chkpPreFlushDb(rocksdb_t* db, rocksdb_column_family_handle_t** cf, int32_t nCf) {
+  if (nCf == 0) return 0;
   int   code = 0;
   char* err = NULL;
 

source/libs/stream/src/streamMeta.c

@@ -238,8 +238,6 @@ int32_t streamMetaReopen(SStreamMeta* pMeta, int64_t chkpId) {
       return -1;
     }
   }
-  taosMemoryFree(defaultPath);
-  taosMemoryFree(newPath);
 
   pMeta->streamBackend = streamBackendInit(pMeta->path, pMeta->chkpId);
   while (pMeta->streamBackend == NULL) {
@@ -254,6 +252,8 @@ int32_t streamMetaReopen(SStreamMeta* pMeta, int64_t chkpId) {
   pMeta->streamBackendRid = taosAddRef(streamBackendId, pMeta->streamBackend);
   streamBackendLoadCheckpointInfo(pMeta);
 
+  taosMemoryFree(defaultPath);
+  taosMemoryFree(newPath);
   return 0;
 }
 

tests/parallel_test/cases.task

@@ -207,7 +207,8 @@
 ,,n,system-test,python3 ./test.py -f 0-others/tag_index_basic.py
 ,,n,system-test,python3 ./test.py -f 0-others/udfpy_main.py
 ,,n,system-test,python3 ./test.py -N 3 -f 0-others/walRetention.py
-#,,n,system-test,python3 ./test.py -f 0-others/splitVGroup.py -N 5
+,,y,system-test,./pytest.sh python3 ./test.py -f 0-others/splitVGroupRep1.py -N 3
+,,y,system-test,./pytest.sh python3 ./test.py -f 0-others/splitVGroupRep3.py -N 3
 ,,n,system-test,python3 ./test.py -f 0-others/timeRangeWise.py -N 3
 ,,y,system-test,./pytest.sh python3 ./test.py -f 1-insert/alter_database.py
 ,,y,system-test,./pytest.sh python3 ./test.py -f 1-insert/alter_replica.py -N 3

tests/system-test/0-others/splitVGroupRep1.py

@@ -0,0 +1,440 @@
+###################################################################
+#           Copyright (c) 2016 by TAOS Technologies, Inc.
+#                     All rights reserved.
+#
+#  This file is proprietary and confidential to TAOS Technologies.
+#  No part of this file may be reproduced, stored, transmitted,
+#  disclosed or used in any form or by any means other than as
+#  expressly provided by the written permission from Jianhui Tao
+#
+###################################################################
+
+# -*- coding: utf-8 -*-
+
+import sys
+import random
+import time
+import copy
+import string
+
+import taos
+from util.log import *
+from util.cases import *
+from util.sql import *
+
+class TDTestCase:
+
+    # random string
+    def random_string(self, count):
+        letters = string.ascii_letters
+        return ''.join(random.choice(letters) for i in range(count))
+
+    # get col value and total max min ...
+    def getColsValue(self, i, j):
+        # c1 value
+        if random.randint(1, 10) == 5:
+            c1 = None
+        else:
+            c1 = 1
+
+        # c2 value
+        if j % 3200 == 0:
+            c2 = 8764231
+        elif random.randint(1, 10) == 5:
+            c2 = None
+        else:
+            c2 = random.randint(-87654297, 98765321)    
+
+
+        value = f"({self.ts}, "
+
+        # c1
+        if c1 is None:
+            value += "null,"
+        else:
+            self.c1Cnt += 1
+            value += f"{c1},"
+        # c2
+        if c2 is None:
+            value += "null,"
+        else:
+            value += f"{c2},"
+            # total count
+            self.c2Cnt += 1
+            # max
+            if self.c2Max is None:
+                self.c2Max = c2
+            else:
+                if c2 > self.c2Max:
+                    self.c2Max = c2
+            # min
+            if self.c2Min is None:
+                self.c2Min = c2
+            else:
+                if c2 < self.c2Min:
+                    self.c2Min = c2
+            # sum
+            if self.c2Sum is None:
+                self.c2Sum = c2
+            else:
+                self.c2Sum += c2
+
+        # c3 same with ts
+        value += f"{self.ts})"
+        
+        # move next
+        self.ts += 1
+
+        return value
+
+    # insert data
+    def insertData(self):
+        tdLog.info("insert data ....")
+        sqls = ""
+        for i in range(self.childCnt):
+            # insert child table
+            values = ""
+            pre_insert = f"insert into @db_name.t{i} values "
+            for j in range(self.childRow):
+                if values == "":
+                    values = self.getColsValue(i, j)
+                else:
+                    values += "," + self.getColsValue(i, j)
+
+                # batch insert    
+                if j % self.batchSize == 0  and values != "":
+                    sql = pre_insert + values
+                    self.exeDouble(sql)
+                    values = ""
+            # append last
+            if values != "":
+                sql = pre_insert + values
+                self.exeDouble(sql)
+                values = ""
+
+        # insert nomal talbe
+        for i in range(20):
+            self.ts += 1000
+            name = self.random_string(20)
+            sql = f"insert into @db_name.ta values({self.ts}, {i}, {self.ts%100000}, '{name}', false)"
+            self.exeDouble(sql)
+
+        # insert finished
+        tdLog.info(f"insert data successfully.\n"
+        f"                            inserted child table = {self.childCnt}\n"
+        f"                            inserted child rows  = {self.childRow}\n"
+        f"                            total inserted rows  = {self.childCnt*self.childRow}\n")
+        return
+    
+    def exeDouble(self, sql):
+        # dbname replace
+        sql1 = sql.replace("@db_name", self.db1)
+
+        if len(sql1) > 100:
+            tdLog.info(sql1[:100])
+        else:
+            tdLog.info(sql1)
+        tdSql.execute(sql1)
+
+        sql2 = sql.replace("@db_name", self.db2)
+        if len(sql2) > 100:
+            tdLog.info(sql2[:100])
+        else:
+            tdLog.info(sql2)
+        tdSql.execute(sql2)
+        
+
+    # prepareEnv
+    def prepareEnv(self):
+        # init                
+        self.ts = 1680000000000
+        self.childCnt = 4
+        self.childRow = 10000
+        self.batchSize = 50000
+        self.vgroups1  = 1
+        self.vgroups2  = 1
+        self.db1 = "db1"
+        self.db2 = "db2"
+        
+        # total
+        self.c1Cnt = 0
+        self.c2Cnt = 0
+        self.c2Max = None
+        self.c2Min = None
+        self.c2Sum = None
+
+        # create database  db
+        sql = f"create database @db_name vgroups {self.vgroups1} replica 1"
+        self.exeDouble(sql)
+
+        # create super talbe st
+        sql = f"create table @db_name.st(ts timestamp, c1 int, c2 bigint, ts1 timestamp) tags(area int)"
+        self.exeDouble(sql)
+
+        # create child table
+        for i in range(self.childCnt):
+            sql = f"create table @db_name.t{i} using @db_name.st tags({i}) "
+            self.exeDouble(sql)
+
+        # create normal table
+        sql = f"create table @db_name.ta(ts timestamp, c1 int, c2 bigint, c3 binary(32), c4 bool)"
+        self.exeDouble(sql)
+
+        # insert data
+        self.insertData()
+
+        # update
+        self.ts = 1680000000000 + 20000
+        self.childRow = 1000
+
+
+        # delete data
+        sql = "delete from @db_name.st where ts > 1680000019000 and ts < 1680000062000"
+        self.exeDouble(sql)
+        sql = "delete from @db_name.st where ts > 1680000099000 and ts < 1680000170000"
+        self.exeDouble(sql)    
+
+    # check data correct
+    def checkExpect(self, sql, expectVal):
+        tdSql.query(sql)
+        rowCnt = tdSql.getRows()
+        for i in range(rowCnt):
+            val = tdSql.getData(i,0)
+            if val != expectVal:
+                tdLog.exit(f"Not expect . query={val} expect={expectVal} i={i} sql={sql}")
+                return False
+
+        tdLog.info(f"check expect ok. sql={sql} expect ={expectVal} rowCnt={rowCnt}")
+        return True
+
+    # init
+    def init(self, conn, logSql, replicaVar=1):
+        seed = time.time() % 10000
+        random.seed(seed)
+        self.replicaVar = int(replicaVar)
+        tdLog.debug(f"start to excute {__file__}")
+        tdSql.init(conn.cursor(), True)
+
+    # check query result same
+    def queryDouble(self, sql):
+        # sql
+        sql1 = sql.replace('@db_name', self.db1)
+        tdLog.info(sql1)
+        start1 = time.time()
+        rows1 = tdSql.query(sql1)
+        spend1 = time.time() - start1
+        res1 = copy.copy(tdSql.queryResult)
+
+        sql2 = sql.replace('@db_name', self.db2)
+        tdLog.info(sql2)
+        start2 = time.time()
+        tdSql.query(sql2)
+        spend2 = time.time() - start2
+        res2 = tdSql.queryResult
+
+        rowlen1 = len(res1)
+        rowlen2 = len(res2)
+
+        if rowlen1 != rowlen2:
+            tdLog.exit(f"both row count not equal. rowlen1={rowlen1} rowlen2={rowlen2} ")
+            return False
+        
+        for i in range(rowlen1):
+            row1 = res1[i]
+            row2 = res2[i]
+            collen1 = len(row1)
+            collen2 = len(row2)
+            if collen1 != collen2:
+                tdLog.exit(f"both col count not equal. collen1={collen1} collen2={collen2}")
+                return False
+            for j in range(collen1):
+                if row1[j] != row2[j]:
+                    tdLog.exit(f"both col not equal. row={i} col={j} col1={row1[j]} col2={row2[j]} .")
+                    return False
+
+        # warning performance
+        diff = (spend2 - spend1)*100/spend1
+        tdLog.info("spend1=%.6fs spend2=%.6fs diff=%.1f%%"%(spend1, spend2, diff))
+        if spend2 > spend1 and diff > 20:
+            tdLog.info("warning: the diff for performance after spliting is over 20%")
+
+        return True
+
+
+    # check result
+    def checkResult(self):
+        # check vgroupid
+        sql = f"select vgroup_id from information_schema.ins_vgroups where db_name='{self.db2}'"
+        tdSql.query(sql)
+        tdSql.checkRows(self.vgroups2)
+
+        # check child table count same
+        sql = "select table_name from information_schema.ins_tables where db_name='@db_name' order by table_name"
+        self.queryDouble(sql)
+
+        # check row value is ok
+        sql = "select * from @db_name.st order by ts"
+        self.queryDouble(sql)
+        
+        # where
+        sql = "select *,tbname from @db_name.st where c1 < 1000 order by ts"
+        self.queryDouble(sql)
+
+        # max
+        sql = "select max(c1) from @db_name.st"
+        self.queryDouble(sql)
+
+        # min
+        sql = "select min(c2) from @db_name.st"
+        self.queryDouble(sql)
+
+        # sum
+        sql = "select sum(c1) from @db_name.st"
+        self.queryDouble(sql)
+
+        # normal table
+
+        # count
+        sql = "select count(*) from @db_name.ta"
+        self.queryDouble(sql)
+
+        # all rows
+        sql = "select * from @db_name.ta"
+        self.queryDouble(sql)
+
+        # sum
+        sql = "select sum(c1) from @db_name.ta"
+        self.queryDouble(sql)
+
+
+    # get vgroup list
+    def getVGroup(self, db_name):
+        vgidList = []
+        sql = f"select vgroup_id from information_schema.ins_vgroups where db_name='{db_name}'"
+        res = tdSql.getResult(sql)
+        rows = len(res)
+        for i in range(rows):
+            vgidList.append(res[i][0])
+
+        return vgidList;        
+
+    # split vgroup on db2
+    def splitVGroup(self, db_name):
+        vgids = self.getVGroup(db_name)
+        selid = random.choice(vgids)
+        sql = f"split vgroup {selid}"
+        tdLog.info(sql)
+        tdSql.execute(sql)
+
+        # wait end
+        seconds = 300
+        for i in range(seconds):
+            sql ="show transactions;"
+            rows = tdSql.query(sql)
+            if rows == 0:
+                tdLog.info("split vgroup finished.")
+                return True
+            #tdLog.info(f"i={i} wait split vgroup ...")
+            time.sleep(1)
+
+        tdLog.exit(f"split vgroup transaction is not finished after executing {seconds}s")
+        return False
+
+    # split error 
+    def expectSplitError(self, dbName):
+        vgids = self.getVGroup(dbName)
+        selid = random.choice(vgids)
+        sql = f"split vgroup {selid}"
+        tdLog.info(sql)
+        tdSql.error(sql)    
+
+    # expect split ok
+    def expectSplitOk(self, dbName):
+        # split vgroup
+        vgList1 = self.getVGroup(dbName)
+        self.splitVGroup(dbName)
+        vgList2 = self.getVGroup(dbName)
+        vgNum1 = len(vgList1) + 1
+        vgNum2 = len(vgList2)
+        if vgNum1 != vgNum2:
+            tdLog.exit(f" vglist len={vgNum1} is not same for expect {vgNum2}")
+            return
+
+    # split empty database
+    def splitEmptyDB(self):        
+        dbName = "emptydb"
+        vgNum = 2
+        # create database
+        sql = f"create database {dbName} vgroups {vgNum} replica 1"
+        tdLog.info(sql)
+        tdSql.execute(sql)
+
+        # split vgroup
+        self.expectSplitOk(dbName)
+
+
+    # forbid
+    def checkForbid(self):
+        # stream
+        tdLog.info("check forbid split having stream...")
+        tdSql.execute("create database streamdb;")
+        tdSql.execute("use streamdb;")
+        tdSql.execute("create table ta(ts timestamp, age int);")
+        tdSql.execute("create stream ma into sta as select count(*) from ta interval(1s);")
+        self.expectSplitError("streamdb")
+        tdSql.execute("drop stream ma;")
+        self.expectSplitOk("streamdb")
+
+        # topic
+        tdLog.info("check forbid split having topic...")
+        tdSql.execute("create database topicdb wal_retention_period 10;")
+        tdSql.execute("use topicdb;")
+        tdSql.execute("create table ta(ts timestamp, age int);")
+        tdSql.execute("create topic toa as select * from ta;")
+        
+        #self.expectSplitError("topicdb")
+        tdSql.execute("drop topic toa;")
+        self.expectSplitOk("topicdb")
+   
+    # compact and check db2
+    def compactAndCheck(self):
+        tdLog.info("compact db2 and check result ...")
+        # compact
+        tdSql.execute(f"compact database {self.db2};")
+        # check result
+        self.checkResult()
+
+    # run
+    def run(self):
+        # prepare env
+        self.prepareEnv()
+
+        for i in range(3):
+            # split vgroup on db2
+            start = time.time()
+            self.splitVGroup(self.db2)
+            end = time.time()
+            self.vgroups2 += 1
+            
+            # check two db query result same
+            self.checkResult()
+            spend = "%.3f"%(end-start)
+            tdLog.info(f"split vgroup i={i} passed. spend = {spend}s")
+
+        # split empty db
+        self.splitEmptyDB()
+
+        # check topic and stream forib
+        self.checkForbid()
+
+        # compact database
+        self.compactAndCheck()
+
+    # stop
+    def stop(self):
+        tdSql.close()
+        tdLog.success(f"{__file__} successfully executed")
+
+
+tdCases.addLinux(__file__, TDTestCase())
+tdCases.addWindows(__file__, TDTestCase())

tests/system-test/0-others/splitVGroupRep3.py

@@ -137,10 +137,10 @@ class TDTestCase:
         tdSql.execute(sql1)
 
         sql2 = sql.replace("@db_name", self.db2)
-        if len(sql1) > 100:
+        if len(sql2) > 100:
-            tdLog.info(sql1[:100])
+            tdLog.info(sql2[:100])
         else:
-            tdLog.info(sql1)
+            tdLog.info(sql2)
         tdSql.execute(sql2)
         
 
@@ -151,8 +151,8 @@ class TDTestCase:
         self.childCnt = 10
         self.childRow = 10000
         self.batchSize = 5000
-        self.vgroups1  = 4
+        self.vgroups1  = 2
-        self.vgroups2  = 4
+        self.vgroups2  = 2
         self.db1 = "db1"
         self.db2 = "db2"
         
@@ -183,6 +183,16 @@ class TDTestCase:
         # insert data
         self.insertData()
 
+        # update
+        self.ts = 1680000000000 + 10000
+        self.childRow = 2000
+
+        # delete data
+        sql = "delete from @db_name.st where ts > 1680000001900 and ts < 1680000012000"
+        self.exeDouble(sql)
+        sql = "delete from @db_name.st where ts > 1680000029000 and ts < 1680000048000"
+        self.exeDouble(sql)
+
     # check data correct
     def checkExpect(self, sql, expectVal):
         tdSql.query(sql)
@@ -225,7 +235,7 @@ class TDTestCase:
         rowlen2 = len(res2)
 
         if rowlen1 != rowlen2:
-            tdLog.exit(f"rowlen1={rowlen1} rowlen2={rowlen2} both not equal.")
+            tdLog.exit(f"both row count not equal. rowlen1={rowlen1} rowlen2={rowlen2} ")
             return False
         
         for i in range(rowlen1):
@@ -234,11 +244,11 @@ class TDTestCase:
             collen1 = len(row1)
             collen2 = len(row2)
             if collen1 != collen2:
-                tdLog.exit(f"collen1={collen1} collen2={collen2} both not equal.")
+                tdLog.exit(f"both col count not equal. collen1={collen1} collen2={collen2}")
                 return False
             for j in range(collen1):
                 if row1[j] != row2[j]:
-                    tdLog.exit(f"col={j} col1={row1[j]} col2={row2[j]} both col not equal.")
+                    tdLog.exit(f"both col not equal. row={i} col={j} col1={row1[j]} col2={row2[j]} .")
                     return False
 
         # warning performance
@@ -354,7 +364,7 @@ class TDTestCase:
         dbName = "emptydb"
         vgNum = 2
         # create database
-        sql = f"create database {dbName} vgroups {vgNum}"
+        sql = f"create database {dbName} vgroups {vgNum} replica 3"
         tdLog.info(sql)
         tdSql.execute(sql)
 
@@ -380,7 +390,7 @@ class TDTestCase:
         tdSql.execute("use topicdb;")
         tdSql.execute("create table ta(ts timestamp, age int);")
         tdSql.execute("create topic toa as select * from ta;")
-        self.expectSplitError("topicdb")
+        #self.expectSplitError("topicdb")
         tdSql.execute("drop topic toa;")
         self.expectSplitOk("topicdb")
    
@@ -397,7 +407,7 @@ class TDTestCase:
         # prepare env
         self.prepareEnv()
 
-        for i in range(5):
+        for i in range(2):
             # split vgroup on db2
             start = time.time()
             self.splitVGroup(self.db2)