diff --git a/examples/c/tmq.c b/examples/c/tmq.c index 2e8aa21da7..7870d7d9a1 100644 --- a/examples/c/tmq.c +++ b/examples/c/tmq.c @@ -47,7 +47,7 @@ int32_t init_env() { return -1; } - TAOS_RES* pRes = taos_query(pConn, "create database if not exists abc1 vgroups 2"); + TAOS_RES* pRes = taos_query(pConn, "create database if not exists abc1 vgroups 1"); if (taos_errno(pRes) != 0) { printf("error in create db, reason:%s\n", taos_errstr(pRes)); return -1; @@ -146,8 +146,8 @@ int32_t create_topic() { return 0; } -void tmq_commit_cb_print(tmq_t* tmq, tmq_resp_err_t resp, tmq_topic_vgroup_list_t* offsets, void* param) { - printf("commit %d tmq %p offsets %p param %p\n", resp, tmq, offsets, param); +void tmq_commit_cb_print(tmq_t* tmq, int32_t code, void* param) { + printf("commit %d tmq %p param %p\n", code, tmq, param); } tmq_t* build_consumer() { @@ -167,7 +167,7 @@ tmq_t* build_consumer() { tmq_conf_set(conf, "td.connect.user", "root"); tmq_conf_set(conf, "td.connect.pass", "taosdata"); tmq_conf_set(conf, "msg.with.table.name", "true"); - tmq_conf_set(conf, "enable.auto.commit", "false"); + tmq_conf_set(conf, "enable.auto.commit", "true"); tmq_conf_set_auto_commit_cb(conf, tmq_commit_cb_print, NULL); tmq_t* tmq = tmq_consumer_new(conf, NULL, 0); assert(tmq); @@ -183,10 +183,10 @@ tmq_list_t* build_topic_list() { } void basic_consume_loop(tmq_t* tmq, tmq_list_t* topics) { - tmq_resp_err_t err; + int32_t code; - if ((err = tmq_subscribe(tmq, topics))) { - fprintf(stderr, "%% Failed to start consuming topics: %s\n", tmq_err2str(err)); + if ((code = tmq_subscribe(tmq, topics))) { + fprintf(stderr, "%% Failed to start consuming topics: %s\n", tmq_err2str(code)); printf("subscribe err\n"); return; } @@ -201,12 +201,13 @@ void basic_consume_loop(tmq_t* tmq, tmq_list_t* topics) { taos_free_result(tmqmessage); /*} else {*/ /*break;*/ + /*tmq_commit_sync(tmq, NULL);*/ } } - err = tmq_consumer_close(tmq); - if (err) - fprintf(stderr, "%% Failed to close consumer: %s\n", tmq_err2str(err)); + code = tmq_consumer_close(tmq); + if (code) + fprintf(stderr, "%% Failed to close consumer: %s\n", tmq_err2str(code)); else fprintf(stderr, "%% Consumer closed\n"); } @@ -214,11 +215,11 @@ void basic_consume_loop(tmq_t* tmq, tmq_list_t* topics) { void sync_consume_loop(tmq_t* tmq, tmq_list_t* topics) { static const int MIN_COMMIT_COUNT = 1; - int msg_count = 0; - tmq_resp_err_t err; + int msg_count = 0; + int32_t code; - if ((err = tmq_subscribe(tmq, topics))) { - fprintf(stderr, "%% Failed to start consuming topics: %s\n", tmq_err2str(err)); + if ((code = tmq_subscribe(tmq, topics))) { + fprintf(stderr, "%% Failed to start consuming topics: %s\n", tmq_err2str(code)); return; } @@ -239,14 +240,14 @@ void sync_consume_loop(tmq_t* tmq, tmq_list_t* topics) { msg_process(tmqmessage); taos_free_result(tmqmessage); - /*tmq_commit_async(tmq, NULL, tmq_commit_cb_print, NULL);*/ + /*tmq_commit_sync(tmq, NULL);*/ /*if ((++msg_count % MIN_COMMIT_COUNT) == 0) tmq_commit(tmq, NULL, 0);*/ } } - err = tmq_consumer_close(tmq); - if (err) - fprintf(stderr, "%% Failed to close consumer: %s\n", tmq_err2str(err)); + code = tmq_consumer_close(tmq); + if (code) + fprintf(stderr, "%% Failed to close consumer: %s\n", tmq_err2str(code)); else fprintf(stderr, "%% Consumer closed\n"); } diff --git a/include/client/taos.h b/include/client/taos.h index deb4276b54..e9f7f88387 100644 --- a/include/client/taos.h +++ b/include/client/taos.h @@ -126,47 +126,47 @@ typedef struct setConfRet { char retMsg[RET_MSG_LENGTH]; } setConfRet; -DLL_EXPORT void taos_cleanup(void); -DLL_EXPORT int taos_options(TSDB_OPTION option, const void *arg, ...); -DLL_EXPORT setConfRet taos_set_config(const char *config); -DLL_EXPORT int taos_init(void); -DLL_EXPORT TAOS *taos_connect(const char *ip, const char *user, const char *pass, const char *db, uint16_t port); -DLL_EXPORT TAOS *taos_connect_auth(const char *ip, const char *user, const char *auth, const char *db, uint16_t port); -DLL_EXPORT void taos_close(TAOS *taos); +DLL_EXPORT void taos_cleanup(void); +DLL_EXPORT int taos_options(TSDB_OPTION option, const void *arg, ...); +DLL_EXPORT setConfRet taos_set_config(const char *config); +DLL_EXPORT int taos_init(void); +DLL_EXPORT TAOS *taos_connect(const char *ip, const char *user, const char *pass, const char *db, uint16_t port); +DLL_EXPORT TAOS *taos_connect_auth(const char *ip, const char *user, const char *auth, const char *db, uint16_t port); +DLL_EXPORT void taos_close(TAOS *taos); -const char *taos_data_type(int type); +const char *taos_data_type(int type); -DLL_EXPORT TAOS_STMT *taos_stmt_init(TAOS *taos); -DLL_EXPORT int taos_stmt_prepare(TAOS_STMT *stmt, const char *sql, unsigned long length); -DLL_EXPORT int taos_stmt_set_tbname_tags(TAOS_STMT *stmt, const char *name, TAOS_MULTI_BIND *tags); -DLL_EXPORT int taos_stmt_set_tbname(TAOS_STMT *stmt, const char *name); -DLL_EXPORT int taos_stmt_set_tags(TAOS_STMT *stmt, TAOS_MULTI_BIND *tags); -DLL_EXPORT int taos_stmt_set_sub_tbname(TAOS_STMT *stmt, const char *name); -DLL_EXPORT int taos_stmt_get_tag_fields(TAOS_STMT *stmt, int *fieldNum, TAOS_FIELD_E **fields); -DLL_EXPORT int taos_stmt_get_col_fields(TAOS_STMT *stmt, int *fieldNum, TAOS_FIELD_E **fields); +DLL_EXPORT TAOS_STMT *taos_stmt_init(TAOS *taos); +DLL_EXPORT int taos_stmt_prepare(TAOS_STMT *stmt, const char *sql, unsigned long length); +DLL_EXPORT int taos_stmt_set_tbname_tags(TAOS_STMT *stmt, const char *name, TAOS_MULTI_BIND *tags); +DLL_EXPORT int taos_stmt_set_tbname(TAOS_STMT *stmt, const char *name); +DLL_EXPORT int taos_stmt_set_tags(TAOS_STMT *stmt, TAOS_MULTI_BIND *tags); +DLL_EXPORT int taos_stmt_set_sub_tbname(TAOS_STMT *stmt, const char *name); +DLL_EXPORT int taos_stmt_get_tag_fields(TAOS_STMT *stmt, int *fieldNum, TAOS_FIELD_E **fields); +DLL_EXPORT int taos_stmt_get_col_fields(TAOS_STMT *stmt, int *fieldNum, TAOS_FIELD_E **fields); -DLL_EXPORT int taos_stmt_is_insert(TAOS_STMT *stmt, int *insert); -DLL_EXPORT int taos_stmt_num_params(TAOS_STMT *stmt, int *nums); -DLL_EXPORT int taos_stmt_get_param(TAOS_STMT *stmt, int idx, int *type, int *bytes); -DLL_EXPORT int taos_stmt_bind_param(TAOS_STMT *stmt, TAOS_MULTI_BIND *bind); -DLL_EXPORT int taos_stmt_bind_param_batch(TAOS_STMT *stmt, TAOS_MULTI_BIND *bind); -DLL_EXPORT int taos_stmt_bind_single_param_batch(TAOS_STMT *stmt, TAOS_MULTI_BIND *bind, int colIdx); -DLL_EXPORT int taos_stmt_add_batch(TAOS_STMT *stmt); -DLL_EXPORT int taos_stmt_execute(TAOS_STMT *stmt); -DLL_EXPORT TAOS_RES *taos_stmt_use_result(TAOS_STMT *stmt); -DLL_EXPORT int taos_stmt_close(TAOS_STMT *stmt); -DLL_EXPORT char *taos_stmt_errstr(TAOS_STMT *stmt); -DLL_EXPORT int taos_stmt_affected_rows(TAOS_STMT *stmt); -DLL_EXPORT int taos_stmt_affected_rows_once(TAOS_STMT *stmt); +DLL_EXPORT int taos_stmt_is_insert(TAOS_STMT *stmt, int *insert); +DLL_EXPORT int taos_stmt_num_params(TAOS_STMT *stmt, int *nums); +DLL_EXPORT int taos_stmt_get_param(TAOS_STMT *stmt, int idx, int *type, int *bytes); +DLL_EXPORT int taos_stmt_bind_param(TAOS_STMT *stmt, TAOS_MULTI_BIND *bind); +DLL_EXPORT int taos_stmt_bind_param_batch(TAOS_STMT *stmt, TAOS_MULTI_BIND *bind); +DLL_EXPORT int taos_stmt_bind_single_param_batch(TAOS_STMT *stmt, TAOS_MULTI_BIND *bind, int colIdx); +DLL_EXPORT int taos_stmt_add_batch(TAOS_STMT *stmt); +DLL_EXPORT int taos_stmt_execute(TAOS_STMT *stmt); +DLL_EXPORT TAOS_RES *taos_stmt_use_result(TAOS_STMT *stmt); +DLL_EXPORT int taos_stmt_close(TAOS_STMT *stmt); +DLL_EXPORT char *taos_stmt_errstr(TAOS_STMT *stmt); +DLL_EXPORT int taos_stmt_affected_rows(TAOS_STMT *stmt); +DLL_EXPORT int taos_stmt_affected_rows_once(TAOS_STMT *stmt); -DLL_EXPORT TAOS_RES *taos_query(TAOS *taos, const char *sql); +DLL_EXPORT TAOS_RES *taos_query(TAOS *taos, const char *sql); -DLL_EXPORT TAOS_ROW taos_fetch_row(TAOS_RES *res); -DLL_EXPORT int taos_result_precision(TAOS_RES *res); // get the time precision of result -DLL_EXPORT void taos_free_result(TAOS_RES *res); -DLL_EXPORT int taos_field_count(TAOS_RES *res); -DLL_EXPORT int taos_num_fields(TAOS_RES *res); -DLL_EXPORT int taos_affected_rows(TAOS_RES *res); +DLL_EXPORT TAOS_ROW taos_fetch_row(TAOS_RES *res); +DLL_EXPORT int taos_result_precision(TAOS_RES *res); // get the time precision of result +DLL_EXPORT void taos_free_result(TAOS_RES *res); +DLL_EXPORT int taos_field_count(TAOS_RES *res); +DLL_EXPORT int taos_num_fields(TAOS_RES *res); +DLL_EXPORT int taos_affected_rows(TAOS_RES *res); DLL_EXPORT TAOS_FIELD *taos_fetch_fields(TAOS_RES *res); DLL_EXPORT int taos_select_db(TAOS *taos, const char *db); @@ -181,8 +181,8 @@ DLL_EXPORT int *taos_get_column_data_offset(TAOS_RES *res, int columnInde DLL_EXPORT int taos_validate_sql(TAOS *taos, const char *sql); DLL_EXPORT void taos_reset_current_db(TAOS *taos); -DLL_EXPORT int *taos_fetch_lengths(TAOS_RES *res); -DLL_EXPORT TAOS_ROW *taos_result_block(TAOS_RES *res); +DLL_EXPORT int *taos_fetch_lengths(TAOS_RES *res); +DLL_EXPORT TAOS_ROW *taos_result_block(TAOS_RES *res); DLL_EXPORT const char *taos_get_server_info(TAOS *taos); DLL_EXPORT const char *taos_get_client_info(); @@ -192,8 +192,8 @@ DLL_EXPORT int taos_errno(TAOS_RES *tres); DLL_EXPORT void taos_query_a(TAOS *taos, const char *sql, __taos_async_fn_t fp, void *param); DLL_EXPORT void taos_fetch_rows_a(TAOS_RES *res, __taos_async_fn_t fp, void *param); -DLL_EXPORT void taos_fetch_raw_block_a(TAOS_RES* res, __taos_async_fn_t fp, void* param); -DLL_EXPORT const void *taos_get_raw_block(TAOS_RES* res); +DLL_EXPORT void taos_fetch_raw_block_a(TAOS_RES *res, __taos_async_fn_t fp, void *param); +DLL_EXPORT const void *taos_get_raw_block(TAOS_RES *res); // Shuduo: temporary enable for app build #if 1 @@ -209,21 +209,20 @@ DLL_EXPORT TAOS_RES *taos_schemaless_insert(TAOS *taos, char *lines[], int numLi /* --------------------------TMQ INTERFACE------------------------------- */ +#if 0 enum { TMQ_RESP_ERR__FAIL = -1, TMQ_RESP_ERR__SUCCESS = 0, }; typedef int32_t tmq_resp_err_t; +#endif -typedef struct tmq_t tmq_t; -typedef struct tmq_topic_vgroup_t tmq_topic_vgroup_t; -typedef struct tmq_topic_vgroup_list_t tmq_topic_vgroup_list_t; - +typedef struct tmq_t tmq_t; typedef struct tmq_conf_t tmq_conf_t; typedef struct tmq_list_t tmq_list_t; -typedef void(tmq_commit_cb(tmq_t *, tmq_resp_err_t, tmq_topic_vgroup_list_t *, void *param)); +typedef void(tmq_commit_cb(tmq_t *, int32_t code, void *param)); DLL_EXPORT tmq_list_t *tmq_list_new(); DLL_EXPORT int32_t tmq_list_append(tmq_list_t *, const char *); @@ -233,25 +232,19 @@ DLL_EXPORT char **tmq_list_to_c_array(const tmq_list_t *); DLL_EXPORT tmq_t *tmq_consumer_new(tmq_conf_t *conf, char *errstr, int32_t errstrLen); -DLL_EXPORT const char *tmq_err2str(tmq_resp_err_t); +DLL_EXPORT const char *tmq_err2str(int32_t code); /* ------------------------TMQ CONSUMER INTERFACE------------------------ */ -DLL_EXPORT tmq_resp_err_t tmq_subscribe(tmq_t *tmq, const tmq_list_t *topic_list); -DLL_EXPORT tmq_resp_err_t tmq_unsubscribe(tmq_t *tmq); -DLL_EXPORT tmq_resp_err_t tmq_subscription(tmq_t *tmq, tmq_list_t **topics); - +DLL_EXPORT int32_t tmq_subscribe(tmq_t *tmq, const tmq_list_t *topic_list); +DLL_EXPORT int32_t tmq_unsubscribe(tmq_t *tmq); +DLL_EXPORT int32_t tmq_subscription(tmq_t *tmq, tmq_list_t **topics); // timeout: -1 means infinitely waiting DLL_EXPORT TAOS_RES *tmq_consumer_poll(tmq_t *tmq, int64_t timeout); +DLL_EXPORT int32_t tmq_consumer_close(tmq_t *tmq); -DLL_EXPORT tmq_resp_err_t tmq_consumer_close(tmq_t *tmq); -DLL_EXPORT tmq_resp_err_t tmq_commit_sync(tmq_t *tmq, const tmq_topic_vgroup_list_t *offsets); -DLL_EXPORT void tmq_commit_async(tmq_t *tmq, const tmq_topic_vgroup_list_t *offsets, tmq_commit_cb *cb, void *param); - -#if 0 -DLL_EXPORT tmq_resp_err_t tmq_commit(tmq_t *tmq, const tmq_topic_vgroup_list_t *offsets, int32_t async); -DLL_EXPORT tmq_resp_err_t tmq_seek(tmq_t *tmq, const tmq_topic_vgroup_t *offset); -#endif +DLL_EXPORT int32_t tmq_commit_sync(tmq_t *tmq, const TAOS_RES *msg); +DLL_EXPORT void tmq_commit_async(tmq_t *tmq, const TAOS_RES *msg, tmq_commit_cb *cb, void *param); /* ----------------------TMQ CONFIGURATION INTERFACE---------------------- */ @@ -275,11 +268,6 @@ DLL_EXPORT const char *tmq_get_db_name(TAOS_RES *res); DLL_EXPORT int32_t tmq_get_vgroup_id(TAOS_RES *res); DLL_EXPORT const char *tmq_get_table_name(TAOS_RES *res); -#if 0 -DLL_EXPORT int64_t tmq_get_request_offset(tmq_message_t *message); -DLL_EXPORT int64_t tmq_get_response_offset(tmq_message_t *message); -#endif - /* ------------------------------ TMQ END -------------------------------- */ #if 1 // Shuduo: temporary enable for app build diff --git a/include/common/tcommon.h b/include/common/tcommon.h index 2e646f4769..a05287761e 100644 --- a/include/common/tcommon.h +++ b/include/common/tcommon.h @@ -35,6 +35,7 @@ enum { TMQ_MSG_TYPE__DUMMY = 0, TMQ_MSG_TYPE__POLL_RSP, TMQ_MSG_TYPE__EP_RSP, + TMQ_MSG_TYPE__END_RSP, }; typedef enum EStreamType { diff --git a/include/common/tglobal.h b/include/common/tglobal.h index 1b44b6d7ea..8c03d3ff42 100644 --- a/include/common/tglobal.h +++ b/include/common/tglobal.h @@ -61,6 +61,7 @@ extern int32_t tsNumOfRpcThreads; extern int32_t tsNumOfCommitThreads; extern int32_t tsNumOfTaskQueueThreads; extern int32_t tsNumOfMnodeQueryThreads; +extern int32_t tsNumOfMnodeFetchThreads; extern int32_t tsNumOfMnodeReadThreads; extern int32_t tsNumOfVnodeQueryThreads; extern int32_t tsNumOfVnodeFetchThreads; diff --git a/include/common/tmsg.h b/include/common/tmsg.h index 2f2b5c5bdd..857a554f62 100644 --- a/include/common/tmsg.h +++ b/include/common/tmsg.h @@ -106,6 +106,7 @@ typedef enum _mgmt_table { TSDB_MGMT_TABLE_CONNS, TSDB_MGMT_TABLE_QUERIES, TSDB_MGMT_TABLE_VNODES, + TSDB_MGMT_TABLE_APPS, TSDB_MGMT_TABLE_MAX, } EShowType; @@ -1316,8 +1317,6 @@ int32_t tSerializeSSetStandbyReq(void* buf, int32_t bufLen, SSetStandbyReq* pReq int32_t tDeserializeSSetStandbyReq(void* buf, int32_t bufLen, SSetStandbyReq* pReq); typedef struct { - int32_t connId; // todo remove - int32_t queryId; // todo remove char queryStrId[TSDB_QUERY_ID_LEN]; } SKillQueryReq; @@ -1325,7 +1324,7 @@ int32_t tSerializeSKillQueryReq(void* buf, int32_t bufLen, SKillQueryReq* pReq); int32_t tDeserializeSKillQueryReq(void* buf, int32_t bufLen, SKillQueryReq* pReq); typedef struct { - int32_t connId; + uint32_t connId; } SKillConnReq; int32_t tSerializeSKillConnReq(void* buf, int32_t bufLen, SKillConnReq* pReq); @@ -2012,7 +2011,6 @@ typedef struct { int64_t useconds; int64_t stime; // timestamp precision ms int64_t reqRid; - int32_t pid; bool stableQuery; char fqdn[TSDB_FQDN_LEN]; int32_t subPlanNum; @@ -2021,8 +2019,6 @@ typedef struct { typedef struct { uint32_t connId; - int32_t pid; - char app[TSDB_APP_NAME_LEN]; SArray* queryDesc; // SArray } SQueryHbReqBasic; @@ -2037,9 +2033,31 @@ typedef struct { SArray* pQnodeList; } SQueryHbRspBasic; +typedef struct SAppClusterSummary { + uint64_t numOfInsertsReq; + uint64_t numOfInsertRows; + uint64_t insertElapsedTime; + uint64_t insertBytes; // submit to tsdb since launched. + + uint64_t fetchBytes; + uint64_t queryElapsedTime; + uint64_t numOfSlowQueries; + uint64_t totalRequests; + uint64_t currentRequests; // the number of SRequestObj +} SAppClusterSummary; + +typedef struct { + int64_t appId; + int32_t pid; + char name[TSDB_APP_NAME_LEN]; + int64_t startTime; + SAppClusterSummary summary; +} SAppHbReq; + typedef struct { SClientHbKey connKey; int64_t clusterId; + SAppHbReq app; SQueryHbReqBasic* query; SHashObj* info; // hash } SClientHbReq; diff --git a/include/libs/catalog/catalog.h b/include/libs/catalog/catalog.h index f3600a509e..c0116051af 100644 --- a/include/libs/catalog/catalog.h +++ b/include/libs/catalog/catalog.h @@ -292,6 +292,8 @@ int32_t catalogUpdateVgEpSet(SCatalog* pCtg, const char* dbFName, int32_t vgId, int32_t ctgdLaunchAsyncCall(SCatalog* pCtg, SRequestConnInfo* pConn, uint64_t reqId, bool forceUpdate); +int32_t catalogClearCache(void); + /** * Destroy catalog and relase all resources */ diff --git a/include/libs/nodes/nodes.h b/include/libs/nodes/nodes.h index 6f239cc892..08190d978e 100644 --- a/include/libs/nodes/nodes.h +++ b/include/libs/nodes/nodes.h @@ -228,6 +228,7 @@ typedef enum ENodeType { QUERY_NODE_PHYSICAL_PLAN_FILL, QUERY_NODE_PHYSICAL_PLAN_MERGE_SESSION, QUERY_NODE_PHYSICAL_PLAN_STREAM_SESSION, + QUERY_NODE_PHYSICAL_PLAN_STREAM_SEMI_SESSION, QUERY_NODE_PHYSICAL_PLAN_STREAM_FINAL_SESSION, QUERY_NODE_PHYSICAL_PLAN_MERGE_STATE, QUERY_NODE_PHYSICAL_PLAN_STREAM_STATE, diff --git a/include/libs/nodes/plannodes.h b/include/libs/nodes/plannodes.h index 749d58b224..a9002b5d19 100644 --- a/include/libs/nodes/plannodes.h +++ b/include/libs/nodes/plannodes.h @@ -130,13 +130,17 @@ typedef struct SMergeLogicNode { typedef enum EWindowType { WINDOW_TYPE_INTERVAL = 1, WINDOW_TYPE_SESSION, WINDOW_TYPE_STATE } EWindowType; -typedef enum EIntervalAlgorithm { +typedef enum EWindowAlgorithm { INTERVAL_ALGO_HASH = 1, INTERVAL_ALGO_MERGE, INTERVAL_ALGO_STREAM_FINAL, INTERVAL_ALGO_STREAM_SEMI, INTERVAL_ALGO_STREAM_SINGLE, -} EIntervalAlgorithm; + SESSION_ALGO_STREAM_SEMI, + SESSION_ALGO_STREAM_FINAL, + SESSION_ALGO_STREAM_SINGLE, + SESSION_ALGO_MERGE, +} EWindowAlgorithm; typedef struct SWindowLogicNode { SLogicNode node; @@ -153,7 +157,7 @@ typedef struct SWindowLogicNode { int8_t triggerType; int64_t watermark; double filesFactor; - EIntervalAlgorithm intervalAlgo; + EWindowAlgorithm windowAlgo; } SWindowLogicNode; typedef struct SFillLogicNode { @@ -371,6 +375,8 @@ typedef struct SSessionWinodwPhysiNode { } SSessionWinodwPhysiNode; typedef SSessionWinodwPhysiNode SStreamSessionWinodwPhysiNode; +typedef SSessionWinodwPhysiNode SStreamSemiSessionWinodwPhysiNode; +typedef SSessionWinodwPhysiNode SStreamFinalSessionWinodwPhysiNode; typedef struct SStateWinodwPhysiNode { SWinodwPhysiNode window; diff --git a/include/libs/qcom/query.h b/include/libs/qcom/query.h index 89eb21556a..b6a3a08f7e 100644 --- a/include/libs/qcom/query.h +++ b/include/libs/qcom/query.h @@ -33,8 +33,6 @@ typedef enum { JOB_TASK_STATUS_PARTIAL_SUCCEED, JOB_TASK_STATUS_SUCCEED, JOB_TASK_STATUS_FAILED, - JOB_TASK_STATUS_CANCELLING, - JOB_TASK_STATUS_CANCELLED, JOB_TASK_STATUS_DROPPING, } EJobTaskType; diff --git a/include/libs/scheduler/scheduler.h b/include/libs/scheduler/scheduler.h index 8087a6d33c..ecb21335b9 100644 --- a/include/libs/scheduler/scheduler.h +++ b/include/libs/scheduler/scheduler.h @@ -73,13 +73,14 @@ typedef void (*schedulerExecCallback)(SQueryResult* pResult, void* param, int32_ typedef void (*schedulerFetchCallback)(void* pResult, void* param, int32_t code); typedef struct SSchedulerReq { - SRequestConnInfo *pConn; - SArray *pNodeList; - SQueryPlan *pDag; - const char *sql; - int64_t startTs; + bool *reqKilled; + SRequestConnInfo *pConn; + SArray *pNodeList; + SQueryPlan *pDag; + const char *sql; + int64_t startTs; schedulerExecCallback fp; - void* cbParam; + void* cbParam; } SSchedulerReq; @@ -127,7 +128,7 @@ void schedulerStopQueryHb(void *pTrans); * Free the query job * @param pJob */ -void schedulerFreeJob(int64_t job); +void schedulerFreeJob(int64_t job, int32_t errCode); void schedulerDestroy(void); diff --git a/include/util/taoserror.h b/include/util/taoserror.h index 675784b565..64cfcccdc1 100644 --- a/include/util/taoserror.h +++ b/include/util/taoserror.h @@ -127,6 +127,7 @@ int32_t* taosGetErrno(); #define TSDB_CODE_TSC_STMT_API_ERROR TAOS_DEF_ERROR_CODE(0, 0X0225) #define TSDB_CODE_TSC_STMT_TBNAME_ERROR TAOS_DEF_ERROR_CODE(0, 0X0226) #define TSDB_CODE_TSC_STMT_CLAUSE_ERROR TAOS_DEF_ERROR_CODE(0, 0X0227) +#define TSDB_CODE_TSC_QUERY_KILLED TAOS_DEF_ERROR_CODE(0, 0X0228) // mnode-common #define TSDB_CODE_MND_APP_ERROR TAOS_DEF_ERROR_CODE(0, 0x0300) @@ -429,7 +430,8 @@ int32_t* taosGetErrno(); #define TSDB_CODE_TQ_META_KEY_NOT_IN_TXN TAOS_DEF_ERROR_CODE(0, 0x0A09) #define TSDB_CODE_TQ_META_KEY_DUP_IN_TXN TAOS_DEF_ERROR_CODE(0, 0x0A0A) #define TSDB_CODE_TQ_GROUP_NOT_SET TAOS_DEF_ERROR_CODE(0, 0x0A0B) -#define TSDB_CODE_TQ_TABLE_SCHEMA_NOT_FOUND TAOS_DEF_ERROR_CODE(0, 0x0A0B) +#define TSDB_CODE_TQ_TABLE_SCHEMA_NOT_FOUND TAOS_DEF_ERROR_CODE(0, 0x0A0C) +#define TSDB_CODE_TQ_NO_COMMITTED_OFFSET TAOS_DEF_ERROR_CODE(0, 0x0A0D) // wal #define TSDB_CODE_WAL_APP_ERROR TAOS_DEF_ERROR_CODE(0, 0x1000) @@ -570,6 +572,7 @@ int32_t* taosGetErrno(); #define TSDB_CODE_SCH_INTERNAL_ERROR TAOS_DEF_ERROR_CODE(0, 0x2502) #define TSDB_CODE_SCH_IGNORE_ERROR TAOS_DEF_ERROR_CODE(0, 0x2503) #define TSDB_CODE_SCH_TIMEOUT_ERROR TAOS_DEF_ERROR_CODE(0, 0x2504) +#define TSDB_CODE_SCH_JOB_IS_DROPPING TAOS_DEF_ERROR_CODE(0, 0x2505) #define TSDB_CODE_QW_MSG_ERROR TAOS_DEF_ERROR_CODE(0, 0x2550) //parser @@ -710,6 +713,8 @@ int32_t* taosGetErrno(); //index #define TSDB_CODE_INDEX_REBUILDING TAOS_DEF_ERROR_CODE(0, 0x3200) +//tmq +#define TSDB_CODE_TMQ_INVALID_MSG TAOS_DEF_ERROR_CODE(0, 0x4000) #ifdef __cplusplus } diff --git a/source/client/inc/clientInt.h b/source/client/inc/clientInt.h index 69a24e20cd..d61262c9dc 100644 --- a/source/client/inc/clientInt.h +++ b/source/client/inc/clientInt.h @@ -76,10 +76,12 @@ typedef int32_t (*FHbReqHandle)(SClientHbKey* connKey, void* param, SClientHbReq typedef struct { int8_t inited; + int64_t appId; // ctl int8_t threadStop; TdThread thread; TdThreadMutex lock; // used when app init and cleanup + SHashObj *appSummary; SArray* appHbMgrs; // SArray one for each cluster FHbReqHandle reqHandle[CONN_TYPE__MAX]; FHbRspHandle rspHandle[CONN_TYPE__MAX]; @@ -92,33 +94,20 @@ typedef struct SQueryExecMetric { int64_t rsp; // receive response from server, us } SQueryExecMetric; -typedef struct SInstanceSummary { - uint64_t numOfInsertsReq; - uint64_t numOfInsertRows; - uint64_t insertElapsedTime; - uint64_t insertBytes; // submit to tsdb since launched. - - uint64_t fetchBytes; - uint64_t queryElapsedTime; - uint64_t numOfSlowQueries; - uint64_t totalRequests; - uint64_t currentRequests; // the number of SRequestObj -} SInstanceSummary; - typedef struct SHeartBeatInfo { void* pTimer; // timer, used to send request msg to mnode } SHeartBeatInfo; struct SAppInstInfo { - int64_t numOfConns; - SCorEpSet mgmtEp; - TdThreadMutex qnodeMutex; - SArray* pQnodeList; - SInstanceSummary summary; - SList* pConnList; // STscObj linked list - uint64_t clusterId; - void* pTransporter; - SAppHbMgr* pAppHbMgr; + int64_t numOfConns; + SCorEpSet mgmtEp; + TdThreadMutex qnodeMutex; + SArray* pQnodeList; + SAppClusterSummary summary; + SList* pConnList; // STscObj linked list + uint64_t clusterId; + void* pTransporter; + SAppHbMgr* pAppHbMgr; }; typedef struct SAppInfo { @@ -215,6 +204,7 @@ typedef struct SRequestObj { SRequestSendRecvBody body; bool stableQuery; + bool killed; uint32_t prevCode; //previous error code: todo refactor, add update flag for catalog uint32_t retry; } SRequestObj; diff --git a/source/client/src/clientEnv.c b/source/client/src/clientEnv.c index bb60624145..24246e5c45 100644 --- a/source/client/src/clientEnv.c +++ b/source/client/src/clientEnv.c @@ -48,7 +48,7 @@ static void registerRequest(SRequestObj *pRequest) { int32_t num = atomic_add_fetch_32(&pTscObj->numOfReqs, 1); if (pTscObj->pAppInfo) { - SInstanceSummary *pSummary = &pTscObj->pAppInfo->summary; + SAppClusterSummary *pSummary = &pTscObj->pAppInfo->summary; int32_t total = atomic_add_fetch_64((int64_t *)&pSummary->totalRequests, 1); int32_t currentInst = atomic_add_fetch_64((int64_t *)&pSummary->currentRequests, 1); @@ -62,7 +62,7 @@ static void deregisterRequest(SRequestObj *pRequest) { assert(pRequest != NULL); STscObj *pTscObj = pRequest->pTscObj; - SInstanceSummary *pActivity = &pTscObj->pAppInfo->summary; + SAppClusterSummary *pActivity = &pTscObj->pAppInfo->summary; int32_t currentInst = atomic_sub_fetch_64((int64_t *)&pActivity->currentRequests, 1); int32_t num = atomic_sub_fetch_32(&pTscObj->numOfReqs, 1); @@ -229,7 +229,7 @@ static void doDestroyRequest(void *p) { taosHashRemove(pRequest->pTscObj->pRequests, &pRequest->self, sizeof(pRequest->self)); if (pRequest->body.queryJob != 0) { - schedulerFreeJob(pRequest->body.queryJob); + schedulerFreeJob(pRequest->body.queryJob, 0); } taosMemoryFreeClear(pRequest->msgBuf); diff --git a/source/client/src/clientHb.c b/source/client/src/clientHb.c index e0b8f322bf..c6d0d6a860 100644 --- a/source/client/src/clientHb.c +++ b/source/client/src/clientHb.c @@ -164,6 +164,7 @@ static int32_t hbQueryHbRspHandle(SAppHbMgr *pAppHbMgr, SClientHbRsp *pRsp) { pTscObj->connId = pRsp->query->connId; if (pRsp->query->killRid) { + tscDebug("request rid %" PRIx64 " need to be killed now", pRsp->query->killRid); SRequestObj *pRequest = acquireRequest(pRsp->query->killRid); if (NULL == pRequest) { tscDebug("request 0x%" PRIx64 " not exist to kill", pRsp->query->killRid); @@ -304,7 +305,7 @@ int32_t hbBuildQueryDesc(SQueryHbReqBasic *hbBasic, STscObj *pObj) { while (pIter != NULL) { int64_t *rid = pIter; SRequestObj *pRequest = acquireRequest(*rid); - if (NULL == pRequest) { + if (NULL == pRequest || pRequest->killed) { pIter = taosHashIterate(pObj->pRequests, pIter); continue; } @@ -314,7 +315,6 @@ int32_t hbBuildQueryDesc(SQueryHbReqBasic *hbBasic, STscObj *pObj) { desc.queryId = pRequest->requestId; desc.useconds = now - pRequest->metric.start; desc.reqRid = pRequest->self; - desc.pid = hbBasic->pid; desc.stableQuery = pRequest->stableQuery; taosGetFqdn(desc.fqdn); desc.subPlanNum = pRequest->body.pDag ? pRequest->body.pDag->numOfSubplans : 0; @@ -360,8 +360,6 @@ int32_t hbGetQueryBasicInfo(SClientHbKey *connKey, SClientHbReq *req) { } hbBasic->connId = pTscObj->connId; - hbBasic->pid = taosGetPId(); - taosGetAppName(hbBasic->app, NULL); int32_t numOfQueries = pTscObj->pRequests ? taosHashGetSize(pTscObj->pRequests) : 0; if (numOfQueries <= 0) { @@ -507,6 +505,21 @@ int32_t hbGetExpiredStbInfo(SClientHbKey *connKey, struct SCatalog *pCatalog, SC return TSDB_CODE_SUCCESS; } +int32_t hbGetAppInfo(int64_t clusterId, SClientHbReq *req) { + SAppHbReq* pApp = taosHashGet(clientHbMgr.appSummary, &clusterId, sizeof(clusterId)); + if (NULL != pApp) { + memcpy(&req->app, pApp, sizeof(*pApp)); + } else { + memset(&req->app.summary, 0, sizeof(req->app.summary)); + req->app.pid = taosGetPId(); + req->app.appId = clientHbMgr.appId; + taosGetAppName(req->app.name, NULL); + } + + return TSDB_CODE_SUCCESS; +} + + int32_t hbQueryHbReqHandle(SClientHbKey *connKey, void *param, SClientHbReq *req) { int64_t *clusterId = (int64_t *)param; struct SCatalog *pCatalog = NULL; @@ -517,6 +530,8 @@ int32_t hbQueryHbReqHandle(SClientHbKey *connKey, void *param, SClientHbReq *req return code; } + hbGetAppInfo(*clusterId, req); + hbGetQueryBasicInfo(connKey, req); code = hbGetExpiredUserInfo(connKey, pCatalog, req); @@ -589,6 +604,50 @@ void hbThreadFuncUnexpectedStopped(void) { atomic_store_8(&clientHbMgr.threadStop, 2); } +void hbMergeSummary(SAppClusterSummary* dst, SAppClusterSummary* src) { + dst->numOfInsertsReq += src->numOfInsertsReq; + dst->numOfInsertRows += src->numOfInsertRows; + dst->insertElapsedTime += src->insertElapsedTime; + dst->insertBytes += src->insertBytes; + dst->fetchBytes += src->fetchBytes; + dst->queryElapsedTime += src->queryElapsedTime; + dst->numOfSlowQueries += src->numOfSlowQueries; + dst->totalRequests += src->totalRequests; + dst->currentRequests += src->currentRequests; +} + +int32_t hbGatherAppInfo(void) { + SAppHbReq req = {0}; + int sz = taosArrayGetSize(clientHbMgr.appHbMgrs); + if (sz > 0) { + req.pid = taosGetPId(); + req.appId = clientHbMgr.appId; + taosGetAppName(req.name, NULL); + } + + taosHashClear(clientHbMgr.appSummary); + + for (int32_t i = 0; i < sz; ++i) { + SAppHbMgr *pAppHbMgr = taosArrayGetP(clientHbMgr.appHbMgrs, i); + uint64_t clusterId = pAppHbMgr->pAppInstInfo->clusterId; + SAppHbReq* pApp = taosHashGet(clientHbMgr.appSummary, &clusterId, sizeof(clusterId)); + if (NULL == pApp) { + memcpy(&req.summary, &pAppHbMgr->pAppInstInfo->summary, sizeof(req.summary)); + req.startTime = pAppHbMgr->startTime; + taosHashPut(clientHbMgr.appSummary, &clusterId, sizeof(clusterId), &req, sizeof(req)); + } else { + if (pAppHbMgr->startTime < pApp->startTime) { + pApp->startTime = pAppHbMgr->startTime; + } + + hbMergeSummary(&pApp->summary, &pAppHbMgr->pAppInstInfo->summary); + } + } + + return TSDB_CODE_SUCCESS; +} + + static void *hbThreadFunc(void *param) { setThreadName("hb"); #ifdef WINDOWS @@ -605,6 +664,10 @@ static void *hbThreadFunc(void *param) { taosThreadMutexLock(&clientHbMgr.lock); int sz = taosArrayGetSize(clientHbMgr.appHbMgrs); + if (sz > 0) { + hbGatherAppInfo(); + } + for (int i = 0; i < sz; i++) { SAppHbMgr *pAppHbMgr = taosArrayGetP(clientHbMgr.appHbMgrs, i); @@ -748,6 +811,10 @@ int hbMgrInit() { int8_t old = atomic_val_compare_exchange_8(&clientHbMgr.inited, 0, 1); if (old == 1) return 0; + clientHbMgr.appId = tGenIdPI64(); + tscDebug("app %" PRIx64 " initialized", clientHbMgr.appId); + + clientHbMgr.appSummary = taosHashInit(10, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BIGINT), false, HASH_NO_LOCK); clientHbMgr.appHbMgrs = taosArrayInit(0, sizeof(void *)); taosThreadMutexInit(&clientHbMgr.lock, NULL); diff --git a/source/client/src/clientImpl.c b/source/client/src/clientImpl.c index 10c2161478..d04cc90ee4 100644 --- a/source/client/src/clientImpl.c +++ b/source/client/src/clientImpl.c @@ -418,7 +418,7 @@ int32_t scheduleAsyncQuery(SRequestObj* pRequest, SQueryPlan* pDag, SArray* pNod while (true) { if (code != TSDB_CODE_SUCCESS) { if (pRequest->body.queryJob != 0) { - schedulerFreeJob(pRequest->body.queryJob); + schedulerFreeJob(pRequest->body.queryJob, 0); } pRequest->code = code; @@ -439,7 +439,7 @@ int32_t scheduleAsyncQuery(SRequestObj* pRequest, SQueryPlan* pDag, SArray* pNod pRequest->body.resInfo.numOfRows = res.numOfRows; if (pRequest->body.queryJob != 0) { - schedulerFreeJob(pRequest->body.queryJob); + schedulerFreeJob(pRequest->body.queryJob, 0); } } @@ -461,14 +461,15 @@ int32_t scheduleQuery(SRequestObj* pRequest, SQueryPlan* pDag, SArray* pNodeList .sql = pRequest->sqlstr, .startTs = pRequest->metric.start, .fp = NULL, - .cbParam = NULL}; + .cbParam = NULL, + .reqKilled = &pRequest->killed}; int32_t code = schedulerExecJob(&req, &pRequest->body.queryJob, &res); pRequest->body.resInfo.execRes = res.res; if (code != TSDB_CODE_SUCCESS) { if (pRequest->body.queryJob != 0) { - schedulerFreeJob(pRequest->body.queryJob); + schedulerFreeJob(pRequest->body.queryJob, 0); } pRequest->code = code; @@ -481,7 +482,7 @@ int32_t scheduleQuery(SRequestObj* pRequest, SQueryPlan* pDag, SArray* pNodeList pRequest->body.resInfo.numOfRows = res.numOfRows; if (pRequest->body.queryJob != 0) { - schedulerFreeJob(pRequest->body.queryJob); + schedulerFreeJob(pRequest->body.queryJob, 0); } } @@ -608,6 +609,9 @@ void schedulerExecCb(SQueryResult* pResult, void* param, int32_t code) { SRequestObj* pRequest = (SRequestObj*)param; pRequest->code = code; + tscDebug("0x%" PRIx64 " enter scheduler exec cb, code:%d - %s, reqId:0x%" PRIx64, + pRequest->self, code, tstrerror(code), pRequest->requestId); + STscObj* pTscObj = pRequest->pTscObj; if (code != TSDB_CODE_SUCCESS && NEED_CLIENT_HANDLE_ERROR(code)) { tscDebug("0x%" PRIx64 " client retry to handle the error, code:%d - %s, tryCount:%d, reqId:0x%" PRIx64, @@ -738,7 +742,8 @@ void launchAsyncQuery(SRequestObj* pRequest, SQuery* pQuery) { .sql = pRequest->sqlstr, .startTs = pRequest->metric.start, .fp = schedulerExecCb, - .cbParam = pRequest}; + .cbParam = pRequest, + .reqKilled = &pRequest->killed}; code = schedulerAsyncExecJob(&req, &pRequest->body.queryJob); } else { tscError("0x%" PRIx64 " failed to create query plan, code:%s 0x%" PRIx64, pRequest->self, tstrerror(code), diff --git a/source/client/src/clientMain.c b/source/client/src/clientMain.c index 13a5acb472..9a5d5bbeca 100644 --- a/source/client/src/clientMain.c +++ b/source/client/src/clientMain.c @@ -246,13 +246,14 @@ TAOS_ROW taos_fetch_row(TAOS_RES *res) { if (TD_RES_QUERY(res)) { SRequestObj *pRequest = (SRequestObj *)res; + if (pRequest->type == TSDB_SQL_RETRIEVE_EMPTY_RESULT || pRequest->type == TSDB_SQL_INSERT || + pRequest->code != TSDB_CODE_SUCCESS || taos_num_fields(res) == 0 || pRequest->killed) { + return NULL; + } + #if SYNC_ON_TOP_OF_ASYNC return doAsyncFetchRows(pRequest, true, true); #else - if (pRequest->type == TSDB_SQL_RETRIEVE_EMPTY_RESULT || pRequest->type == TSDB_SQL_INSERT || - pRequest->code != TSDB_CODE_SUCCESS || taos_num_fields(res) == 0) { - return NULL; - } return doFetchRows(pRequest, true, true); #endif @@ -482,14 +483,20 @@ void taos_stop_query(TAOS_RES *res) { } SRequestObj *pRequest = (SRequestObj *)res; + pRequest->killed = true; + int32_t numOfFields = taos_num_fields(pRequest); - // It is not a query, no need to stop. if (numOfFields == 0) { + tscDebug("request %" PRIx64 " no need to be killed since not query", pRequest->requestId); return; } - schedulerFreeJob(pRequest->body.queryJob); + if (pRequest->body.queryJob) { + schedulerFreeJob(pRequest->body.queryJob, TSDB_CODE_TSC_QUERY_KILLED); + } + + tscDebug("request %" PRIx64 " killed", pRequest->requestId); } bool taos_is_null(TAOS_RES *res, int32_t row, int32_t col) { @@ -830,6 +837,9 @@ static void fetchCallback(void *pResult, void *param, int32_t code) { SReqResultInfo *pResultInfo = &pRequest->body.resInfo; + tscDebug("0x%" PRIx64 " enter scheduler fetch cb, code:%d - %s, reqId:0x%" PRIx64, + pRequest->self, code, tstrerror(code), pRequest->requestId); + pResultInfo->pData = pResult; pResultInfo->numOfRows = 0; diff --git a/source/client/src/tmq.c b/source/client/src/tmq.c index dd900e6045..638b4f1ea5 100644 --- a/source/client/src/tmq.c +++ b/source/client/src/tmq.c @@ -48,14 +48,6 @@ struct tmq_list_t { SArray container; }; -struct tmq_topic_vgroup_t { - SMqOffset offset; -}; - -struct tmq_topic_vgroup_list_t { - SArray container; // SArray -}; - struct tmq_conf_t { char clientId[256]; char groupId[TSDB_CGROUP_LEN]; @@ -161,9 +153,9 @@ typedef struct { } SMqPollRspWrapper; typedef struct { - tmq_t* tmq; - tsem_t rspSem; - tmq_resp_err_t rspErr; + tmq_t* tmq; + tsem_t rspSem; + int32_t rspErr; } SMqSubscribeCbParam; typedef struct { @@ -189,7 +181,7 @@ typedef struct { int8_t freeOffsets; tmq_commit_cb* userCb; tsem_t rspSem; - tmq_resp_err_t rspErr; + int32_t rspErr; SArray* offsets; void* userParam; } SMqCommitCbParam; @@ -201,12 +193,12 @@ typedef struct { int8_t freeOffsets; int32_t waitingRspNum; int32_t totalRspNum; - tmq_resp_err_t rspErr; + int32_t rspErr; tmq_commit_cb* userCb; - SArray* successfulOffsets; - SArray* failedOffsets; - void* userParam; - tsem_t rspSem; + /*SArray* successfulOffsets;*/ + /*SArray* failedOffsets;*/ + void* userParam; + tsem_t rspSem; } SMqCommitCbParamSet; typedef struct { @@ -347,10 +339,9 @@ int32_t tmqCommitCb(void* param, const SDataBuf* pMsg, int32_t code) { pParam->rspErr = code; if (pParam->async) { if (pParam->automatic && pParam->tmq->commitCb) { - pParam->tmq->commitCb(pParam->tmq, pParam->rspErr, (tmq_topic_vgroup_list_t*)pParam->offsets, - pParam->tmq->commitCbUserParam); + pParam->tmq->commitCb(pParam->tmq, pParam->rspErr, pParam->tmq->commitCbUserParam); } else if (!pParam->automatic && pParam->userCb) { - pParam->userCb(pParam->tmq, pParam->rspErr, (tmq_topic_vgroup_list_t*)pParam->offsets, pParam->userParam); + pParam->userCb(pParam->tmq, pParam->rspErr, pParam->userParam); } if (pParam->freeOffsets) { @@ -368,11 +359,16 @@ int32_t tmqCommitCb2(void* param, const SDataBuf* pBuf, int32_t code) { SMqCommitCbParam2* pParam = (SMqCommitCbParam2*)param; SMqCommitCbParamSet* pParamSet = (SMqCommitCbParamSet*)pParam->params; // push into array +#if 0 if (code == 0) { taosArrayPush(pParamSet->failedOffsets, &pParam->pOffset); } else { taosArrayPush(pParamSet->successfulOffsets, &pParam->pOffset); } +#endif + + /*tscDebug("receive offset commit cb of %s on vg %d, offset is %ld", pParam->pOffset->subKey, pParam->->vgId, + * pOffset->version);*/ // count down waiting rsp int32_t waitingRspNum = atomic_sub_fetch_32(&pParamSet->waitingRspNum, 1); @@ -383,23 +379,149 @@ int32_t tmqCommitCb2(void* param, const SDataBuf* pBuf, int32_t code) { if (pParamSet->async) { // call async cb func if (pParamSet->automatic && pParamSet->tmq->commitCb) { - pParamSet->tmq->commitCb(pParamSet->tmq, pParamSet->rspErr, NULL, pParamSet->tmq->commitCbUserParam); + pParamSet->tmq->commitCb(pParamSet->tmq, pParamSet->rspErr, pParamSet->tmq->commitCbUserParam); } else if (!pParamSet->automatic && pParamSet->userCb) { // sem post - pParamSet->userCb(pParamSet->tmq, pParamSet->rspErr, NULL, pParamSet->userParam); + pParamSet->userCb(pParamSet->tmq, pParamSet->rspErr, pParamSet->userParam); } + } else { + tsem_post(&pParamSet->rspSem); } +#if 0 taosArrayDestroyP(pParamSet->successfulOffsets, taosMemoryFree); taosArrayDestroyP(pParamSet->failedOffsets, taosMemoryFree); +#endif } return 0; } -int32_t tmqCommitInner2(tmq_t* tmq, const tmq_topic_vgroup_list_t* offsets, int8_t automatic, int8_t async, - tmq_commit_cb* userCb, void* userParam) { +int32_t tmqCommitInner2(tmq_t* tmq, const TAOS_RES* msg, int8_t automatic, int8_t async, tmq_commit_cb* userCb, + void* userParam) { int32_t code = -1; + if (msg != NULL) { + SMqRspObj* pRspObj = (SMqRspObj*)msg; + if (!TD_RES_TMQ(pRspObj)) { + return TSDB_CODE_TMQ_INVALID_MSG; + } + + SMqCommitCbParamSet* pParamSet = taosMemoryCalloc(1, sizeof(SMqCommitCbParamSet)); + if (pParamSet == NULL) { + terrno = TSDB_CODE_OUT_OF_MEMORY; + return -1; + } + pParamSet->tmq = tmq; + pParamSet->automatic = automatic; + pParamSet->async = async; + pParamSet->freeOffsets = 1; + pParamSet->userCb = userCb; + pParamSet->userParam = userParam; + tsem_init(&pParamSet->rspSem, 0, 0); + + for (int32_t i = 0; i < taosArrayGetSize(tmq->clientTopics); i++) { + SMqClientTopic* pTopic = taosArrayGet(tmq->clientTopics, i); + if (strcmp(pTopic->topicName, pRspObj->topic) == 0) { + for (int32_t j = 0; j < taosArrayGetSize(pTopic->vgs); j++) { + SMqClientVg* pVg = taosArrayGet(pTopic->vgs, j); + if (pVg->vgId == pRspObj->vgId) { + if (pVg->currentOffset < 0 || pVg->committedOffset == pVg->currentOffset) { + tscDebug("consumer %ld skip commit for topic %s vg %d, current offset is %ld, committed offset is %ld", + tmq->consumerId, pTopic->topicName, pVg->vgId, pVg->currentOffset, pVg->committedOffset); + + return 0; + } + + STqOffset* pOffset = taosMemoryCalloc(1, sizeof(STqOffset)); + if (pOffset == NULL) { + terrno = TSDB_CODE_OUT_OF_MEMORY; + return -1; + } + pOffset->type = TMQ_OFFSET__LOG; + pOffset->version = pVg->currentOffset; + + int32_t groupLen = strlen(tmq->groupId); + memcpy(pOffset->subKey, tmq->groupId, groupLen); + pOffset->subKey[groupLen] = TMQ_SEPARATOR; + strcpy(pOffset->subKey + groupLen + 1, pTopic->topicName); + + int32_t len; + int32_t code; + tEncodeSize(tEncodeSTqOffset, pOffset, len, code); + if (code < 0) { + ASSERT(0); + } + void* buf = taosMemoryCalloc(1, sizeof(SMsgHead) + len); + ((SMsgHead*)buf)->vgId = htonl(pVg->vgId); + + void* abuf = POINTER_SHIFT(buf, sizeof(SMsgHead)); + + SEncoder encoder; + tEncoderInit(&encoder, abuf, len); + tEncodeSTqOffset(&encoder, pOffset); + + // build param + SMqCommitCbParam2* pParam = taosMemoryCalloc(1, sizeof(SMqCommitCbParam2)); + pParam->params = pParamSet; + pParam->pOffset = pOffset; + + // build send info + SMsgSendInfo* pMsgSendInfo = taosMemoryCalloc(1, sizeof(SMsgSendInfo)); + if (pMsgSendInfo == NULL) { + // TODO + continue; + } + pMsgSendInfo->msgInfo = (SDataBuf){ + .pData = buf, + .len = sizeof(SMsgHead) + len, + .handle = NULL, + }; + + tscDebug("consumer %ld commit offset of %s on vg %d, offset is %ld", tmq->consumerId, pOffset->subKey, + pVg->vgId, pOffset->version); + + // TODO: put into cb + pVg->committedOffset = pVg->currentOffset; + + pMsgSendInfo->requestId = generateRequestId(); + pMsgSendInfo->requestObjRefId = 0; + pMsgSendInfo->param = pParam; + pMsgSendInfo->fp = tmqCommitCb2; + pMsgSendInfo->msgType = TDMT_VND_MQ_COMMIT_OFFSET; + // send msg + + int64_t transporterId = 0; + asyncSendMsgToServer(tmq->pTscObj->pAppInfo->pTransporter, &pVg->epSet, &transporterId, pMsgSendInfo); + pParamSet->waitingRspNum++; + pParamSet->totalRspNum++; + } + } + } + } + if (pParamSet->totalRspNum == 0) { + tsem_destroy(&pParamSet->rspSem); + taosMemoryFree(pParamSet); + return 0; + } + + if (!async) { + tsem_wait(&pParamSet->rspSem); + code = pParamSet->rspErr; + tsem_destroy(&pParamSet->rspSem); + } else { + code = 0; + } + + if (code != 0 && async) { + if (automatic) { + tmq->commitCb(tmq, code, tmq->commitCbUserParam); + } else { + userCb(tmq, code, userParam); + } + } + return 0; + } + SMqCommitCbParamSet* pParamSet = taosMemoryCalloc(1, sizeof(SMqCommitCbParamSet)); if (pParamSet == NULL) { terrno = TSDB_CODE_OUT_OF_MEMORY; @@ -415,17 +537,35 @@ int32_t tmqCommitInner2(tmq_t* tmq, const tmq_topic_vgroup_list_t* offsets, int8 for (int32_t i = 0; i < taosArrayGetSize(tmq->clientTopics); i++) { SMqClientTopic* pTopic = taosArrayGet(tmq->clientTopics, i); + + tscDebug("consumer %ld begin commit for topic %s, vgNum %d", tmq->consumerId, pTopic->topicName, + (int32_t)taosArrayGetSize(pTopic->vgs)); + for (int32_t j = 0; j < taosArrayGetSize(pTopic->vgs); j++) { - SMqClientVg* pVg = taosArrayGet(pTopic->vgs, i); - STqOffset* pOffset = taosMemoryCalloc(1, sizeof(STqOffset)); + SMqClientVg* pVg = taosArrayGet(pTopic->vgs, j); + + tscDebug("consumer %ld begin commit for topic %s, vgId %d", tmq->consumerId, pTopic->topicName, pVg->vgId); + + /*if (pVg->currentOffset < 0) {*/ + if (pVg->currentOffset < 0 || pVg->committedOffset == pVg->currentOffset) { + tscDebug("consumer %ld skip commit for topic %s vg %d, current offset is %ld, committed offset is %ld", + tmq->consumerId, pTopic->topicName, pVg->vgId, pVg->currentOffset, pVg->committedOffset); + + continue; + } + STqOffset* pOffset = taosMemoryCalloc(1, sizeof(STqOffset)); if (pOffset == NULL) { terrno = TSDB_CODE_OUT_OF_MEMORY; return -1; } - int32_t tlen = strlen(tmq->groupId); - memcpy(pOffset->subKey, tmq->groupId, tlen); - pOffset->subKey[tlen] = TMQ_SEPARATOR; - strcpy(pOffset->subKey + tlen + 1, pTopic->topicName); + pOffset->type = TMQ_OFFSET__LOG; + pOffset->version = pVg->currentOffset; + + int32_t groupLen = strlen(tmq->groupId); + memcpy(pOffset->subKey, tmq->groupId, groupLen); + pOffset->subKey[groupLen] = TMQ_SEPARATOR; + strcpy(pOffset->subKey + groupLen + 1, pTopic->topicName); + int32_t len; int32_t code; tEncodeSize(tEncodeSTqOffset, pOffset, len, code); @@ -454,25 +594,36 @@ int32_t tmqCommitInner2(tmq_t* tmq, const tmq_topic_vgroup_list_t* offsets, int8 } pMsgSendInfo->msgInfo = (SDataBuf){ .pData = buf, - .len = len, + .len = sizeof(SMsgHead) + len, .handle = NULL, }; + tscDebug("consumer %ld commit offset of %s on vg %d, offset is %ld", tmq->consumerId, pOffset->subKey, pVg->vgId, + pOffset->version); + + // TODO: put into cb + pVg->committedOffset = pVg->currentOffset; + pMsgSendInfo->requestId = generateRequestId(); pMsgSendInfo->requestObjRefId = 0; pMsgSendInfo->param = pParam; pMsgSendInfo->fp = tmqCommitCb2; - pMsgSendInfo->msgType = TDMT_MND_MQ_COMMIT_OFFSET; + pMsgSendInfo->msgType = TDMT_VND_MQ_COMMIT_OFFSET; // send msg - SEpSet epSet = getEpSet_s(&tmq->pTscObj->pAppInfo->mgmtEp); int64_t transporterId = 0; - asyncSendMsgToServer(tmq->pTscObj->pAppInfo->pTransporter, &epSet, &transporterId, pMsgSendInfo); + asyncSendMsgToServer(tmq->pTscObj->pAppInfo->pTransporter, &pVg->epSet, &transporterId, pMsgSendInfo); pParamSet->waitingRspNum++; pParamSet->totalRspNum++; } } + if (pParamSet->totalRspNum == 0) { + tsem_destroy(&pParamSet->rspSem); + taosMemoryFree(pParamSet); + return 0; + } + if (!async) { tsem_wait(&pParamSet->rspSem); code = pParamSet->rspErr; @@ -483,21 +634,24 @@ int32_t tmqCommitInner2(tmq_t* tmq, const tmq_topic_vgroup_list_t* offsets, int8 if (code != 0 && async) { if (automatic) { - tmq->commitCb(tmq, code, NULL, tmq->commitCbUserParam); + tmq->commitCb(tmq, code, tmq->commitCbUserParam); } else { - userCb(tmq, code, NULL, userParam); + userCb(tmq, code, userParam); } } +#if 0 if (!async) { taosArrayDestroyP(pParamSet->successfulOffsets, taosMemoryFree); taosArrayDestroyP(pParamSet->failedOffsets, taosMemoryFree); } +#endif return 0; } -int32_t tmqCommitInner(tmq_t* tmq, const tmq_topic_vgroup_list_t* offsets, int8_t automatic, int8_t async, +#if 0 +int32_t tmqCommitInner(tmq_t* tmq, const TAOS_RES* msg, int8_t automatic, int8_t async, tmq_commit_cb* userCb, void* userParam) { SMqCMCommitOffsetReq req; SArray* pOffsets = NULL; @@ -507,7 +661,7 @@ int32_t tmqCommitInner(tmq_t* tmq, const tmq_topic_vgroup_list_t* offsets, int8_ int8_t freeOffsets; int32_t code = -1; - if (offsets == NULL) { + if (msg == NULL) { freeOffsets = 1; pOffsets = taosArrayInit(0, sizeof(SMqOffset)); for (int32_t i = 0; i < taosArrayGetSize(tmq->clientTopics); i++) { @@ -524,7 +678,7 @@ int32_t tmqCommitInner(tmq_t* tmq, const tmq_topic_vgroup_list_t* offsets, int8_ } } else { freeOffsets = 0; - pOffsets = (SArray*)&offsets->container; + pOffsets = (SArray*)&msg->container; } req.num = (int32_t)pOffsets->size; @@ -611,6 +765,7 @@ END: } return code; } +#endif void tmqAssignDelayedHbTask(void* param, void* tmrId) { tmq_t* tmq = (tmq_t*)param; @@ -648,7 +803,7 @@ int32_t tmqHandleAllDelayedTask(tmq_t* tmq) { tmqAskEp(tmq, true); taosTmrReset(tmqAssignDelayedHbTask, 1000, tmq, tmqMgmt.timer, &tmq->hbTimer); } else if (*pTaskType == TMQ_DELAYED_TASK__COMMIT) { - tmqCommitInner(tmq, NULL, 1, 1, tmq->commitCb, tmq->commitCbUserParam); + tmqCommitInner2(tmq, NULL, 1, 1, tmq->commitCb, tmq->commitCbUserParam); taosTmrReset(tmqAssignDelayedCommitTask, tmq->autoCommitInterval, tmq, tmqMgmt.timer, &tmq->commitTimer); } else if (*pTaskType == TMQ_DELAYED_TASK__REPORT) { } else { @@ -689,7 +844,7 @@ int32_t tmqSubscribeCb(void* param, const SDataBuf* pMsg, int32_t code) { return 0; } -tmq_resp_err_t tmq_subscription(tmq_t* tmq, tmq_list_t** topics) { +int32_t tmq_subscription(tmq_t* tmq, tmq_list_t** topics) { if (*topics == NULL) { *topics = tmq_list_new(); } @@ -697,12 +852,12 @@ tmq_resp_err_t tmq_subscription(tmq_t* tmq, tmq_list_t** topics) { SMqClientTopic* topic = taosArrayGet(tmq->clientTopics, i); tmq_list_append(*topics, strchr(topic->topicName, '.') + 1); } - return TMQ_RESP_ERR__SUCCESS; + return 0; } -tmq_resp_err_t tmq_unsubscribe(tmq_t* tmq) { - tmq_list_t* lst = tmq_list_new(); - tmq_resp_err_t rsp = tmq_subscribe(tmq, lst); +int32_t tmq_unsubscribe(tmq_t* tmq) { + tmq_list_t* lst = tmq_list_new(); + int32_t rsp = tmq_subscribe(tmq, lst); tmq_list_destroy(lst); return rsp; } @@ -818,11 +973,13 @@ FAIL: return NULL; } -tmq_resp_err_t tmq_commit(tmq_t* tmq, const tmq_topic_vgroup_list_t* offsets, int32_t async) { - return tmqCommitInner(tmq, offsets, 0, async, tmq->commitCb, tmq->commitCbUserParam); +#if 0 +int32_t tmq_commit(tmq_t* tmq, const tmq_topic_vgroup_list_t* offsets, int32_t async) { + return tmqCommitInner2(tmq, offsets, 0, async, tmq->commitCb, tmq->commitCbUserParam); } +#endif -tmq_resp_err_t tmq_subscribe(tmq_t* tmq, const tmq_list_t* topic_list) { +int32_t tmq_subscribe(tmq_t* tmq, const tmq_list_t* topic_list) { const SArray* container = &topic_list->container; int32_t sz = taosArrayGetSize(container); void* buf = NULL; @@ -862,7 +1019,7 @@ tmq_resp_err_t tmq_subscribe(tmq_t* tmq, const tmq_list_t* topic_list) { if (sendInfo == NULL) goto FAIL; SMqSubscribeCbParam param = { - .rspErr = TMQ_RESP_ERR__SUCCESS, + .rspErr = 0, .tmq = tmq, }; @@ -943,6 +1100,19 @@ int32_t tmqPollCb(void* param, const SDataBuf* pMsg, int32_t code) { if (code != 0) { tscWarn("msg discard from vg %d, epoch %d, code:%x", vgId, epoch, code); if (pMsg->pData) taosMemoryFree(pMsg->pData); + if (code == TSDB_CODE_TQ_NO_COMMITTED_OFFSET) { + SMqPollRspWrapper* pRspWrapper = taosAllocateQitem(sizeof(SMqPollRspWrapper), DEF_QITEM); + if (pRspWrapper == NULL) { + taosMemoryFree(pMsg->pData); + tscWarn("msg discard from vg %d, epoch %d since out of memory", vgId, epoch); + goto CREATE_MSG_FAIL; + } + pRspWrapper->tmqRspType = TMQ_MSG_TYPE__END_RSP; + /*pRspWrapper->vgHandle = pVg;*/ + /*pRspWrapper->topicHandle = pTopic;*/ + taosWriteQitem(tmq->mqueue, pRspWrapper); + tsem_post(&tmq->rspSem); + } goto CREATE_MSG_FAIL; } @@ -1017,14 +1187,13 @@ bool tmqUpdateEp2(tmq_t* tmq, int32_t epoch, SMqAskEpRsp* pRsp) { if (pTopicCur->vgs) { int32_t vgNumCur = taosArrayGetSize(pTopicCur->vgs); tscDebug("consumer %ld new vg num: %d", tmq->consumerId, vgNumCur); - if (vgNumCur == 0) break; for (int32_t j = 0; j < vgNumCur; j++) { SMqClientVg* pVgCur = taosArrayGet(pTopicCur->vgs, j); sprintf(vgKey, "%s:%d", pTopicCur->topicName, pVgCur->vgId); - tscDebug("consumer %ld epoch %d vg %d build %s", tmq->consumerId, epoch, pVgCur->vgId, vgKey); + tscDebug("consumer %ld epoch %d vg %d vgKey is %s, offset is %ld", tmq->consumerId, epoch, pVgCur->vgId, vgKey, + pVgCur->currentOffset); taosHashPut(pHash, vgKey, strlen(vgKey), &pVgCur->currentOffset, sizeof(int64_t)); } - break; } } @@ -1032,7 +1201,6 @@ bool tmqUpdateEp2(tmq_t* tmq, int32_t epoch, SMqAskEpRsp* pRsp) { SMqClientTopic topic = {0}; SMqSubTopicEp* pTopicEp = taosArrayGet(pRsp->topics, i); topic.schema = pTopicEp->schema; - taosHashClear(pHash); topic.topicName = strdup(pTopicEp->topic); tstrncpy(topic.db, pTopicEp->db, TSDB_DB_FNAME_LEN); @@ -1049,7 +1217,8 @@ bool tmqUpdateEp2(tmq_t* tmq, int32_t epoch, SMqAskEpRsp* pRsp) { offset = *pOffset; } - tscDebug("consumer %ld(epoch %d) offset of vg %d updated to %ld", tmq->consumerId, epoch, pVgEp->vgId, offset); + tscDebug("consumer %ld(epoch %d) offset of vg %d updated to %ld, vgKey is %s", tmq->consumerId, epoch, + pVgEp->vgId, offset, vgKey); SMqClientVg clientVg = { .pollCnt = 0, .currentOffset = offset, @@ -1076,6 +1245,7 @@ bool tmqUpdateEp2(tmq_t* tmq, int32_t epoch, SMqAskEpRsp* pRsp) { return set; } +#if 1 bool tmqUpdateEp(tmq_t* tmq, int32_t epoch, SMqAskEpRsp* pRsp) { /*printf("call update ep %d\n", epoch);*/ bool set = false; @@ -1160,6 +1330,7 @@ bool tmqUpdateEp(tmq_t* tmq, int32_t epoch, SMqAskEpRsp* pRsp) { atomic_store_32(&tmq->epoch, epoch); return set; } +#endif int32_t tmqAskEpCb(void* param, const SDataBuf* pMsg, int32_t code) { SMqAskEpCbParam* pParam = (SMqAskEpCbParam*)param; @@ -1186,7 +1357,7 @@ int32_t tmqAskEpCb(void* param, const SDataBuf* pMsg, int32_t code) { tDecodeSMqAskEpRsp(POINTER_SHIFT(pMsg->pData, sizeof(SMqRspHead)), &rsp); /*printf("rsp epoch %ld sz %ld\n", rsp.epoch, rsp.topics->size);*/ /*printf("tmq epoch %ld sz %ld\n", tmq->epoch, tmq->clientTopics->size);*/ - tmqUpdateEp(tmq, head->epoch, &rsp); + tmqUpdateEp2(tmq, head->epoch, &rsp); tDeleteSMqAskEpRsp(&rsp); } else { SMqAskEpRspWrapper* pWrapper = taosAllocateQitem(sizeof(SMqAskEpRspWrapper), DEF_QITEM); @@ -1283,7 +1454,8 @@ int32_t tmqAskEp(tmq_t* tmq, bool async) { return code; } -tmq_resp_err_t tmq_seek(tmq_t* tmq, const tmq_topic_vgroup_t* offset) { +#if 0 +int32_t tmq_seek(tmq_t* tmq, const tmq_topic_vgroup_t* offset) { const SMqOffset* pOffset = &offset->offset; if (strcmp(pOffset->cgroup, tmq->groupId) != 0) { return TMQ_RESP_ERR__FAIL; @@ -1305,16 +1477,17 @@ tmq_resp_err_t tmq_seek(tmq_t* tmq, const tmq_topic_vgroup_t* offset) { } return TMQ_RESP_ERR__FAIL; } +#endif SMqPollReq* tmqBuildConsumeReqImpl(tmq_t* tmq, int64_t timeout, SMqClientTopic* pTopic, SMqClientVg* pVg) { int64_t reqOffset; if (pVg->currentOffset >= 0) { reqOffset = pVg->currentOffset; } else { - if (tmq->resetOffsetCfg == TMQ_CONF__RESET_OFFSET__NONE) { - tscError("unable to poll since no committed offset but reset offset is set to none"); - return NULL; - } + /*if (tmq->resetOffsetCfg == TMQ_CONF__RESET_OFFSET__NONE) {*/ + /*tscError("unable to poll since no committed offset but reset offset is set to none");*/ + /*return NULL;*/ + /*}*/ reqOffset = tmq->resetOffsetCfg; } @@ -1326,10 +1499,10 @@ SMqPollReq* tmqBuildConsumeReqImpl(tmq_t* tmq, int64_t timeout, SMqClientTopic* /*strcpy(pReq->topic, pTopic->topicName);*/ /*strcpy(pReq->cgroup, tmq->groupId);*/ - int32_t tlen = strlen(tmq->groupId); - memcpy(pReq->subKey, tmq->groupId, tlen); - pReq->subKey[tlen] = TMQ_SEPARATOR; - strcpy(pReq->subKey + tlen + 1, pTopic->topicName); + int32_t groupLen = strlen(tmq->groupId); + memcpy(pReq->subKey, tmq->groupId, groupLen); + pReq->subKey[groupLen] = TMQ_SEPARATOR; + strcpy(pReq->subKey + groupLen + 1, pTopic->topicName); pReq->withTbName = tmq->withTbName; pReq->timeout = timeout; @@ -1440,7 +1613,7 @@ int32_t tmqHandleNoPollRsp(tmq_t* tmq, SMqRspWrapper* rspWrapper, bool* pReset) if (rspWrapper->epoch > atomic_load_32(&tmq->epoch)) { SMqAskEpRspWrapper* pEpRspWrapper = (SMqAskEpRspWrapper*)rspWrapper; SMqAskEpRsp* rspMsg = &pEpRspWrapper->msg; - tmqUpdateEp(tmq, rspWrapper->epoch, rspMsg); + tmqUpdateEp2(tmq, rspWrapper->epoch, rspMsg); /*tmqClearUnhandleMsg(tmq);*/ *pReset = true; } else { @@ -1462,7 +1635,11 @@ SMqRspObj* tmqHandleAllRsp(tmq_t* tmq, int64_t timeout, bool pollIfReset) { if (rspWrapper == NULL) return NULL; } - if (rspWrapper->tmqRspType == TMQ_MSG_TYPE__POLL_RSP) { + if (rspWrapper->tmqRspType == TMQ_MSG_TYPE__END_RSP) { + taosFreeQitem(rspWrapper); + terrno = TSDB_CODE_TQ_NO_COMMITTED_OFFSET; + return NULL; + } else if (rspWrapper->tmqRspType == TMQ_MSG_TYPE__POLL_RSP) { SMqPollRspWrapper* pollRspWrapper = (SMqPollRspWrapper*)rspWrapper; /*atomic_sub_fetch_32(&tmq->readyRequest, 1);*/ int32_t consumerEpoch = atomic_load_32(&tmq->epoch); @@ -1523,6 +1700,8 @@ TAOS_RES* tmq_consumer_poll(tmq_t* tmq, int64_t timeout) { rspObj = tmqHandleAllRsp(tmq, timeout, false); if (rspObj) { return (TAOS_RES*)rspObj; + } else if (terrno == TSDB_CODE_TQ_NO_COMMITTED_OFFSET) { + return NULL; } if (timeout != -1) { int64_t endTime = taosGetTimestampMs(); @@ -1539,10 +1718,10 @@ TAOS_RES* tmq_consumer_poll(tmq_t* tmq, int64_t timeout) { } } -tmq_resp_err_t tmq_consumer_close(tmq_t* tmq) { +int32_t tmq_consumer_close(tmq_t* tmq) { if (tmq->status == TMQ_CONSUMER_STATUS__READY) { - tmq_resp_err_t rsp = tmq_commit_sync(tmq, NULL); - if (rsp != TMQ_RESP_ERR__SUCCESS) { + int32_t rsp = tmq_commit_sync(tmq, NULL); + if (rsp != 0) { return rsp; } @@ -1550,18 +1729,18 @@ tmq_resp_err_t tmq_consumer_close(tmq_t* tmq) { rsp = tmq_subscribe(tmq, lst); tmq_list_destroy(lst); - if (rsp != TMQ_RESP_ERR__SUCCESS) { + if (rsp != 0) { return rsp; } } // TODO: free resources - return TMQ_RESP_ERR__SUCCESS; + return 0; } -const char* tmq_err2str(tmq_resp_err_t err) { - if (err == TMQ_RESP_ERR__SUCCESS) { +const char* tmq_err2str(int32_t err) { + if (err == 0) { return "success"; - } else if (err == TMQ_RESP_ERR__FAIL) { + } else if (err == -1) { return "fail"; } else { return tstrerror(err); @@ -1607,10 +1786,8 @@ const char* tmq_get_table_name(TAOS_RES* res) { return NULL; } -void tmq_commit_async(tmq_t* tmq, const tmq_topic_vgroup_list_t* offsets, tmq_commit_cb* cb, void* param) { - tmqCommitInner(tmq, offsets, 0, 1, cb, param); +void tmq_commit_async(tmq_t* tmq, const TAOS_RES* msg, tmq_commit_cb* cb, void* param) { + tmqCommitInner2(tmq, msg, 0, 1, cb, param); } -tmq_resp_err_t tmq_commit_sync(tmq_t* tmq, const tmq_topic_vgroup_list_t* offsets) { - return tmqCommitInner(tmq, offsets, 0, 0, NULL, NULL); -} +int32_t tmq_commit_sync(tmq_t* tmq, const TAOS_RES* msg) { return tmqCommitInner2(tmq, msg, 0, 0, NULL, NULL); } diff --git a/source/common/src/systable.c b/source/common/src/systable.c index bcc248d122..192d1d9922 100644 --- a/source/common/src/systable.c +++ b/source/common/src/systable.c @@ -77,7 +77,7 @@ static const SSysDbTableSchema userDBSchema[] = { {.name = "ntables", .bytes = 8, .type = TSDB_DATA_TYPE_BIGINT}, {.name = "replica", .bytes = 1, .type = TSDB_DATA_TYPE_TINYINT}, {.name = "strict", .bytes = 9 + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_VARCHAR}, - {.name = "duration", .bytes = 4, .type = TSDB_DATA_TYPE_INT}, + {.name = "duration", .bytes = 10 + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_VARCHAR}, {.name = "keep", .bytes = 32 + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_VARCHAR}, {.name = "buffer", .bytes = 4, .type = TSDB_DATA_TYPE_INT}, {.name = "pagesize", .bytes = 4, .type = TSDB_DATA_TYPE_INT}, @@ -302,7 +302,7 @@ static const SSysDbTableSchema offsetSchema[] = { }; static const SSysDbTableSchema querySchema[] = { - {.name = "query_id", .bytes = 26 + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_VARCHAR}, + {.name = "query_id", .bytes = TSDB_QUERY_ID_LEN + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_VARCHAR}, {.name = "req_id", .bytes = 8, .type = TSDB_DATA_TYPE_UBIGINT}, {.name = "connId", .bytes = 4, .type = TSDB_DATA_TYPE_UINT}, {.name = "app", .bytes = TSDB_APP_NAME_LEN + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_VARCHAR}, @@ -317,6 +317,24 @@ static const SSysDbTableSchema querySchema[] = { {.name = "sql", .bytes = TSDB_SHOW_SQL_LEN + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_VARCHAR}, }; +static const SSysDbTableSchema appSchema[] = { + {.name = "app_id", .bytes = 8, .type = TSDB_DATA_TYPE_UBIGINT}, + {.name = "ip", .bytes = TSDB_IPv4ADDR_LEN + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_VARCHAR}, + {.name = "pid", .bytes = 4, .type = TSDB_DATA_TYPE_INT}, + {.name = "name", .bytes = TSDB_APP_NAME_LEN + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_VARCHAR}, + {.name = "start_time", .bytes = 8 , .type = TSDB_DATA_TYPE_TIMESTAMP}, + {.name = "insert_req", .bytes = 8 , .type = TSDB_DATA_TYPE_UBIGINT}, + {.name = "insert_row", .bytes = 8 , .type = TSDB_DATA_TYPE_UBIGINT}, + {.name = "insert_time", .bytes = 8 , .type = TSDB_DATA_TYPE_UBIGINT}, + {.name = "insert_bytes", .bytes = 8 , .type = TSDB_DATA_TYPE_UBIGINT}, + {.name = "fetch_bytes", .bytes = 8 , .type = TSDB_DATA_TYPE_UBIGINT}, + {.name = "query_time", .bytes = 8 , .type = TSDB_DATA_TYPE_UBIGINT}, + {.name = "show_query", .bytes = 8 , .type = TSDB_DATA_TYPE_UBIGINT}, + {.name = "total_req", .bytes = 8 , .type = TSDB_DATA_TYPE_UBIGINT}, + {.name = "current_req", .bytes = 8 , .type = TSDB_DATA_TYPE_UBIGINT}, + {.name = "last_access", .bytes = 8, .type = TSDB_DATA_TYPE_TIMESTAMP}, +}; + static const SSysTableMeta perfsMeta[] = { {TSDB_PERFS_TABLE_CONNECTIONS, connectionsSchema, tListLen(connectionsSchema)}, {TSDB_PERFS_TABLE_QUERIES, querySchema, tListLen(querySchema)}, @@ -327,6 +345,7 @@ static const SSysTableMeta perfsMeta[] = { {TSDB_PERFS_TABLE_TRANS, transSchema, tListLen(transSchema)}, {TSDB_PERFS_TABLE_SMAS, smaSchema, tListLen(smaSchema)}, {TSDB_PERFS_TABLE_STREAMS, streamSchema, tListLen(streamSchema)}, + {TSDB_PERFS_TABLE_APPS, appSchema, tListLen(appSchema)} }; void getInfosDbMeta(const SSysTableMeta** pInfosTableMeta, size_t* size) { diff --git a/source/common/src/tglobal.c b/source/common/src/tglobal.c index fbb4f78425..7457fe7eb6 100644 --- a/source/common/src/tglobal.c +++ b/source/common/src/tglobal.c @@ -51,15 +51,16 @@ int32_t tsNumOfShmThreads = 1; int32_t tsNumOfRpcThreads = 1; int32_t tsNumOfCommitThreads = 2; int32_t tsNumOfTaskQueueThreads = 1; -int32_t tsNumOfMnodeQueryThreads = 1; +int32_t tsNumOfMnodeQueryThreads = 2; +int32_t tsNumOfMnodeFetchThreads = 1; int32_t tsNumOfMnodeReadThreads = 1; int32_t tsNumOfVnodeQueryThreads = 2; -int32_t tsNumOfVnodeFetchThreads = 2; +int32_t tsNumOfVnodeFetchThreads = 1; int32_t tsNumOfVnodeWriteThreads = 2; int32_t tsNumOfVnodeSyncThreads = 2; int32_t tsNumOfVnodeMergeThreads = 2; int32_t tsNumOfQnodeQueryThreads = 2; -int32_t tsNumOfQnodeFetchThreads = 2; +int32_t tsNumOfQnodeFetchThreads = 1; int32_t tsNumOfSnodeSharedThreads = 2; int32_t tsNumOfSnodeUniqueThreads = 2; @@ -417,8 +418,7 @@ static int32_t taosAddServerCfg(SConfig *pCfg) { tsNumOfVnodeQueryThreads = TMAX(tsNumOfVnodeQueryThreads, 1); if (cfgAddInt32(pCfg, "numOfVnodeQueryThreads", tsNumOfVnodeQueryThreads, 1, 1024, 0) != 0) return -1; - tsNumOfVnodeFetchThreads = tsNumOfCores / 2; - tsNumOfVnodeFetchThreads = TRANGE(tsNumOfVnodeFetchThreads, 2, 4); + tsNumOfVnodeFetchThreads = TRANGE(tsNumOfVnodeFetchThreads, 1, 1); if (cfgAddInt32(pCfg, "numOfVnodeFetchThreads", tsNumOfVnodeFetchThreads, 1, 1024, 0) != 0) return -1; tsNumOfVnodeWriteThreads = tsNumOfCores; @@ -437,8 +437,7 @@ static int32_t taosAddServerCfg(SConfig *pCfg) { tsNumOfQnodeQueryThreads = TMAX(tsNumOfQnodeQueryThreads, 1); if (cfgAddInt32(pCfg, "numOfQnodeQueryThreads", tsNumOfQnodeQueryThreads, 1, 1024, 0) != 0) return -1; - tsNumOfQnodeFetchThreads = tsNumOfCores / 2; - tsNumOfQnodeFetchThreads = TRANGE(tsNumOfQnodeFetchThreads, 2, 4); + tsNumOfQnodeFetchThreads = TRANGE(tsNumOfQnodeFetchThreads, 1, 1); if (cfgAddInt32(pCfg, "numOfQnodeFetchThreads", tsNumOfQnodeFetchThreads, 1, 1024, 0) != 0) return -1; tsNumOfSnodeSharedThreads = tsNumOfCores / 4; diff --git a/source/common/src/tmsg.c b/source/common/src/tmsg.c index c22b9e5f2e..2443ffbb7f 100644 --- a/source/common/src/tmsg.c +++ b/source/common/src/tmsg.c @@ -191,13 +191,25 @@ static int32_t tSerializeSClientHbReq(SEncoder *pEncoder, const SClientHbReq *pR if (tEncodeSClientHbKey(pEncoder, &pReq->connKey) < 0) return -1; if (pReq->connKey.connType == CONN_TYPE__QUERY) { + if (tEncodeI64(pEncoder, pReq->app.appId) < 0) return -1; + if (tEncodeI32(pEncoder, pReq->app.pid) < 0) return -1; + if (tEncodeCStr(pEncoder, pReq->app.name) < 0) return -1; + if (tEncodeI64(pEncoder, pReq->app.startTime) < 0) return -1; + if (tEncodeU64(pEncoder, pReq->app.summary.numOfInsertsReq) < 0) return -1; + if (tEncodeU64(pEncoder, pReq->app.summary.numOfInsertRows) < 0) return -1; + if (tEncodeU64(pEncoder, pReq->app.summary.insertElapsedTime) < 0) return -1; + if (tEncodeU64(pEncoder, pReq->app.summary.insertBytes) < 0) return -1; + if (tEncodeU64(pEncoder, pReq->app.summary.fetchBytes) < 0) return -1; + if (tEncodeU64(pEncoder, pReq->app.summary.queryElapsedTime) < 0) return -1; + if (tEncodeU64(pEncoder, pReq->app.summary.numOfSlowQueries) < 0) return -1; + if (tEncodeU64(pEncoder, pReq->app.summary.totalRequests) < 0) return -1; + if (tEncodeU64(pEncoder, pReq->app.summary.currentRequests) < 0) return -1; + int32_t queryNum = 0; if (pReq->query) { queryNum = 1; if (tEncodeI32(pEncoder, queryNum) < 0) return -1; if (tEncodeU32(pEncoder, pReq->query->connId) < 0) return -1; - if (tEncodeI32(pEncoder, pReq->query->pid) < 0) return -1; - if (tEncodeCStr(pEncoder, pReq->query->app) < 0) return -1; int32_t num = taosArrayGetSize(pReq->query->queryDesc); if (tEncodeI32(pEncoder, num) < 0) return -1; @@ -209,7 +221,6 @@ static int32_t tSerializeSClientHbReq(SEncoder *pEncoder, const SClientHbReq *pR if (tEncodeI64(pEncoder, desc->useconds) < 0) return -1; if (tEncodeI64(pEncoder, desc->stime) < 0) return -1; if (tEncodeI64(pEncoder, desc->reqRid) < 0) return -1; - if (tEncodeI32(pEncoder, desc->pid) < 0) return -1; if (tEncodeI8(pEncoder, desc->stableQuery) < 0) return -1; if (tEncodeCStr(pEncoder, desc->fqdn) < 0) return -1; if (tEncodeI32(pEncoder, desc->subPlanNum) < 0) return -1; @@ -243,14 +254,26 @@ static int32_t tDeserializeSClientHbReq(SDecoder *pDecoder, SClientHbReq *pReq) if (tDecodeSClientHbKey(pDecoder, &pReq->connKey) < 0) return -1; if (pReq->connKey.connType == CONN_TYPE__QUERY) { + if (tDecodeI64(pDecoder, &pReq->app.appId) < 0) return -1; + if (tDecodeI32(pDecoder, &pReq->app.pid) < 0) return -1; + if (tDecodeCStrTo(pDecoder, pReq->app.name) < 0) return -1; + if (tDecodeI64(pDecoder, &pReq->app.startTime) < 0) return -1; + if (tDecodeU64(pDecoder, &pReq->app.summary.numOfInsertsReq) < 0) return -1; + if (tDecodeU64(pDecoder, &pReq->app.summary.numOfInsertRows) < 0) return -1; + if (tDecodeU64(pDecoder, &pReq->app.summary.insertElapsedTime) < 0) return -1; + if (tDecodeU64(pDecoder, &pReq->app.summary.insertBytes) < 0) return -1; + if (tDecodeU64(pDecoder, &pReq->app.summary.fetchBytes) < 0) return -1; + if (tDecodeU64(pDecoder, &pReq->app.summary.queryElapsedTime) < 0) return -1; + if (tDecodeU64(pDecoder, &pReq->app.summary.numOfSlowQueries) < 0) return -1; + if (tDecodeU64(pDecoder, &pReq->app.summary.totalRequests) < 0) return -1; + if (tDecodeU64(pDecoder, &pReq->app.summary.currentRequests) < 0) return -1; + int32_t queryNum = 0; if (tDecodeI32(pDecoder, &queryNum) < 0) return -1; if (queryNum) { pReq->query = taosMemoryCalloc(1, sizeof(*pReq->query)); if (NULL == pReq->query) return -1; if (tDecodeU32(pDecoder, &pReq->query->connId) < 0) return -1; - if (tDecodeI32(pDecoder, &pReq->query->pid) < 0) return -1; - if (tDecodeCStrTo(pDecoder, pReq->query->app) < 0) return -1; int32_t num = 0; if (tDecodeI32(pDecoder, &num) < 0) return -1; @@ -265,7 +288,6 @@ static int32_t tDeserializeSClientHbReq(SDecoder *pDecoder, SClientHbReq *pReq) if (tDecodeI64(pDecoder, &desc.useconds) < 0) return -1; if (tDecodeI64(pDecoder, &desc.stime) < 0) return -1; if (tDecodeI64(pDecoder, &desc.reqRid) < 0) return -1; - if (tDecodeI32(pDecoder, &desc.pid) < 0) return -1; if (tDecodeI8(pDecoder, (int8_t *)&desc.stableQuery) < 0) return -1; if (tDecodeCStrTo(pDecoder, desc.fqdn) < 0) return -1; if (tDecodeI32(pDecoder, &desc.subPlanNum) < 0) return -1; @@ -3414,7 +3436,7 @@ int32_t tSerializeSKillConnReq(void *buf, int32_t bufLen, SKillConnReq *pReq) { tEncoderInit(&encoder, buf, bufLen); if (tStartEncode(&encoder) < 0) return -1; - if (tEncodeI32(&encoder, pReq->connId) < 0) return -1; + if (tEncodeU32(&encoder, pReq->connId) < 0) return -1; tEndEncode(&encoder); int32_t tlen = encoder.pos; @@ -3427,7 +3449,7 @@ int32_t tDeserializeSKillConnReq(void *buf, int32_t bufLen, SKillConnReq *pReq) tDecoderInit(&decoder, buf, bufLen); if (tStartDecode(&decoder) < 0) return -1; - if (tDecodeI32(&decoder, &pReq->connId) < 0) return -1; + if (tDecodeU32(&decoder, &pReq->connId) < 0) return -1; tEndDecode(&decoder); tDecoderClear(&decoder); diff --git a/source/dnode/mgmt/mgmt_mnode/inc/mmInt.h b/source/dnode/mgmt/mgmt_mnode/inc/mmInt.h index c5c3d76f1e..a4c37dd334 100644 --- a/source/dnode/mgmt/mgmt_mnode/inc/mmInt.h +++ b/source/dnode/mgmt/mgmt_mnode/inc/mmInt.h @@ -30,6 +30,7 @@ typedef struct SMnodeMgmt { const char *path; const char *name; SSingleWorker queryWorker; + SSingleWorker fetchWorker; SSingleWorker readWorker; SSingleWorker writeWorker; SSingleWorker syncWorker; @@ -57,6 +58,7 @@ int32_t mmPutMsgToWriteQueue(SMnodeMgmt *pMgmt, SRpcMsg *pMsg); int32_t mmPutMsgToSyncQueue(SMnodeMgmt *pMgmt, SRpcMsg *pMsg); int32_t mmPutMsgToReadQueue(SMnodeMgmt *pMgmt, SRpcMsg *pMsg); int32_t mmPutMsgToQueryQueue(SMnodeMgmt *pMgmt, SRpcMsg *pMsg); +int32_t mmPutMsgToFetchQueue(SMnodeMgmt *pMgmt, SRpcMsg *pMsg); int32_t mmPutMsgToMonitorQueue(SMnodeMgmt *pMgmt, SRpcMsg *pMsg); int32_t mmPutMsgToQueue(SMnodeMgmt *pMgmt, EQueueType qtype, SRpcMsg *pRpc); diff --git a/source/dnode/mgmt/mgmt_mnode/src/mmWorker.c b/source/dnode/mgmt/mgmt_mnode/src/mmWorker.c index 493f4ab85f..88f667ce9e 100644 --- a/source/dnode/mgmt/mgmt_mnode/src/mmWorker.c +++ b/source/dnode/mgmt/mgmt_mnode/src/mmWorker.c @@ -122,6 +122,13 @@ int32_t mmPutMsgToQueryQueue(SMnodeMgmt *pMgmt, SRpcMsg *pMsg) { return mmPutMsgToWorker(pMgmt, &pMgmt->queryWorker, pMsg); } +int32_t mmPutMsgToFetchQueue(SMnodeMgmt *pMgmt, SRpcMsg *pMsg) { + pMsg->info.node = pMgmt->pMnode; + + return mmPutMsgToWorker(pMgmt, &pMgmt->fetchWorker, pMsg); +} + + int32_t mmPutMsgToMonitorQueue(SMnodeMgmt *pMgmt, SRpcMsg *pMsg) { return mmPutMsgToWorker(pMgmt, &pMgmt->monitorWorker, pMsg); } @@ -135,6 +142,9 @@ int32_t mmPutMsgToQueue(SMnodeMgmt *pMgmt, EQueueType qtype, SRpcMsg *pRpc) { case QUERY_QUEUE: pWorker = &pMgmt->queryWorker; break; + case FETCH_QUEUE: + pWorker = &pMgmt->fetchWorker; + break; case READ_QUEUE: pWorker = &pMgmt->readWorker; break; @@ -167,6 +177,18 @@ int32_t mmStartWorker(SMnodeMgmt *pMgmt) { return -1; } + SSingleWorkerCfg fCfg = { + .min = tsNumOfMnodeFetchThreads, + .max = tsNumOfMnodeFetchThreads, + .name = "mnode-fetch", + .fp = (FItem)mmProcessRpcMsg, + .param = pMgmt, + }; + if (tSingleWorkerInit(&pMgmt->fetchWorker, &fCfg) != 0) { + dError("failed to start mnode-fetch worker since %s", terrstr()); + return -1; + } + SSingleWorkerCfg rCfg = { .min = tsNumOfMnodeReadThreads, .max = tsNumOfMnodeReadThreads, @@ -224,6 +246,7 @@ void mmStopWorker(SMnodeMgmt *pMgmt) { tSingleWorkerCleanup(&pMgmt->monitorWorker); tSingleWorkerCleanup(&pMgmt->queryWorker); + tSingleWorkerCleanup(&pMgmt->fetchWorker); tSingleWorkerCleanup(&pMgmt->readWorker); tSingleWorkerCleanup(&pMgmt->writeWorker); tSingleWorkerCleanup(&pMgmt->syncWorker); diff --git a/source/dnode/mnode/impl/inc/mndInt.h b/source/dnode/mnode/impl/inc/mndInt.h index cc9bc5b634..37bae1d5c0 100644 --- a/source/dnode/mnode/impl/inc/mndInt.h +++ b/source/dnode/mnode/impl/inc/mndInt.h @@ -67,7 +67,8 @@ typedef struct { } SShowMgmt; typedef struct { - SCacheObj *cache; + SCacheObj *connCache; + SCacheObj *appCache; } SProfileMgmt; typedef struct { diff --git a/source/dnode/mnode/impl/src/mndDb.c b/source/dnode/mnode/impl/src/mndDb.c index 2eeff9cb33..49f6aa1edf 100644 --- a/source/dnode/mnode/impl/src/mndDb.c +++ b/source/dnode/mnode/impl/src/mndDb.c @@ -1391,11 +1391,13 @@ static void dumpDbInfoData(SSDataBlock *pBlock, SDbObj *pDb, SShowObj *pShow, in pColInfo = taosArrayGet(pBlock->pDataBlock, cols++); colDataAppend(pColInfo, rows, (const char *)strict, false); - pColInfo = taosArrayGet(pBlock->pDataBlock, cols++); - colDataAppend(pColInfo, rows, (const char *)&pDb->cfg.daysPerFile, false); - char tmp[128] = {0}; int32_t len = 0; + len = sprintf(&tmp[VARSTR_HEADER_SIZE], "%dm", pDb->cfg.daysPerFile); + varDataSetLen(tmp, len); + pColInfo = taosArrayGet(pBlock->pDataBlock, cols++); + colDataAppend(pColInfo, rows, (const char *)tmp, false); + if (pDb->cfg.daysToKeep0 > pDb->cfg.daysToKeep1 || pDb->cfg.daysToKeep0 > pDb->cfg.daysToKeep2) { len = sprintf(&tmp[VARSTR_HEADER_SIZE], "%dm,%dm,%dm", pDb->cfg.daysToKeep1, pDb->cfg.daysToKeep2, pDb->cfg.daysToKeep0); diff --git a/source/dnode/mnode/impl/src/mndDnode.c b/source/dnode/mnode/impl/src/mndDnode.c index d1bed90175..58c3570c36 100644 --- a/source/dnode/mnode/impl/src/mndDnode.c +++ b/source/dnode/mnode/impl/src/mndDnode.c @@ -761,11 +761,12 @@ static int32_t mndRetrieveDnodes(SRpcMsg *pReq, SShowObj *pShow, SSDataBlock *pB pColInfo = taosArrayGet(pBlock->pDataBlock, cols++); colDataAppend(pColInfo, numOfRows, (const char *)&pDnode->createdTime, false); - char b[tListLen(offlineReason) + VARSTR_HEADER_SIZE] = {0}; + char *b = taosMemoryCalloc(VARSTR_HEADER_SIZE + strlen(offlineReason[pDnode->offlineReason]) + 1, 1); STR_TO_VARSTR(b, online ? "" : offlineReason[pDnode->offlineReason]); pColInfo = taosArrayGet(pBlock->pDataBlock, cols++); colDataAppend(pColInfo, numOfRows, b, false); + taosMemoryFreeClear(b); numOfRows++; sdbRelease(pSdb, pDnode); diff --git a/source/dnode/mnode/impl/src/mndMain.c b/source/dnode/mnode/impl/src/mndMain.c index 7b5025a986..bc418e0f53 100644 --- a/source/dnode/mnode/impl/src/mndMain.c +++ b/source/dnode/mnode/impl/src/mndMain.c @@ -510,8 +510,7 @@ static int32_t mndCheckMnodeState(SRpcMsg *pMsg) { if (IsReq(pMsg) && pMsg->msgType != TDMT_MND_MQ_TIMER && pMsg->msgType != TDMT_MND_TELEM_TIMER && pMsg->msgType != TDMT_MND_TRANS_TIMER) { - mError("msg:%p, failed to check mnode state since %s, app:%p type:%s", pMsg, terrstr(), pMsg->info.ahandle, - TMSG_INFO(pMsg->msgType)); + mError("msg:%p, failed to check mnode state since %s, type:%s", pMsg, terrstr(), TMSG_INFO(pMsg->msgType)); SEpSet epSet = {0}; mndGetMnodeEpSet(pMsg->info.node, &epSet); @@ -534,7 +533,8 @@ static int32_t mndCheckMsgContent(SRpcMsg *pMsg) { if (!IsReq(pMsg)) return 0; if (pMsg->contLen != 0 && pMsg->pCont != NULL) return 0; - mError("msg:%p, failed to check msg content, app:%p type:%s", pMsg, pMsg->info.ahandle, TMSG_INFO(pMsg->msgType)); + mError("msg:%p, failed to check msg, cont:%p contLen:%d, app:%p type:%s", pMsg, pMsg->pCont, pMsg->contLen, + pMsg->info.ahandle, TMSG_INFO(pMsg->msgType)); terrno = TSDB_CODE_INVALID_MSG_LEN; return -1; } @@ -558,7 +558,7 @@ int32_t mndProcessRpcMsg(SRpcMsg *pMsg) { if (code == TSDB_CODE_ACTION_IN_PROGRESS) { mTrace("msg:%p, won't response immediately since in progress", pMsg); } else if (code == 0) { - mTrace("msg:%p, successfully processed and response", pMsg); + mTrace("msg:%p, successfully processed", pMsg); } else { mError("msg:%p, failed to process since %s, app:%p type:%s", pMsg, terrstr(), pMsg->info.ahandle, TMSG_INFO(pMsg->msgType)); diff --git a/source/dnode/mnode/impl/src/mndProfile.c b/source/dnode/mnode/impl/src/mndProfile.c index 39d21aad49..fd80679316 100644 --- a/source/dnode/mnode/impl/src/mndProfile.c +++ b/source/dnode/mnode/impl/src/mndProfile.c @@ -43,6 +43,16 @@ typedef struct { SArray *pQueries; // SArray } SConnObj; +typedef struct { + int64_t appId; + uint32_t ip; + int32_t pid; + char name[TSDB_APP_NAME_LEN]; + int64_t startTime; + SAppClusterSummary summary; + int64_t lastAccessTimeMs; +} SAppObj; + static SConnObj *mndCreateConn(SMnode *pMnode, const char *user, int8_t connType, uint32_t ip, uint16_t port, int32_t pid, const char *app, int64_t startTime); static void mndFreeConn(SConnObj *pConn); @@ -57,14 +67,24 @@ static int32_t mndProcessKillConnReq(SRpcMsg *pReq); static int32_t mndRetrieveConns(SRpcMsg *pReq, SShowObj *pShow, SSDataBlock *pBlock, int32_t rows); static int32_t mndRetrieveQueries(SRpcMsg *pReq, SShowObj *pShow, SSDataBlock *pBlock, int32_t rows); static void mndCancelGetNextQuery(SMnode *pMnode, void *pIter); +static void mndFreeApp(SAppObj *pApp); +static int32_t mndRetrieveApps(SRpcMsg *pReq, SShowObj *pShow, SSDataBlock *pBlock, int32_t rows); +static void mndCancelGetNextApp(SMnode *pMnode, void *pIter); int32_t mndInitProfile(SMnode *pMnode) { SProfileMgmt *pMgmt = &pMnode->profileMgmt; // in ms - int32_t connCheckTime = tsShellActivityTimer * 2 * 1000; - pMgmt->cache = taosCacheInit(TSDB_DATA_TYPE_INT, connCheckTime, true, (__cache_free_fn_t)mndFreeConn, "conn"); - if (pMgmt->cache == NULL) { + int32_t checkTime = tsShellActivityTimer * 2 * 1000; + pMgmt->connCache = taosCacheInit(TSDB_DATA_TYPE_UINT, checkTime, true, (__cache_free_fn_t)mndFreeConn, "conn"); + if (pMgmt->connCache == NULL) { + terrno = TSDB_CODE_OUT_OF_MEMORY; + mError("failed to alloc profile cache since %s", terrstr()); + return -1; + } + + pMgmt->appCache = taosCacheInit(TSDB_DATA_TYPE_BIGINT, checkTime, true, (__cache_free_fn_t)mndFreeApp, "app"); + if (pMgmt->appCache == NULL) { terrno = TSDB_CODE_OUT_OF_MEMORY; mError("failed to alloc profile cache since %s", terrstr()); return -1; @@ -79,15 +99,22 @@ int32_t mndInitProfile(SMnode *pMnode) { mndAddShowFreeIterHandle(pMnode, TSDB_MGMT_TABLE_CONNS, mndCancelGetNextConn); mndAddShowRetrieveHandle(pMnode, TSDB_MGMT_TABLE_QUERIES, mndRetrieveQueries); mndAddShowFreeIterHandle(pMnode, TSDB_MGMT_TABLE_QUERIES, mndCancelGetNextQuery); + mndAddShowRetrieveHandle(pMnode, TSDB_MGMT_TABLE_APPS, mndRetrieveApps); + mndAddShowFreeIterHandle(pMnode, TSDB_MGMT_TABLE_APPS, mndCancelGetNextApp); return 0; } void mndCleanupProfile(SMnode *pMnode) { SProfileMgmt *pMgmt = &pMnode->profileMgmt; - if (pMgmt->cache != NULL) { - taosCacheCleanup(pMgmt->cache); - pMgmt->cache = NULL; + if (pMgmt->connCache != NULL) { + taosCacheCleanup(pMgmt->connCache); + pMgmt->connCache = NULL; + } + + if (pMgmt->appCache != NULL) { + taosCacheCleanup(pMgmt->appCache); + pMgmt->appCache = NULL; } } @@ -97,7 +124,7 @@ static SConnObj *mndCreateConn(SMnode *pMnode, const char *user, int8_t connType char connStr[255] = {0}; int32_t len = snprintf(connStr, sizeof(connStr), "%s%d%d%d%s", user, ip, port, pid, app); - int32_t connId = mndGenerateUid(connStr, len); + uint32_t connId = mndGenerateUid(connStr, len); if (startTime == 0) startTime = taosGetTimestampMs(); SConnObj connObj = {.id = connId, @@ -118,7 +145,7 @@ static SConnObj *mndCreateConn(SMnode *pMnode, const char *user, int8_t connType tstrncpy(connObj.app, app, TSDB_APP_NAME_LEN); int32_t keepTime = tsShellActivityTimer * 3; - SConnObj *pConn = taosCachePut(pMgmt->cache, &connId, sizeof(int32_t), &connObj, sizeof(connObj), keepTime * 1000); + SConnObj *pConn = taosCachePut(pMgmt->connCache, &connId, sizeof(uint32_t), &connObj, sizeof(connObj), keepTime * 1000); if (pConn == NULL) { terrno = TSDB_CODE_OUT_OF_MEMORY; mError("conn:%d, failed to put into cache since %s, user:%s", connId, user, terrstr()); @@ -140,14 +167,13 @@ static void mndFreeConn(SConnObj *pConn) { static SConnObj *mndAcquireConn(SMnode *pMnode, uint32_t connId) { SProfileMgmt *pMgmt = &pMnode->profileMgmt; - SConnObj *pConn = taosCacheAcquireByKey(pMgmt->cache, &connId, sizeof(connId)); + SConnObj *pConn = taosCacheAcquireByKey(pMgmt->connCache, &connId, sizeof(connId)); if (pConn == NULL) { mDebug("conn:%u, already destroyed", connId); return NULL; } - int32_t keepTime = tsShellActivityTimer * 3; - pConn->lastAccessTimeMs = keepTime * 1000 + (uint64_t)taosGetTimestampMs(); + pConn->lastAccessTimeMs = taosGetTimestampMs(); mTrace("conn:%u, acquired from cache, data:%p", pConn->id, pConn); return pConn; @@ -158,7 +184,7 @@ static void mndReleaseConn(SMnode *pMnode, SConnObj *pConn) { mTrace("conn:%u, released from cache, data:%p", pConn->id, pConn); SProfileMgmt *pMgmt = &pMnode->profileMgmt; - taosCacheRelease(pMgmt->cache, (void **)&pConn, false); + taosCacheRelease(pMgmt->connCache, (void **)&pConn, false); } void *mndGetNextConn(SMnode *pMnode, SCacheIter *pIter) { @@ -276,6 +302,77 @@ static int32_t mndSaveQueryList(SConnObj *pConn, SQueryHbReqBasic *pBasic) { return TSDB_CODE_SUCCESS; } +static SAppObj *mndCreateApp(SMnode *pMnode, uint32_t clientIp, SAppHbReq* pReq) { + SProfileMgmt *pMgmt = &pMnode->profileMgmt; + + SAppObj app; + app.appId = pReq->appId; + app.ip = clientIp; + app.pid = pReq->pid; + strcpy(app.name, pReq->name); + app.startTime = pReq->startTime; + memcpy(&app.summary, &pReq->summary, sizeof(pReq->summary)); + app.lastAccessTimeMs = taosGetTimestampMs(); + + int32_t keepTime = tsShellActivityTimer * 3; + SAppObj *pApp = taosCachePut(pMgmt->appCache, &pReq->appId, sizeof(pReq->appId), &app, sizeof(app), keepTime * 1000); + if (pApp == NULL) { + terrno = TSDB_CODE_OUT_OF_MEMORY; + mError("failed to app %" PRIx64 " into cache since %s", pReq->appId, terrstr()); + return NULL; + } + + mTrace("app %" PRIx64 " is put into cache", pReq->appId); + return pApp; +} + +static void mndFreeApp(SAppObj *pApp) { + mTrace("app %" PRIx64 " is destroyed", pApp->appId); +} + + +static SAppObj *mndAcquireApp(SMnode *pMnode, int64_t appId) { + SProfileMgmt *pMgmt = &pMnode->profileMgmt; + + SAppObj *pApp = taosCacheAcquireByKey(pMgmt->appCache, &appId, sizeof(appId)); + if (pApp == NULL) { + mDebug("app %" PRIx64 " not in cache", appId); + return NULL; + } + + pApp->lastAccessTimeMs = (uint64_t)taosGetTimestampMs(); + + mTrace("app %" PRIx64 " acquired from cache", appId); + return pApp; +} + +static void mndReleaseApp(SMnode *pMnode, SAppObj *pApp) { + if (pApp == NULL) return; + mTrace("release app %" PRIx64 " to cache", pApp->appId); + + SProfileMgmt *pMgmt = &pMnode->profileMgmt; + taosCacheRelease(pMgmt->appCache, (void **)&pApp, false); +} + +void *mndGetNextApp(SMnode *pMnode, SCacheIter *pIter) { + SAppObj *pApp = NULL; + bool hasNext = taosCacheIterNext(pIter); + if (hasNext) { + size_t dataLen = 0; + pApp = taosCacheIterGetData(pIter, &dataLen); + } else { + taosCacheDestroyIter(pIter); + } + + return pApp; +} + +static void mndCancelGetNextApp(SMnode *pMnode, void *pIter) { + if (pIter != NULL) { + taosCacheDestroyIter(pIter); + } +} + static SClientHbRsp *mndMqHbBuildRsp(SMnode *pMnode, SClientHbReq *pReq) { #if 0 SClientHbRsp* pRsp = taosMemoryMalloc(sizeof(SClientHbRsp)); @@ -341,25 +438,48 @@ static SClientHbRsp *mndMqHbBuildRsp(SMnode *pMnode, SClientHbReq *pReq) { return NULL; } +static int32_t mndUpdateAppInfo(SMnode *pMnode, SClientHbReq *pHbReq, SRpcConnInfo *connInfo) { + SAppHbReq* pReq = &pHbReq->app; + SAppObj *pApp = mndAcquireApp(pMnode, pReq->appId); + if (pApp == NULL) { + pApp = mndCreateApp(pMnode, connInfo->clientIp, pReq); + if (pApp == NULL) { + mError("failed to create new app %" PRIx64 " since %s", pReq->appId, terrstr()); + return -1; + } else { + mDebug("a new app %" PRIx64 "created", pReq->appId); + mndReleaseApp(pMnode, pApp); + return TSDB_CODE_SUCCESS; + } + } + + memcpy(&pApp->summary, &pReq->summary, sizeof(pReq->summary)); + + mndReleaseApp(pMnode, pApp); + + return TSDB_CODE_SUCCESS; +} + static int32_t mndProcessQueryHeartBeat(SMnode *pMnode, SRpcMsg *pMsg, SClientHbReq *pHbReq, SClientHbBatchRsp *pBatchRsp) { SProfileMgmt *pMgmt = &pMnode->profileMgmt; SClientHbRsp hbRsp = {.connKey = pHbReq->connKey, .status = 0, .info = NULL, .query = NULL}; + SRpcConnInfo connInfo = pMsg->info.conn; + + mndUpdateAppInfo(pMnode, pHbReq, &connInfo); if (pHbReq->query) { SQueryHbReqBasic *pBasic = pHbReq->query; - SRpcConnInfo connInfo = pMsg->info.conn; - SConnObj *pConn = mndAcquireConn(pMnode, pBasic->connId); if (pConn == NULL) { pConn = mndCreateConn(pMnode, connInfo.user, CONN_TYPE__QUERY, connInfo.clientIp, connInfo.clientPort, - pBasic->pid, pBasic->app, 0); + pHbReq->app.pid, pHbReq->app.name, 0); if (pConn == NULL) { mError("user:%s, conn:%u is freed and failed to create new since %s", connInfo.user, pBasic->connId, terrstr()); return -1; } else { - mDebug("user:%s, conn:%u is freed and create a new conn:%u", connInfo.user, pBasic->connId, pConn->id); + mDebug("user:%s, conn:%u is freed, will create a new conn:%u", connInfo.user, pBasic->connId, pConn->id); } } @@ -516,17 +636,28 @@ static int32_t mndProcessKillQueryReq(SRpcMsg *pReq) { return -1; } - mInfo("kill query msg is received, queryId:%d", killReq.queryId); + mInfo("kill query msg is received, queryId:%s", killReq.queryStrId); + int32_t connId = 0; + uint64_t queryId = 0; + char* p = strchr(killReq.queryStrId, ':'); + if (NULL == p) { + mError("invalid query id %s", killReq.queryStrId); + terrno = TSDB_CODE_MND_INVALID_QUERY_ID; + return -1; + } + *p = 0; + connId = taosStr2Int32(killReq.queryStrId, NULL, 16); + queryId = taosStr2UInt64(p + 1, NULL, 16); - SConnObj *pConn = taosCacheAcquireByKey(pMgmt->cache, &killReq.connId, sizeof(int32_t)); + SConnObj *pConn = taosCacheAcquireByKey(pMgmt->connCache, &connId, sizeof(int32_t)); if (pConn == NULL) { - mError("connId:%d, failed to kill queryId:%d, conn not exist", killReq.connId, killReq.queryId); + mError("connId:%x, failed to kill queryId:%" PRIx64 ", conn not exist", connId, queryId); terrno = TSDB_CODE_MND_INVALID_CONN_ID; return -1; } else { - mInfo("connId:%d, queryId:%d is killed by user:%s", killReq.connId, killReq.queryId, pReq->info.conn.user); - pConn->killId = killReq.queryId; - taosCacheRelease(pMgmt->cache, (void **)&pConn, false); + mInfo("connId:%x, queryId:%" PRIx64 " is killed by user:%s", connId, queryId, pReq->info.conn.user); + pConn->killId = queryId; + taosCacheRelease(pMgmt->connCache, (void **)&pConn, false); return 0; } } @@ -550,15 +681,15 @@ static int32_t mndProcessKillConnReq(SRpcMsg *pReq) { return -1; } - SConnObj *pConn = taosCacheAcquireByKey(pMgmt->cache, &killReq.connId, sizeof(int32_t)); + SConnObj *pConn = taosCacheAcquireByKey(pMgmt->connCache, &killReq.connId, sizeof(uint32_t)); if (pConn == NULL) { - mError("connId:%d, failed to kill connection, conn not exist", killReq.connId); + mError("connId:%u, failed to kill connection, conn not exist", killReq.connId); terrno = TSDB_CODE_MND_INVALID_CONN_ID; return -1; } else { - mInfo("connId:%d, is killed by user:%s", killReq.connId, pReq->info.conn.user); + mInfo("connId:%u, is killed by user:%s", killReq.connId, pReq->info.conn.user); pConn->killed = 1; - taosCacheRelease(pMgmt->cache, (void **)&pConn, false); + taosCacheRelease(pMgmt->connCache, (void **)&pConn, false); return TSDB_CODE_SUCCESS; } } @@ -572,7 +703,7 @@ static int32_t mndRetrieveConns(SRpcMsg *pReq, SShowObj *pShow, SSDataBlock *pBl if (pShow->pIter == NULL) { SProfileMgmt *pMgmt = &pMnode->profileMgmt; - pShow->pIter = taosCacheCreateIter(pMgmt->cache); + pShow->pIter = taosCacheCreateIter(pMgmt->connCache); } while (numOfRows < rows) { @@ -628,7 +759,7 @@ static int32_t mndRetrieveQueries(SRpcMsg *pReq, SShowObj *pShow, SSDataBlock *p if (pShow->pIter == NULL) { SProfileMgmt *pMgmt = &pMnode->profileMgmt; - pShow->pIter = taosCacheCreateIter(pMgmt->cache); + pShow->pIter = taosCacheCreateIter(pMgmt->connCache); } while (numOfRows < rows) { @@ -667,7 +798,7 @@ static int32_t mndRetrieveQueries(SRpcMsg *pReq, SShowObj *pShow, SSDataBlock *p colDataAppend(pColInfo, numOfRows, (const char *)app, false); pColInfo = taosArrayGet(pBlock->pDataBlock, cols++); - colDataAppend(pColInfo, numOfRows, (const char *)&pQuery->pid, false); + colDataAppend(pColInfo, numOfRows, (const char *)&pConn->pid, false); char user[TSDB_USER_LEN + VARSTR_HEADER_SIZE] = {0}; STR_TO_VARSTR(user, pConn->user); @@ -721,6 +852,86 @@ static int32_t mndRetrieveQueries(SRpcMsg *pReq, SShowObj *pShow, SSDataBlock *p return numOfRows; } +static int32_t mndRetrieveApps(SRpcMsg *pReq, SShowObj *pShow, SSDataBlock *pBlock, int32_t rows) { + SMnode *pMnode = pReq->info.node; + SSdb *pSdb = pMnode->pSdb; + int32_t numOfRows = 0; + int32_t cols = 0; + SAppObj *pApp = NULL; + + if (pShow->pIter == NULL) { + SProfileMgmt *pMgmt = &pMnode->profileMgmt; + pShow->pIter = taosCacheCreateIter(pMgmt->appCache); + } + + while (numOfRows < rows) { + pApp = mndGetNextApp(pMnode, pShow->pIter); + if (pApp == NULL) { + pShow->pIter = NULL; + break; + } + + cols = 0; + + SColumnInfoData *pColInfo = taosArrayGet(pBlock->pDataBlock, cols++); + colDataAppend(pColInfo, numOfRows, (const char *)&pApp->appId, false); + + char ip[TSDB_IPv4ADDR_LEN + 6 + VARSTR_HEADER_SIZE] = {0}; + sprintf(&ip[VARSTR_HEADER_SIZE], "%s", taosIpStr(pApp->ip)); + varDataLen(ip) = strlen(&ip[VARSTR_HEADER_SIZE]); + pColInfo = taosArrayGet(pBlock->pDataBlock, cols++); + colDataAppend(pColInfo, numOfRows, (const char *)ip, false); + + pColInfo = taosArrayGet(pBlock->pDataBlock, cols++); + colDataAppend(pColInfo, numOfRows, (const char *)&pApp->pid, false); + + char name[TSDB_APP_NAME_LEN + 6 + VARSTR_HEADER_SIZE] = {0}; + sprintf(&name[VARSTR_HEADER_SIZE], "%s", pApp->name); + varDataLen(name) = strlen(&name[VARSTR_HEADER_SIZE]); + pColInfo = taosArrayGet(pBlock->pDataBlock, cols++); + colDataAppend(pColInfo, numOfRows, (const char *)name, false); + + pColInfo = taosArrayGet(pBlock->pDataBlock, cols++); + colDataAppend(pColInfo, numOfRows, (const char *)&pApp->startTime, false); + + pColInfo = taosArrayGet(pBlock->pDataBlock, cols++); + colDataAppend(pColInfo, numOfRows, (const char *)&pApp->summary.numOfInsertsReq, false); + + pColInfo = taosArrayGet(pBlock->pDataBlock, cols++); + colDataAppend(pColInfo, numOfRows, (const char *)&pApp->summary.numOfInsertRows, false); + + pColInfo = taosArrayGet(pBlock->pDataBlock, cols++); + colDataAppend(pColInfo, numOfRows, (const char *)&pApp->summary.insertElapsedTime, false); + + pColInfo = taosArrayGet(pBlock->pDataBlock, cols++); + colDataAppend(pColInfo, numOfRows, (const char *)&pApp->summary.insertBytes, false); + + pColInfo = taosArrayGet(pBlock->pDataBlock, cols++); + colDataAppend(pColInfo, numOfRows, (const char *)&pApp->summary.fetchBytes, false); + + pColInfo = taosArrayGet(pBlock->pDataBlock, cols++); + colDataAppend(pColInfo, numOfRows, (const char *)&pApp->summary.queryElapsedTime, false); + + pColInfo = taosArrayGet(pBlock->pDataBlock, cols++); + colDataAppend(pColInfo, numOfRows, (const char *)&pApp->summary.numOfSlowQueries, false); + + pColInfo = taosArrayGet(pBlock->pDataBlock, cols++); + colDataAppend(pColInfo, numOfRows, (const char *)&pApp->summary.totalRequests, false); + + pColInfo = taosArrayGet(pBlock->pDataBlock, cols++); + colDataAppend(pColInfo, numOfRows, (const char *)&pApp->summary.currentRequests, false); + + pColInfo = taosArrayGet(pBlock->pDataBlock, cols++); + colDataAppend(pColInfo, numOfRows, (const char *)&pApp->lastAccessTimeMs, false); + + numOfRows++; + } + + pShow->numOfRows += numOfRows; + return numOfRows; +} + + static void mndCancelGetNextQuery(SMnode *pMnode, void *pIter) { if (pIter != NULL) { taosCacheDestroyIter(pIter); @@ -729,5 +940,5 @@ static void mndCancelGetNextQuery(SMnode *pMnode, void *pIter) { int32_t mndGetNumOfConnections(SMnode *pMnode) { SProfileMgmt *pMgmt = &pMnode->profileMgmt; - return taosCacheGetNumOfObj(pMgmt->cache); + return taosCacheGetNumOfObj(pMgmt->connCache); } diff --git a/source/dnode/mnode/impl/src/mndShow.c b/source/dnode/mnode/impl/src/mndShow.c index a6447ed405..fed6a5a721 100644 --- a/source/dnode/mnode/impl/src/mndShow.c +++ b/source/dnode/mnode/impl/src/mndShow.c @@ -104,6 +104,8 @@ static int32_t convertToRetrieveType(char *name, int32_t len) { type = TSDB_MGMT_TABLE_TOPICS; } else if (strncasecmp(name, TSDB_PERFS_TABLE_STREAMS, len) == 0) { type = TSDB_MGMT_TABLE_STREAMS; + } else if (strncasecmp(name, TSDB_PERFS_TABLE_APPS, len) == 0) { + type = TSDB_MGMT_TABLE_APPS; } else { // ASSERT(0); } diff --git a/source/dnode/mnode/impl/src/mndTrans.c b/source/dnode/mnode/impl/src/mndTrans.c index 858c6f956f..11e07cb9fa 100644 --- a/source/dnode/mnode/impl/src/mndTrans.c +++ b/source/dnode/mnode/impl/src/mndTrans.c @@ -661,7 +661,7 @@ static int32_t mndTransSync(SMnode *pMnode, STrans *pTrans) { } sdbSetRawStatus(pRaw, SDB_STATUS_READY); - mDebug("trans:%d, sync to other mnodes", pTrans->id); + mDebug("trans:%d, sync to other mnodes, stage:%s", pTrans->id, mndTransStr(pTrans->stage)); int32_t code = mndSyncPropose(pMnode, pRaw, pTrans->id); if (code != 0) { mError("trans:%d, failed to sync since %s", pTrans->id, terrstr()); diff --git a/source/dnode/mnode/impl/src/mndVgroup.c b/source/dnode/mnode/impl/src/mndVgroup.c index cec83d1af5..94ddbcd409 100644 --- a/source/dnode/mnode/impl/src/mndVgroup.c +++ b/source/dnode/mnode/impl/src/mndVgroup.c @@ -1058,7 +1058,7 @@ int32_t mndSetMoveVgroupsInfoToTrans(SMnode *pMnode, STrans *pTrans, int32_t del static int32_t mndAddIncVgroupReplicaToTrans(SMnode *pMnode, STrans *pTrans, SDbObj *pDb, SVgObj *pVgroup, int32_t newDnodeId) { - mDebug("vgId:%d, will add 1 vnode, replica:%d, dnode:%d", pVgroup->vgId, pVgroup->replica, newDnodeId); + mDebug("vgId:%d, will add 1 vnode, replica:%d dnode:%d", pVgroup->vgId, pVgroup->replica, newDnodeId); SVnodeGid *pGid = &pVgroup->vnodeGid[pVgroup->replica]; pVgroup->replica++; @@ -1074,7 +1074,7 @@ static int32_t mndAddIncVgroupReplicaToTrans(SMnode *pMnode, STrans *pTrans, SDb static int32_t mndAddDecVgroupReplicaFromTrans(SMnode *pMnode, STrans *pTrans, SDbObj *pDb, SVgObj *pVgroup, int32_t delDnodeId) { - mDebug("vgId:%d, will remove 1 vnode, replica:%d, dnode:%d", pVgroup->vgId, pVgroup->replica, delDnodeId); + mDebug("vgId:%d, will remove 1 vnode, replica:%d dnode:%d", pVgroup->vgId, pVgroup->replica, delDnodeId); SVnodeGid *pGid = NULL; SVnodeGid delGid = {0}; @@ -1116,7 +1116,8 @@ static int32_t mndRedistributeVgroup(SMnode *pMnode, SRpcMsg *pReq, SDbObj *pDb, memcpy(&newVg, pVgroup, sizeof(SVgObj)); mInfo("vgId:%d, vgroup info before redistribute, replica:%d", newVg.vgId, newVg.replica); for (int32_t i = 0; i < newVg.replica; ++i) { - mInfo("vgId:%d, vnode:%d dnode:%d", newVg.vgId, i, newVg.vnodeGid[i].dnodeId); + mInfo("vgId:%d, vnode:%d dnode:%d role:%s", newVg.vgId, i, newVg.vnodeGid[i].dnodeId, + syncStr(newVg.vnodeGid[i].role)); } if (pNew1 != NULL && pOld1 != NULL) { @@ -1198,7 +1199,7 @@ static int32_t mndProcessRedistributeVgroupMsg(SRpcMsg *pReq) { goto _OVER; } - mInfo("vgId:%d, start to redistribute to dnode %d:%d:%d", req.vgId, req.dnodeId1, req.dnodeId2, req.dnodeId3); + mInfo("vgId:%d, start to redistribute vgroup to dnode %d:%d:%d", req.vgId, req.dnodeId1, req.dnodeId2, req.dnodeId3); if (mndCheckOperAuth(pMnode, pReq->info.conn.user, MND_OPER_REDISTRIBUTE_VGROUP) != 0) goto _OVER; diff --git a/source/dnode/mnode/impl/test/profile/profile.cpp b/source/dnode/mnode/impl/test/profile/profile.cpp index 039700ae9e..e784a41d6f 100644 --- a/source/dnode/mnode/impl/test/profile/profile.cpp +++ b/source/dnode/mnode/impl/test/profile/profile.cpp @@ -295,8 +295,7 @@ TEST_F(MndTestProfile, 07_KillQueryMsg) { TEST_F(MndTestProfile, 08_KillQueryMsg_InvalidConn) { SKillQueryReq killReq = {0}; - killReq.connId = 2345; - killReq.queryId = 2345; + strcpy(killReq.queryStrId, "2345:2345"); int32_t contLen = tSerializeSKillQueryReq(NULL, 0, &killReq); void* pReq = rpcMallocCont(contLen); diff --git a/source/dnode/vnode/src/meta/metaQuery.c b/source/dnode/vnode/src/meta/metaQuery.c index a6339125c4..8cfebe1b98 100644 --- a/source/dnode/vnode/src/meta/metaQuery.c +++ b/source/dnode/vnode/src/meta/metaQuery.c @@ -628,7 +628,7 @@ int32_t metaFilteTableIds(SMeta *pMeta, SMetaFltParam *param, SArray *pUids) { void * tagData = NULL; if (param->val == NULL) { - metaError("vgId:%d failed to filter NULL data", TD_VID(pMeta->pVnode)); + metaError("vgId:%d, failed to filter NULL data", TD_VID(pMeta->pVnode)); return -1; } else { if (IS_VAR_DATA_TYPE(param->type)) { diff --git a/source/dnode/vnode/src/tq/tq.c b/source/dnode/vnode/src/tq/tq.c index c9d3267adb..70b09ec701 100644 --- a/source/dnode/vnode/src/tq/tq.c +++ b/source/dnode/vnode/src/tq/tq.c @@ -124,17 +124,20 @@ int32_t tqProcessOffsetCommitReq(STQ* pTq, char* msg, int32_t msgLen) { tDecoderClear(&decoder); if (offset.type == TMQ_OFFSET__SNAPSHOT) { - tqDebug("receive offset commit msg to %s, offset(type:snapshot) uid: %ld, ts: %ld", offset.subKey, offset.uid, - offset.ts); + tqDebug("receive offset commit msg to %s on vg %d, offset(type:snapshot) uid: %ld, ts: %ld", offset.subKey, + pTq->pVnode->config.vgId, offset.uid, offset.ts); } else if (offset.type == TMQ_OFFSET__LOG) { - tqDebug("receive offset commit msg to %s, offset(type:log) version: %ld", offset.subKey, offset.version); + tqDebug("receive offset commit msg to %s on vg %d, offset(type:log) version: %ld", offset.subKey, + pTq->pVnode->config.vgId, offset.version); } else { ASSERT(0); } - - if (tqOffsetWrite(pTq->pOffsetStore, &offset) < 0) { - ASSERT(0); - return -1; + STqOffset* pOffset = tqOffsetRead(pTq->pOffsetStore, offset.subKey); + if (pOffset == NULL || pOffset->version < offset.version) { + if (tqOffsetWrite(pTq->pOffsetStore, &offset) < 0) { + ASSERT(0); + return -1; + } } return 0; @@ -149,16 +152,33 @@ int32_t tqProcessPollReq(STQ* pTq, SRpcMsg* pMsg, int32_t workerId) { int32_t code = 0; // get offset to fetch message - if (pReq->currentOffset == TMQ_CONF__RESET_OFFSET__EARLIEAST) { - fetchOffset = walGetFirstVer(pTq->pWal); - } else if (pReq->currentOffset == TMQ_CONF__RESET_OFFSET__LATEST) { - fetchOffset = walGetCommittedVer(pTq->pWal); - } else { + if (pReq->currentOffset >= 0) { fetchOffset = pReq->currentOffset + 1; + } else { + STqOffset* pOffset = tqOffsetRead(pTq->pOffsetStore, pReq->subKey); + if (pOffset != NULL) { + ASSERT(pOffset->type == TMQ_OFFSET__LOG); + tqDebug("consumer %ld, restore offset of %s on vg %d, offset(type:log) version: %ld", consumerId, pReq->subKey, + pTq->pVnode->config.vgId, pOffset->version); + fetchOffset = pOffset->version + 1; + } else { + if (pReq->currentOffset == TMQ_CONF__RESET_OFFSET__EARLIEAST) { + fetchOffset = walGetFirstVer(pTq->pWal); + } else if (pReq->currentOffset == TMQ_CONF__RESET_OFFSET__LATEST) { + fetchOffset = walGetCommittedVer(pTq->pWal); + } else if (pReq->currentOffset == TMQ_CONF__RESET_OFFSET__NONE) { + tqError("tmq poll: no offset committed for consumer %ld in vg %d, subkey %s", consumerId, + pTq->pVnode->config.vgId, pReq->subKey); + terrno = TSDB_CODE_TQ_NO_COMMITTED_OFFSET; + return -1; + } + tqDebug("consumer %ld, restore offset of %s on vg %d failed, config is %ld, set to %ld", consumerId, pReq->subKey, + pTq->pVnode->config.vgId, pReq->currentOffset, fetchOffset); + } } - tqDebug("tmq poll: consumer %ld (epoch %d) recv poll req in vg %d, req %ld %ld", consumerId, pReq->epoch, - TD_VID(pTq->pVnode), pReq->currentOffset, fetchOffset); + tqDebug("tmq poll: consumer %ld (epoch %d) recv poll req in vg %d, req offset %ld fetch offset %ld", consumerId, + pReq->epoch, TD_VID(pTq->pVnode), pReq->currentOffset, fetchOffset); STqHandle* pHandle = taosHashGet(pTq->handles, pReq->subKey, strlen(pReq->subKey)); /*ASSERT(pHandle);*/ diff --git a/source/dnode/vnode/src/tq/tqCommit.c b/source/dnode/vnode/src/tq/tqCommit.c index 7b116bff2e..639da22b1c 100644 --- a/source/dnode/vnode/src/tq/tqCommit.c +++ b/source/dnode/vnode/src/tq/tqCommit.c @@ -15,7 +15,4 @@ #include "tq.h" -int tqCommit(STQ* pTq) { - // do nothing - return 0; -} +int tqCommit(STQ* pTq) { return tqOffsetSnapshot(pTq->pOffsetStore); } diff --git a/source/dnode/vnode/src/tq/tqOffset.c b/source/dnode/vnode/src/tq/tqOffset.c index 8d6cb28065..e5475d7c30 100644 --- a/source/dnode/vnode/src/tq/tqOffset.c +++ b/source/dnode/vnode/src/tq/tqOffset.c @@ -92,6 +92,8 @@ STqOffset* tqOffsetRead(STqOffsetStore* pStore, const char* subscribeKey) { } int32_t tqOffsetWrite(STqOffsetStore* pStore, const STqOffset* pOffset) { + ASSERT(pOffset->type == TMQ_OFFSET__LOG); + ASSERT(pOffset->version >= 0); return taosHashPut(pStore->pHash, pOffset->subKey, strlen(pOffset->subKey), pOffset, sizeof(STqOffset)); } @@ -129,7 +131,7 @@ int32_t tqOffsetSnapshot(STqOffsetStore* pStore) { tEncodeSTqOffset(&encoder, pOffset); // write file int64_t writeLen; - if ((writeLen = taosWriteFile(pFile, buf, totLen)) != bodyLen) { + if ((writeLen = taosWriteFile(pFile, buf, totLen)) != totLen) { ASSERT(0); tqError("write offset incomplete, len %d, write len %ld", bodyLen, writeLen); taosHashCancelIterate(pStore->pHash, pIter); diff --git a/source/dnode/vnode/src/vnd/vnodeSync.c b/source/dnode/vnode/src/vnd/vnodeSync.c index f46114746e..fded723f1a 100644 --- a/source/dnode/vnode/src/vnd/vnodeSync.c +++ b/source/dnode/vnode/src/vnd/vnodeSync.c @@ -61,6 +61,7 @@ static int32_t vnodeSetStandBy(SVnode *pVnode) { return -1; } + vInfo("vgId:%d, start to transfer leader", TD_VID(pVnode)); if (syncLeaderTransfer(pVnode->sync) != 0) { vError("vgId:%d, failed to transfer leader since:%s", TD_VID(pVnode), terrstr()); return -1; @@ -297,7 +298,7 @@ int32_t vnodeProcessSyncReq(SVnode *pVnode, SRpcMsg *pMsg, SRpcMsg **pRsp) { ret = -1; } - if (ret != 0) { + if (ret != 0 && terrno == 0) { terrno = TSDB_CODE_SYN_INTERNAL_ERROR; } return ret; diff --git a/source/libs/catalog/inc/catalogInt.h b/source/libs/catalog/inc/catalogInt.h index df4b9b6ea6..7c90c3538c 100644 --- a/source/libs/catalog/inc/catalogInt.h +++ b/source/libs/catalog/inc/catalogInt.h @@ -59,6 +59,7 @@ enum { CTG_OP_UPDATE_VG_EPSET, CTG_OP_UPDATE_TB_INDEX, CTG_OP_DROP_TB_INDEX, + CTG_OP_CLEAR_CACHE, CTG_OP_MAX }; @@ -328,6 +329,10 @@ typedef struct SCtgDropTbIndexMsg { char tbName[TSDB_TABLE_NAME_LEN]; } SCtgDropTbIndexMsg; +typedef struct SCtgClearCacheMsg { + SCatalog* pCtg; +} SCtgClearCacheMsg; + typedef struct SCtgUpdateEpsetMsg { SCatalog* pCtg; char dbFName[TSDB_DB_FNAME_LEN]; @@ -471,8 +476,8 @@ typedef struct SCtgOperation { #define CTG_API_LEAVE(c) do { int32_t __code = c; CTG_UNLOCK(CTG_READ, &gCtgMgmt.lock); CTG_API_DEBUG("CTG API leave %s", __FUNCTION__); CTG_RET(__code); } while (0) #define CTG_API_ENTER() do { CTG_API_DEBUG("CTG API enter %s", __FUNCTION__); CTG_LOCK(CTG_READ, &gCtgMgmt.lock); if (atomic_load_8((int8_t*)&gCtgMgmt.exit)) { CTG_API_LEAVE(TSDB_CODE_CTG_OUT_OF_SERVICE); } } while (0) -void ctgdShowTableMeta(SCatalog* pCtg, const char *tbName, STableMeta* p); -void ctgdShowClusterCache(SCatalog* pCtg); +void ctgdShowTableMeta(SCatalog* pCtg, const char *tbName, STableMeta* p); +void ctgdShowClusterCache(SCatalog* pCtg); int32_t ctgdShowCacheInfo(void); int32_t ctgRemoveTbMetaFromCache(SCatalog* pCtg, SName* pTableName, bool syncReq); @@ -487,8 +492,8 @@ int32_t ctgOpDropTbMeta(SCtgCacheOperation *action); int32_t ctgOpUpdateUser(SCtgCacheOperation *action); int32_t ctgOpUpdateEpset(SCtgCacheOperation *operation); int32_t ctgAcquireVgInfoFromCache(SCatalog* pCtg, const char *dbFName, SCtgDBCache **pCache); -void ctgReleaseDBCache(SCatalog *pCtg, SCtgDBCache *dbCache); -void ctgRUnlockVgInfo(SCtgDBCache *dbCache); +void ctgReleaseDBCache(SCatalog *pCtg, SCtgDBCache *dbCache); +void ctgRUnlockVgInfo(SCtgDBCache *dbCache); int32_t ctgTbMetaExistInCache(SCatalog* pCtg, char *dbFName, char* tbName, int32_t *exist); int32_t ctgReadTbMetaFromCache(SCatalog* pCtg, SCtgTbMetaCtx* ctx, STableMeta** pTableMeta); int32_t ctgReadTbVerFromCache(SCatalog *pCtg, SName *pTableName, int32_t *sver, int32_t *tver, int32_t *tbType, uint64_t *suid, char *stbName); @@ -502,17 +507,20 @@ int32_t ctgUpdateTbMetaEnqueue(SCatalog* pCtg, STableMetaOutput *output, bool sy int32_t ctgUpdateUserEnqueue(SCatalog* pCtg, SGetUserAuthRsp *pAuth, bool syncReq); int32_t ctgUpdateVgEpsetEnqueue(SCatalog* pCtg, char *dbFName, int32_t vgId, SEpSet* pEpSet); int32_t ctgUpdateTbIndexEnqueue(SCatalog* pCtg, STableIndex **pIndex, bool syncOp); +int32_t ctgClearCacheEnqueue(SCatalog* pCtg, bool syncOp); int32_t ctgMetaRentInit(SCtgRentMgmt *mgmt, uint32_t rentSec, int8_t type); int32_t ctgMetaRentAdd(SCtgRentMgmt *mgmt, void *meta, int64_t id, int32_t size); int32_t ctgMetaRentGet(SCtgRentMgmt *mgmt, void **res, uint32_t *num, int32_t size); int32_t ctgUpdateTbMetaToCache(SCatalog* pCtg, STableMetaOutput* pOut, bool syncReq); int32_t ctgStartUpdateThread(); int32_t ctgRelaunchGetTbMetaTask(SCtgTask *pTask); -void ctgReleaseVgInfoToCache(SCatalog* pCtg, SCtgDBCache *dbCache); +void ctgReleaseVgInfoToCache(SCatalog* pCtg, SCtgDBCache *dbCache); int32_t ctgReadTbIndexFromCache(SCatalog* pCtg, SName* pTableName, SArray** pRes); int32_t ctgDropTbIndexEnqueue(SCatalog* pCtg, SName* pName, bool syncOp); int32_t ctgOpDropTbIndex(SCtgCacheOperation *operation); int32_t ctgOpUpdateTbIndex(SCtgCacheOperation *operation); +int32_t ctgOpClearCache(SCtgCacheOperation *operation); + @@ -535,22 +543,22 @@ int32_t ctgMakeAsyncRes(SCtgJob *pJob); int32_t ctgCloneVgInfo(SDBVgInfo *src, SDBVgInfo **dst); int32_t ctgCloneMetaOutput(STableMetaOutput *output, STableMetaOutput **pOutput); int32_t ctgGenerateVgList(SCatalog *pCtg, SHashObj *vgHash, SArray** pList); -void ctgFreeJob(void* job); -void ctgFreeHandle(SCatalog* pCtg); -void ctgFreeVgInfo(SDBVgInfo *vgInfo); +void ctgFreeJob(void* job); +void ctgFreeHandle(SCatalog* pCtg); +void ctgFreeVgInfo(SDBVgInfo *vgInfo); int32_t ctgGetVgInfoFromHashValue(SCatalog *pCtg, SDBVgInfo *dbInfo, const SName *pTableName, SVgroupInfo *pVgroup); -void ctgResetTbMetaTask(SCtgTask* pTask); -void ctgFreeDbCache(SCtgDBCache *dbCache); +void ctgResetTbMetaTask(SCtgTask* pTask); +void ctgFreeDbCache(SCtgDBCache *dbCache); int32_t ctgStbVersionSortCompare(const void* key1, const void* key2); int32_t ctgDbVgVersionSortCompare(const void* key1, const void* key2); int32_t ctgStbVersionSearchCompare(const void* key1, const void* key2); int32_t ctgDbVgVersionSearchCompare(const void* key1, const void* key2); -void ctgFreeSTableMetaOutput(STableMetaOutput* pOutput); +void ctgFreeSTableMetaOutput(STableMetaOutput* pOutput); int32_t ctgUpdateMsgCtx(SCtgMsgCtx* pCtx, int32_t reqType, void* out, char* target); -char *ctgTaskTypeStr(CTG_TASK_TYPE type); +char * ctgTaskTypeStr(CTG_TASK_TYPE type); int32_t ctgUpdateSendTargetInfo(SMsgSendInfo *pMsgSendInfo, int32_t msgType, SCtgTask* pTask); int32_t ctgCloneTableIndex(SArray* pIndex, SArray** pRes); -void ctgFreeSTableIndex(void *info); +void ctgFreeSTableIndex(void *info); extern SCatalogMgmt gCtgMgmt; diff --git a/source/libs/catalog/src/catalog.c b/source/libs/catalog/src/catalog.c index a3576f8738..bb02895569 100644 --- a/source/libs/catalog/src/catalog.c +++ b/source/libs/catalog/src/catalog.c @@ -105,7 +105,7 @@ int32_t ctgRefreshDBVgInfo(SCatalog* pCtg, SRequestConnInfo *pConn, const char* code = ctgGetDBVgInfoFromMnode(pCtg, pConn, &input, &DbOut, NULL); if (code) { if (CTG_DB_NOT_EXIST(code) && (NULL != dbCache)) { - ctgDebug("db no longer exist, dbFName:%s, dbId:%" PRIx64, input.db, input.dbId); + ctgDebug("db no longer exist, dbFName:%s, dbId:0x%" PRIx64, input.db, input.dbId); ctgDropDbCacheEnqueue(pCtg, input.db, input.dbId); } @@ -571,7 +571,7 @@ int32_t catalogGetHandle(uint64_t clusterId, SCatalog** catalogHandle) { } if (NULL == gCtgMgmt.pCluster) { - qError("catalog cluster cache are not ready, clusterId:%" PRIx64, clusterId); + qError("catalog cluster cache are not ready, clusterId:0x%" PRIx64, clusterId); CTG_ERR_RET(TSDB_CODE_CTG_NOT_READY); } @@ -583,7 +583,7 @@ int32_t catalogGetHandle(uint64_t clusterId, SCatalog** catalogHandle) { if (ctg && (*ctg)) { *catalogHandle = *ctg; - qDebug("got catalog handle from cache, clusterId:%" PRIx64 ", CTG:%p", clusterId, *ctg); + qDebug("got catalog handle from cache, clusterId:0x%" PRIx64 ", CTG:%p", clusterId, *ctg); return TSDB_CODE_SUCCESS; } @@ -612,11 +612,11 @@ int32_t catalogGetHandle(uint64_t clusterId, SCatalog** catalogHandle) { continue; } - qError("taosHashPut CTG to cache failed, clusterId:%" PRIx64, clusterId); + qError("taosHashPut CTG to cache failed, clusterId:0x%" PRIx64, clusterId); CTG_ERR_JRET(TSDB_CODE_CTG_INTERNAL_ERROR); } - qDebug("add CTG to cache, clusterId:%" PRIx64 ", CTG:%p", clusterId, clusterCtg); + qDebug("add CTG to cache, clusterId:0x%" PRIx64 ", CTG:%p", clusterId, clusterCtg); break; } @@ -640,7 +640,7 @@ void catalogFreeHandle(SCatalog* pCtg) { } if (taosHashRemove(gCtgMgmt.pCluster, &pCtg->clusterId, sizeof(pCtg->clusterId))) { - ctgWarn("taosHashRemove from cluster failed, may already be freed, clusterId:%" PRIx64, pCtg->clusterId); + ctgWarn("taosHashRemove from cluster failed, may already be freed, clusterId:0x%" PRIx64, pCtg->clusterId); return; } @@ -650,7 +650,7 @@ void catalogFreeHandle(SCatalog* pCtg) { ctgFreeHandle(pCtg); - ctgInfo("handle freed, culsterId:%" PRIx64, clusterId); + ctgInfo("handle freed, culsterId:0x%" PRIx64, clusterId); } int32_t catalogGetDBVgVersion(SCatalog* pCtg, const char* dbFName, int32_t* version, int64_t* dbId, int32_t* tableNum) { @@ -1247,6 +1247,23 @@ int32_t catalogUpdateUserAuthInfo(SCatalog* pCtg, SGetUserAuthRsp* pAuth) { CTG_API_LEAVE(ctgUpdateUserEnqueue(pCtg, pAuth, false)); } +int32_t catalogClearCache(void) { + CTG_API_ENTER(); + + qInfo("start to clear catalog cache"); + + if (NULL == gCtgMgmt.pCluster || atomic_load_8((int8_t*)&gCtgMgmt.exit)) { + CTG_API_LEAVE(TSDB_CODE_SUCCESS); + } + + int32_t code = ctgClearCacheEnqueue(NULL, true); + + qInfo("clear catalog cache end, code: %s", tstrerror(code)); + + CTG_API_LEAVE(code); +} + + void catalogDestroy(void) { qInfo("start to destroy catalog"); diff --git a/source/libs/catalog/src/ctgAsync.c b/source/libs/catalog/src/ctgAsync.c index 8dabd56934..6adadf5045 100644 --- a/source/libs/catalog/src/ctgAsync.c +++ b/source/libs/catalog/src/ctgAsync.c @@ -622,7 +622,7 @@ int32_t ctgHandleTaskEnd(SCtgTask* pTask, int32_t rspCode) { SCtgJob* pJob = pTask->pJob; int32_t code = 0; - qDebug("QID:0x%" PRIx64 " task %d end with rsp %s", pJob->queryId, pTask->taskId, tstrerror(rspCode)); + qDebug("QID:0x%" PRIx64 " task %d end with res %s", pJob->queryId, pTask->taskId, tstrerror(rspCode)); pTask->code = rspCode; @@ -1276,7 +1276,7 @@ int32_t ctgLaunchJob(SCtgJob *pJob) { for (int32_t i = 0; i < taskNum; ++i) { SCtgTask *pTask = taosArrayGet(pJob->pTasks, i); - qDebug("QID:0x%" PRIx64 " start to launch task %d", pJob->queryId, pTask->taskId); + qDebug("QID:0x%" PRIx64 " ctg start to launch task %d", pJob->queryId, pTask->taskId); CTG_ERR_RET((*gCtgAsyncFps[pTask->type].launchFp)(pTask)); } diff --git a/source/libs/catalog/src/ctgCache.c b/source/libs/catalog/src/ctgCache.c index 0948c01270..1de5ee3d7d 100644 --- a/source/libs/catalog/src/ctgCache.c +++ b/source/libs/catalog/src/ctgCache.c @@ -69,6 +69,11 @@ SCtgOperation gCtgCacheOperation[CTG_OP_MAX] = { CTG_OP_DROP_TB_INDEX, "drop tbIndex", ctgOpDropTbIndex + }, + { + CTG_OP_CLEAR_CACHE, + "clear cache", + ctgOpClearCache } }; @@ -81,7 +86,7 @@ int32_t ctgRLockVgInfo(SCatalog *pCtg, SCtgDBCache *dbCache, bool *inCache) { if (dbCache->deleted) { CTG_UNLOCK(CTG_READ, &dbCache->vgCache.vgLock); - ctgDebug("db is dropping, dbId:%"PRIx64, dbCache->dbId); + ctgDebug("db is dropping, dbId:0x%"PRIx64, dbCache->dbId); *inCache = false; return TSDB_CODE_SUCCESS; @@ -92,7 +97,7 @@ int32_t ctgRLockVgInfo(SCatalog *pCtg, SCtgDBCache *dbCache, bool *inCache) { CTG_UNLOCK(CTG_READ, &dbCache->vgCache.vgLock); *inCache = false; - ctgDebug("db vgInfo is empty, dbId:%"PRIx64, dbCache->dbId); + ctgDebug("db vgInfo is empty, dbId:0x%"PRIx64, dbCache->dbId); return TSDB_CODE_SUCCESS; } @@ -105,7 +110,7 @@ int32_t ctgWLockVgInfo(SCatalog *pCtg, SCtgDBCache *dbCache) { CTG_LOCK(CTG_WRITE, &dbCache->vgCache.vgLock); if (dbCache->deleted) { - ctgDebug("db is dropping, dbId:%"PRIx64, dbCache->dbId); + ctgDebug("db is dropping, dbId:0x%"PRIx64, dbCache->dbId); CTG_UNLOCK(CTG_WRITE, &dbCache->vgCache.vgLock); CTG_ERR_RET(TSDB_CODE_CTG_DB_DROPPED); } @@ -280,27 +285,27 @@ int32_t ctgAcquireStbMetaFromCache(SCatalog* pCtg, char *dbFName, uint64_t suid, int32_t sz = 0; char* stName = taosHashAcquire(dbCache->stbCache, &suid, sizeof(suid)); if (NULL == stName) { - ctgDebug("stb %" PRIx64 " not in cache, dbFName:%s", suid, dbFName); + ctgDebug("stb 0x%" PRIx64 " not in cache, dbFName:%s", suid, dbFName); goto _return; } pCache = taosHashAcquire(dbCache->tbCache, stName, strlen(stName)); if (NULL == pCache) { - ctgDebug("stb %" PRIx64 " name %s not in cache, dbFName:%s", suid, stName, dbFName); + ctgDebug("stb 0x%" PRIx64 " name %s not in cache, dbFName:%s", suid, stName, dbFName); taosHashRelease(dbCache->stbCache, stName); goto _return; } CTG_LOCK(CTG_READ, &pCache->metaLock); if (NULL == pCache->pMeta) { - ctgDebug("stb %" PRIx64 " meta not in cache, dbFName:%s", suid, dbFName); + ctgDebug("stb 0x%" PRIx64 " meta not in cache, dbFName:%s", suid, dbFName); goto _return; } *pDb = dbCache; *pTb = pCache; - ctgDebug("stb %" PRIx64 " meta got in cache, dbFName:%s", suid, dbFName); + ctgDebug("stb 0x%" PRIx64 " meta got in cache, dbFName:%s", suid, dbFName); CTG_CACHE_STAT_INC(tbMetaHitNum, 1); @@ -434,14 +439,14 @@ int32_t ctgReadTbMetaFromCache(SCatalog* pCtg, SCtgTbMetaCtx* ctx, STableMeta** if (NULL == tbCache) { ctgReleaseTbMetaToCache(pCtg, dbCache, tbCache); taosMemoryFreeClear(*pTableMeta); - ctgDebug("stb %" PRIx64 " meta not in cache", ctx->tbInfo.suid); + ctgDebug("stb 0x%" PRIx64 " meta not in cache", ctx->tbInfo.suid); return TSDB_CODE_SUCCESS; } STableMeta* stbMeta = tbCache->pMeta; if (stbMeta->suid != ctx->tbInfo.suid) { ctgReleaseTbMetaToCache(pCtg, dbCache, tbCache); - ctgError("stb suid %" PRIx64 " in stbCache mis-match, expected suid:%"PRIx64 , stbMeta->suid, ctx->tbInfo.suid); + ctgError("stb suid 0x%" PRIx64 " in stbCache mis-match, expected suid 0x%"PRIx64 , stbMeta->suid, ctx->tbInfo.suid); CTG_ERR_JRET(TSDB_CODE_CTG_INTERNAL_ERROR); } @@ -492,7 +497,7 @@ int32_t ctgReadTbVerFromCache(SCatalog *pCtg, SName *pTableName, int32_t *sver, *sver = tbMeta->sversion; *tver = tbMeta->tversion; - ctgDebug("Got tb %s ver from cache, dbFName:%s, tbType:%d, sver:%d, tver:%d, suid:%" PRIx64, + ctgDebug("Got tb %s ver from cache, dbFName:%s, tbType:%d, sver:%d, tver:%d, suid:0x%" PRIx64, pTableName->tname, dbFName, *tbType, *sver, *tver, *suid); ctgReleaseTbMetaToCache(pCtg, dbCache, tbCache); @@ -507,14 +512,14 @@ int32_t ctgReadTbVerFromCache(SCatalog *pCtg, SName *pTableName, int32_t *sver, ctgAcquireStbMetaFromCache(pCtg, dbFName, *suid, &dbCache, &tbCache); if (NULL == tbCache) { ctgReleaseTbMetaToCache(pCtg, dbCache, tbCache); - ctgDebug("stb %" PRIx64 " meta not in cache", *suid); + ctgDebug("stb 0x%" PRIx64 " meta not in cache", *suid); return TSDB_CODE_SUCCESS; } STableMeta* stbMeta = tbCache->pMeta; if (stbMeta->suid != *suid) { ctgReleaseTbMetaToCache(pCtg, dbCache, tbCache); - ctgError("stb suid %" PRIx64 " in stbCache mis-match, expected suid:%" PRIx64 , stbMeta->suid, *suid); + ctgError("stb suid 0x%" PRIx64 " in stbCache mis-match, expected suid:0x%" PRIx64 , stbMeta->suid, *suid); CTG_ERR_RET(TSDB_CODE_CTG_INTERNAL_ERROR); } @@ -990,6 +995,33 @@ _return: } +int32_t ctgClearCacheEnqueue(SCatalog* pCtg, bool syncOp) { + int32_t code = 0; + SCtgCacheOperation *op = taosMemoryCalloc(1, sizeof(SCtgCacheOperation)); + op->opId = CTG_OP_CLEAR_CACHE; + op->syncOp = syncOp; + + SCtgClearCacheMsg *msg = taosMemoryMalloc(sizeof(SCtgClearCacheMsg)); + if (NULL == msg) { + ctgError("malloc %d failed", (int32_t)sizeof(SCtgClearCacheMsg)); + CTG_ERR_RET(TSDB_CODE_CTG_MEM_ERROR); + } + + msg->pCtg = pCtg; + op->data = msg; + + CTG_ERR_JRET(ctgEnqueue(pCtg, op)); + + return TSDB_CODE_SUCCESS; + +_return: + + taosMemoryFreeClear(msg); + + CTG_RET(code); +} + + int32_t ctgMetaRentInit(SCtgRentMgmt *mgmt, uint32_t rentSec, int8_t type) { mgmt->slotRIdx = 0; mgmt->slotNum = rentSec / CTG_RENT_SLOT_SECOND; @@ -1019,19 +1051,19 @@ int32_t ctgMetaRentAdd(SCtgRentMgmt *mgmt, void *meta, int64_t id, int32_t size) if (NULL == slot->meta) { slot->meta = taosArrayInit(CTG_DEFAULT_RENT_SLOT_SIZE, size); if (NULL == slot->meta) { - qError("taosArrayInit %d failed, id:%"PRIx64", slot idx:%d, type:%d", CTG_DEFAULT_RENT_SLOT_SIZE, id, widx, mgmt->type); + qError("taosArrayInit %d failed, id:0x%"PRIx64", slot idx:%d, type:%d", CTG_DEFAULT_RENT_SLOT_SIZE, id, widx, mgmt->type); CTG_ERR_JRET(TSDB_CODE_CTG_MEM_ERROR); } } if (NULL == taosArrayPush(slot->meta, meta)) { - qError("taosArrayPush meta to rent failed, id:%"PRIx64", slot idx:%d, type:%d", id, widx, mgmt->type); + qError("taosArrayPush meta to rent failed, id:0x%"PRIx64", slot idx:%d, type:%d", id, widx, mgmt->type); CTG_ERR_JRET(TSDB_CODE_CTG_MEM_ERROR); } slot->needSort = true; - qDebug("add meta to rent, id:%"PRIx64", slot idx:%d, type:%d", id, widx, mgmt->type); + qDebug("add meta to rent, id:0x%"PRIx64", slot idx:%d, type:%d", id, widx, mgmt->type); _return: @@ -1047,7 +1079,7 @@ int32_t ctgMetaRentUpdate(SCtgRentMgmt *mgmt, void *meta, int64_t id, int32_t si CTG_LOCK(CTG_WRITE, &slot->lock); if (NULL == slot->meta) { - qError("empty meta slot, id:%"PRIx64", slot idx:%d, type:%d", id, widx, mgmt->type); + qError("empty meta slot, id:0x%"PRIx64", slot idx:%d, type:%d", id, widx, mgmt->type); CTG_ERR_JRET(TSDB_CODE_CTG_INTERNAL_ERROR); } @@ -1060,20 +1092,20 @@ int32_t ctgMetaRentUpdate(SCtgRentMgmt *mgmt, void *meta, int64_t id, int32_t si void *orig = taosArraySearch(slot->meta, &id, searchCompare, TD_EQ); if (NULL == orig) { - qDebug("meta not found in slot, id:%"PRIx64", slot idx:%d, type:%d, size:%d", id, widx, mgmt->type, (int32_t)taosArrayGetSize(slot->meta)); + qDebug("meta not found in slot, id:0x%"PRIx64", slot idx:%d, type:%d, size:%d", id, widx, mgmt->type, (int32_t)taosArrayGetSize(slot->meta)); CTG_ERR_JRET(TSDB_CODE_CTG_INTERNAL_ERROR); } memcpy(orig, meta, size); - qDebug("meta in rent updated, id:%"PRIx64", slot idx:%d, type:%d", id, widx, mgmt->type); + qDebug("meta in rent updated, id:0x%"PRIx64", slot idx:%d, type:%d", id, widx, mgmt->type); _return: CTG_UNLOCK(CTG_WRITE, &slot->lock); if (code) { - qDebug("meta in rent update failed, will try to add it, code:%x, id:%"PRIx64", slot idx:%d, type:%d", code, id, widx, mgmt->type); + qDebug("meta in rent update failed, will try to add it, code:%x, id:0x%"PRIx64", slot idx:%d, type:%d", code, id, widx, mgmt->type); CTG_RET(ctgMetaRentAdd(mgmt, meta, id, size)); } @@ -1088,7 +1120,7 @@ int32_t ctgMetaRentRemove(SCtgRentMgmt *mgmt, int64_t id, __compar_fn_t sortComp CTG_LOCK(CTG_WRITE, &slot->lock); if (NULL == slot->meta) { - qError("empty meta slot, id:%"PRIx64", slot idx:%d, type:%d", id, widx, mgmt->type); + qError("empty meta slot, id:0x%"PRIx64", slot idx:%d, type:%d", id, widx, mgmt->type); CTG_ERR_JRET(TSDB_CODE_CTG_INTERNAL_ERROR); } @@ -1100,13 +1132,13 @@ int32_t ctgMetaRentRemove(SCtgRentMgmt *mgmt, int64_t id, __compar_fn_t sortComp int32_t idx = taosArraySearchIdx(slot->meta, &id, searchCompare, TD_EQ); if (idx < 0) { - qError("meta not found in slot, id:%"PRIx64", slot idx:%d, type:%d", id, widx, mgmt->type); + qError("meta not found in slot, id:0x%"PRIx64", slot idx:%d, type:%d", id, widx, mgmt->type); CTG_ERR_JRET(TSDB_CODE_CTG_INTERNAL_ERROR); } taosArrayRemove(slot->meta, idx); - qDebug("meta in rent removed, id:%"PRIx64", slot idx:%d, type:%d", id, widx, mgmt->type); + qDebug("meta in rent removed, id:0x%"PRIx64", slot idx:%d, type:%d", id, widx, mgmt->type); _return: @@ -1219,11 +1251,11 @@ int32_t ctgAddNewDBCache(SCatalog *pCtg, const char *dbFName, uint64_t dbId) { SDbVgVersion vgVersion = {.dbId = newDBCache.dbId, .vgVersion = -1}; strncpy(vgVersion.dbFName, dbFName, sizeof(vgVersion.dbFName)); - ctgDebug("db added to cache, dbFName:%s, dbId:%"PRIx64, dbFName, dbId); + ctgDebug("db added to cache, dbFName:%s, dbId:0x%"PRIx64, dbFName, dbId); CTG_ERR_RET(ctgMetaRentAdd(&pCtg->dbRent, &vgVersion, dbId, sizeof(SDbVgVersion))); - ctgDebug("db added to rent, dbFName:%s, vgVersion:%d, dbId:%"PRIx64, dbFName, vgVersion.vgVersion, dbId); + ctgDebug("db added to rent, dbFName:%s, vgVersion:%d, dbId:0x%"PRIx64, dbFName, vgVersion.vgVersion, dbId); return TSDB_CODE_SUCCESS; @@ -1246,7 +1278,7 @@ void ctgRemoveStbRent(SCatalog* pCtg, SCtgDBCache *dbCache) { suid = taosHashGetKey(pIter, NULL); if (TSDB_CODE_SUCCESS == ctgMetaRentRemove(&pCtg->stbRent, *suid, ctgStbVersionSortCompare, ctgStbVersionSearchCompare)) { - ctgDebug("stb removed from rent, suid:%"PRIx64, *suid); + ctgDebug("stb removed from rent, suid:0x%"PRIx64, *suid); } pIter = taosHashIterate(dbCache->stbCache, pIter); @@ -1257,7 +1289,7 @@ void ctgRemoveStbRent(SCatalog* pCtg, SCtgDBCache *dbCache) { int32_t ctgRemoveDBFromCache(SCatalog* pCtg, SCtgDBCache *dbCache, const char* dbFName) { uint64_t dbId = dbCache->dbId; - ctgInfo("start to remove db from cache, dbFName:%s, dbId:%"PRIx64, dbFName, dbCache->dbId); + ctgInfo("start to remove db from cache, dbFName:%s, dbId:0x%"PRIx64, dbFName, dbCache->dbId); CTG_LOCK(CTG_WRITE, &dbCache->dbLock); @@ -1268,7 +1300,7 @@ int32_t ctgRemoveDBFromCache(SCatalog* pCtg, SCtgDBCache *dbCache, const char* d CTG_UNLOCK(CTG_WRITE, &dbCache->dbLock); CTG_ERR_RET(ctgMetaRentRemove(&pCtg->dbRent, dbId, ctgDbVgVersionSortCompare, ctgDbVgVersionSearchCompare)); - ctgDebug("db removed from rent, dbFName:%s, dbId:%"PRIx64, dbFName, dbId); + ctgDebug("db removed from rent, dbFName:%s, dbId:0x%"PRIx64, dbFName, dbId); if (taosHashRemove(pCtg->dbCache, dbFName, strlen(dbFName))) { ctgInfo("taosHashRemove from dbCache failed, may be removed, dbFName:%s", dbFName); @@ -1276,7 +1308,7 @@ int32_t ctgRemoveDBFromCache(SCatalog* pCtg, SCtgDBCache *dbCache, const char* d } CTG_CACHE_STAT_DEC(dbNum, 1); - ctgInfo("db removed from cache, dbFName:%s, dbId:%"PRIx64, dbFName, dbId); + ctgInfo("db removed from cache, dbFName:%s, dbId:0x%"PRIx64, dbFName, dbId); return TSDB_CODE_SUCCESS; } @@ -1339,7 +1371,7 @@ int32_t ctgUpdateRentStbVersion(SCatalog *pCtg, char* dbFName, char* tbName, uin CTG_ERR_RET(ctgMetaRentUpdate(&pCtg->stbRent, &metaRent, metaRent.suid, sizeof(SSTableVersion), ctgStbVersionSortCompare, ctgStbVersionSearchCompare)); - ctgDebug("db %s,%" PRIx64 " stb %s,%" PRIx64 " sver %d tver %d smaVer %d updated to stbRent", + ctgDebug("db %s,0x%" PRIx64 " stb %s,0x%" PRIx64 " sver %d tver %d smaVer %d updated to stbRent", dbFName, dbId, tbName, suid, metaRent.sversion, metaRent.tversion, metaRent.smaVer); return TSDB_CODE_SUCCESS; @@ -1349,7 +1381,7 @@ int32_t ctgUpdateRentStbVersion(SCatalog *pCtg, char* dbFName, char* tbName, uin int32_t ctgWriteTbMetaToCache(SCatalog *pCtg, SCtgDBCache *dbCache, char *dbFName, uint64_t dbId, char *tbName, STableMeta *meta, int32_t metaSize) { if (NULL == dbCache->tbCache || NULL == dbCache->stbCache) { taosMemoryFree(meta); - ctgError("db is dropping, dbId:%"PRIx64, dbCache->dbId); + ctgError("db is dropping, dbId:0x%"PRIx64, dbCache->dbId); CTG_ERR_RET(TSDB_CODE_CTG_DB_DROPPED); } @@ -1370,10 +1402,10 @@ int32_t ctgWriteTbMetaToCache(SCatalog *pCtg, SCtgDBCache *dbCache, char *dbFNam if (origType == TSDB_SUPER_TABLE) { if (taosHashRemove(dbCache->stbCache, &orig->suid, sizeof(orig->suid))) { - ctgError("stb not exist in stbCache, dbFName:%s, stb:%s, suid:%"PRIx64, dbFName, tbName, orig->suid); + ctgError("stb not exist in stbCache, dbFName:%s, stb:%s, suid:0x%"PRIx64, dbFName, tbName, orig->suid); } else { CTG_CACHE_STAT_DEC(stblNum, 1); - ctgDebug("stb removed from stbCache, dbFName:%s, stb:%s, suid:%"PRIx64, dbFName, tbName, orig->suid); + ctgDebug("stb removed from stbCache, dbFName:%s, stb:%s, suid:0x%"PRIx64, dbFName, tbName, orig->suid); } origSuid = orig->suid; @@ -1407,13 +1439,13 @@ int32_t ctgWriteTbMetaToCache(SCatalog *pCtg, SCtgDBCache *dbCache, char *dbFNam } if (origSuid != meta->suid && taosHashPut(dbCache->stbCache, &meta->suid, sizeof(meta->suid), tbName, strlen(tbName) + 1) != 0) { - ctgError("taosHashPut to stable cache failed, suid:%"PRIx64, meta->suid); + ctgError("taosHashPut to stable cache failed, suid:0x%"PRIx64, meta->suid); CTG_ERR_RET(TSDB_CODE_CTG_MEM_ERROR); } CTG_CACHE_STAT_INC(stblNum, 1); - ctgDebug("stb %" PRIx64 " updated to cache, dbFName:%s, tbName:%s, tbType:%d", meta->suid, dbFName, tbName, meta->tableType); + ctgDebug("stb 0x%" PRIx64 " updated to cache, dbFName:%s, tbName:%s, tbType:%d", meta->suid, dbFName, tbName, meta->tableType); CTG_ERR_RET(ctgUpdateRentStbVersion(pCtg, dbFName, tbName, dbId, meta->suid, pCache)); @@ -1424,7 +1456,7 @@ int32_t ctgWriteTbIndexToCache(SCatalog *pCtg, SCtgDBCache *dbCache, char* dbFNa if (NULL == dbCache->tbCache) { ctgFreeSTableIndex(*index); taosMemoryFreeClear(*index); - ctgError("db is dropping, dbId:%"PRIx64, dbCache->dbId); + ctgError("db is dropping, dbId:0x%"PRIx64, dbCache->dbId); CTG_ERR_RET(TSDB_CODE_CTG_DB_DROPPED); } @@ -1510,7 +1542,7 @@ int32_t ctgOpUpdateVgroup(SCtgCacheOperation *operation) { SCtgDBCache *dbCache = NULL; CTG_ERR_RET(ctgGetAddDBCache(msg->pCtg, dbFName, msg->dbId, &dbCache)); if (NULL == dbCache) { - ctgInfo("conflict db update, ignore this update, dbFName:%s, dbId:%"PRIx64, dbFName, msg->dbId); + ctgInfo("conflict db update, ignore this update, dbFName:%s, dbId:0x%"PRIx64, dbFName, msg->dbId); CTG_ERR_RET(TSDB_CODE_CTG_INTERNAL_ERROR); } @@ -1540,7 +1572,7 @@ int32_t ctgOpUpdateVgroup(SCtgCacheOperation *operation) { vgCache->vgInfo = dbInfo; msg->dbInfo = NULL; - ctgDebug("db vgInfo updated, dbFName:%s, vgVer:%d, dbId:%"PRIx64, dbFName, vgVersion.vgVersion, vgVersion.dbId); + ctgDebug("db vgInfo updated, dbFName:%s, vgVer:%d, dbId:0x%"PRIx64, dbFName, vgVersion.vgVersion, vgVersion.dbId); ctgWUnlockVgInfo(dbCache); @@ -1569,7 +1601,7 @@ int32_t ctgOpDropDbCache(SCtgCacheOperation *operation) { } if (dbCache->dbId != msg->dbId) { - ctgInfo("dbId already updated, dbFName:%s, dbId:%"PRIx64 ", targetId:%"PRIx64, msg->dbFName, dbCache->dbId, msg->dbId); + ctgInfo("dbId already updated, dbFName:%s, dbId:0x%"PRIx64 ", targetId:0x%"PRIx64, msg->dbFName, dbCache->dbId, msg->dbId); goto _return; } @@ -1629,7 +1661,7 @@ int32_t ctgOpUpdateTbMeta(SCtgCacheOperation *operation) { CTG_ERR_JRET(ctgGetAddDBCache(pCtg, pMeta->dbFName, pMeta->dbId, &dbCache)); if (NULL == dbCache) { - ctgInfo("conflict db update, ignore this update, dbFName:%s, dbId:%" PRIx64, pMeta->dbFName, pMeta->dbId); + ctgInfo("conflict db update, ignore this update, dbFName:%s, dbId:0x%" PRIx64, pMeta->dbFName, pMeta->dbId); CTG_ERR_JRET(TSDB_CODE_CTG_INTERNAL_ERROR); } @@ -1673,27 +1705,28 @@ int32_t ctgOpDropStbMeta(SCtgCacheOperation *operation) { } if (msg->dbId && (dbCache->dbId != msg->dbId)) { - ctgDebug("dbId already modified, dbFName:%s, current:%"PRIx64", dbId:%"PRIx64", stb:%s, suid:%"PRIx64, msg->dbFName, dbCache->dbId, msg->dbId, msg->stbName, msg->suid); + ctgDebug("dbId already modified, dbFName:%s, current:0x%"PRIx64", dbId:0x%"PRIx64", stb:%s, suid:0x%"PRIx64, + msg->dbFName, dbCache->dbId, msg->dbId, msg->stbName, msg->suid); return TSDB_CODE_SUCCESS; } if (taosHashRemove(dbCache->stbCache, &msg->suid, sizeof(msg->suid))) { - ctgDebug("stb not exist in stbCache, may be removed, dbFName:%s, stb:%s, suid:%"PRIx64, msg->dbFName, msg->stbName, msg->suid); + ctgDebug("stb not exist in stbCache, may be removed, dbFName:%s, stb:%s, suid:0x%"PRIx64, msg->dbFName, msg->stbName, msg->suid); } else { CTG_CACHE_STAT_DEC(stblNum, 1); } if (taosHashRemove(dbCache->tbCache, msg->stbName, strlen(msg->stbName))) { - ctgError("stb not exist in cache, dbFName:%s, stb:%s, suid:%"PRIx64, msg->dbFName, msg->stbName, msg->suid); + ctgError("stb not exist in cache, dbFName:%s, stb:%s, suid:0x%"PRIx64, msg->dbFName, msg->stbName, msg->suid); } else { CTG_CACHE_STAT_DEC(tblNum, 1); } - ctgInfo("stb removed from cache, dbFName:%s, stbName:%s, suid:%"PRIx64, msg->dbFName, msg->stbName, msg->suid); + ctgInfo("stb removed from cache, dbFName:%s, stbName:%s, suid:0x%"PRIx64, msg->dbFName, msg->stbName, msg->suid); CTG_ERR_JRET(ctgMetaRentRemove(&msg->pCtg->stbRent, msg->suid, ctgStbVersionSortCompare, ctgStbVersionSearchCompare)); - ctgDebug("stb removed from rent, dbFName:%s, stbName:%s, suid:%"PRIx64, msg->dbFName, msg->stbName, msg->suid); + ctgDebug("stb removed from rent, dbFName:%s, stbName:%s, suid:0x%"PRIx64, msg->dbFName, msg->stbName, msg->suid); _return: @@ -1714,7 +1747,7 @@ int32_t ctgOpDropTbMeta(SCtgCacheOperation *operation) { } if (dbCache->dbId != msg->dbId) { - ctgDebug("dbId %" PRIx64 " not match with curId %"PRIx64", dbFName:%s, tbName:%s", msg->dbId, dbCache->dbId, msg->dbFName, msg->tbName); + ctgDebug("dbId 0x%" PRIx64 " not match with curId 0x%"PRIx64", dbFName:%s, tbName:%s", msg->dbId, dbCache->dbId, msg->dbFName, msg->tbName); return TSDB_CODE_SUCCESS; } @@ -1898,6 +1931,37 @@ _return: } +int32_t ctgOpClearCache(SCtgCacheOperation *operation) { + int32_t code = 0; + SCtgClearCacheMsg *msg = operation->data; + SCatalog* pCtg = msg->pCtg; + + if (pCtg) { + catalogFreeHandle(pCtg); + goto _return; + } + + void* pIter = taosHashIterate(gCtgMgmt.pCluster, NULL); + while (pIter) { + pCtg = *(SCatalog**)pIter; + + if (pCtg) { + catalogFreeHandle(pCtg); + } + + pIter = taosHashIterate(gCtgMgmt.pCluster, pIter); + } + + taosHashClear(gCtgMgmt.pCluster); + +_return: + + taosMemoryFreeClear(msg); + + CTG_RET(code); +} + + void ctgUpdateThreadUnexpectedStopped(void) { if (!atomic_load_8((int8_t*)&gCtgMgmt.exit) && CTG_IS_LOCKED(&gCtgMgmt.lock) > 0) CTG_UNLOCK(CTG_READ, &gCtgMgmt.lock); } @@ -1971,11 +2035,10 @@ void* ctgUpdateThreadFunc(void* param) { CTG_RT_STAT_INC(qDoneNum, 1); + ctgdShowCacheInfo(); ctgdShowClusterCache(pCtg); } - if (CTG_IS_LOCKED(&gCtgMgmt.lock)) CTG_UNLOCK(CTG_READ, &gCtgMgmt.lock); - qInfo("catalog update thread stopped"); return NULL; diff --git a/source/libs/catalog/src/ctgDbg.c b/source/libs/catalog/src/ctgDbg.c index ff93bedb21..7f2b919f17 100644 --- a/source/libs/catalog/src/ctgDbg.c +++ b/source/libs/catalog/src/ctgDbg.c @@ -19,7 +19,7 @@ #include "catalogInt.h" extern SCatalogMgmt gCtgMgmt; -SCtgDebug gCTGDebug = {.lockEnable = true, .apiEnable = true}; +SCtgDebug gCTGDebug = {.cacheEnable = true}; void ctgdUserCallback(SMetaData* pResult, void* param, int32_t code) { ASSERT(*(int32_t*)param == 1); @@ -40,9 +40,9 @@ void ctgdUserCallback(SMetaData* pResult, void* param, int32_t code) { STableComInfo *c = &p->tableInfo; if (TSDB_CHILD_TABLE == p->tableType) { - qDebug("table meta: type:%d, vgId:%d, uid:%" PRIx64 ",suid:%" PRIx64, p->tableType, p->vgId, p->uid, p->suid); + qDebug("table meta: type:%d, vgId:%d, uid:0x%" PRIx64 ",suid:0x%" PRIx64, p->tableType, p->vgId, p->uid, p->suid); } else { - qDebug("table meta: type:%d, vgId:%d, uid:%" PRIx64 ",suid:%" PRIx64 ",sv:%d, tv:%d, tagNum:%d, precision:%d, colNum:%d, rowSize:%d", + qDebug("table meta: type:%d, vgId:%d, uid:0x%" PRIx64 ",suid:0x%" PRIx64 ",sv:%d, tv:%d, tagNum:%d, precision:%d, colNum:%d, rowSize:%d", p->tableType, p->vgId, p->uid, p->suid, p->sversion, p->tversion, c->numOfTags, c->precision, c->numOfColumns, c->rowSize); } @@ -75,7 +75,7 @@ void ctgdUserCallback(SMetaData* pResult, void* param, int32_t code) { num = taosArrayGetSize(pResult->pDbInfo); for (int32_t i = 0; i < num; ++i) { SDbInfo *pDb = taosArrayGet(pResult->pDbInfo, i); - qDebug("db %d dbInfo: vgVer:%d, tbNum:%d, dbId:%" PRIx64, i, pDb->vgVer, pDb->tbNum, pDb->dbId); + qDebug("db %d dbInfo: vgVer:%d, tbNum:%d, dbId:0x%" PRIx64, i, pDb->vgVer, pDb->tbNum, pDb->dbId); } } else { qDebug("empty db info"); @@ -333,10 +333,10 @@ void ctgdShowTableMeta(SCatalog* pCtg, const char *tbName, STableMeta* p) { STableComInfo *c = &p->tableInfo; if (TSDB_CHILD_TABLE == p->tableType) { - ctgDebug("table [%s] meta: type:%d, vgId:%d, uid:%" PRIx64 ",suid:%" PRIx64, tbName, p->tableType, p->vgId, p->uid, p->suid); + ctgDebug("table [%s] meta: type:%d, vgId:%d, uid:0x%" PRIx64 ",suid:0x%" PRIx64, tbName, p->tableType, p->vgId, p->uid, p->suid); return; } else { - ctgDebug("table [%s] meta: type:%d, vgId:%d, uid:%" PRIx64 ",suid:%" PRIx64 ",sv:%d, tv:%d, tagNum:%d, precision:%d, colNum:%d, rowSize:%d", + ctgDebug("table [%s] meta: type:%d, vgId:%d, uid:0x%" PRIx64 ",suid:0x%" PRIx64 ",sv:%d, tv:%d, tagNum:%d, precision:%d, colNum:%d, rowSize:%d", tbName, p->tableType, p->vgId, p->uid, p->suid, p->sversion, p->tversion, c->numOfTags, c->precision, c->numOfColumns, c->rowSize); } @@ -377,7 +377,7 @@ void ctgdShowDBCache(SCatalog* pCtg, SHashObj *dbHash) { } } - ctgDebug("[%d] db [%.*s][%"PRIx64"] %s: metaNum:%d, stbNum:%d, vgVersion:%d, hashMethod:%d, vgNum:%d", + ctgDebug("[%d] db [%.*s][0x%"PRIx64"] %s: metaNum:%d, stbNum:%d, vgVersion:%d, hashMethod:%d, vgNum:%d", i, (int32_t)len, dbFName, dbCache->dbId, dbCache->deleted?"deleted":"", metaNum, stbNum, vgVersion, hashMethod, vgNum); pIter = taosHashIterate(dbHash, pIter); @@ -392,13 +392,13 @@ void ctgdShowClusterCache(SCatalog* pCtg) { return; } - ctgDebug("## cluster %"PRIx64" %p cache Info BEGIN ##", pCtg->clusterId, pCtg); + ctgDebug("## cluster 0x%"PRIx64" %p cache Info BEGIN ##", pCtg->clusterId, pCtg); ctgDebug("db:%d meta:%d stb:%d dbRent:%d stbRent:%d", ctgdGetClusterCacheNum(pCtg, CTG_DBG_DB_NUM), ctgdGetClusterCacheNum(pCtg, CTG_DBG_META_NUM), ctgdGetClusterCacheNum(pCtg, CTG_DBG_STB_NUM), ctgdGetClusterCacheNum(pCtg, CTG_DBG_DB_RENT_NUM), ctgdGetClusterCacheNum(pCtg, CTG_DBG_STB_RENT_NUM)); ctgdShowDBCache(pCtg, pCtg->dbCache); - ctgDebug("## cluster %"PRIx64" %p cache Info END ##", pCtg->clusterId, pCtg); + ctgDebug("## cluster 0x%"PRIx64" %p cache Info END ##", pCtg->clusterId, pCtg); } int32_t ctgdShowCacheInfo(void) { @@ -407,6 +407,8 @@ int32_t ctgdShowCacheInfo(void) { } CTG_API_ENTER(); + + qDebug("# total catalog cluster number %d #", taosHashGetSize(gCtgMgmt.pCluster)); SCatalog *pCtg = NULL; void *pIter = taosHashIterate(gCtgMgmt.pCluster, NULL); diff --git a/source/libs/catalog/src/ctgRemote.c b/source/libs/catalog/src/ctgRemote.c index 178025704f..fa1a262832 100644 --- a/source/libs/catalog/src/ctgRemote.c +++ b/source/libs/catalog/src/ctgRemote.c @@ -186,13 +186,13 @@ int32_t ctgHandleMsgCallback(void *param, const SDataBuf *pMsg, int32_t rspCode) SCtgJob* pJob = taosAcquireRef(gCtgMgmt.jobPool, cbParam->refId); if (NULL == pJob) { - qDebug("job refId %" PRIx64 " already dropped", cbParam->refId); + qDebug("ctg job refId 0x%" PRIx64 " already dropped", cbParam->refId); goto _return; } SCtgTask *pTask = taosArrayGet(pJob->pTasks, cbParam->taskId); - qDebug("QID:0x%" PRIx64 " task %d start to handle rsp %s", pJob->queryId, pTask->taskId, TMSG_INFO(cbParam->reqType + 1)); + qDebug("QID:0x%" PRIx64 " ctg task %d start to handle rsp %s", pJob->queryId, pTask->taskId, TMSG_INFO(cbParam->reqType + 1)); CTG_ERR_JRET((*gCtgAsyncFps[pTask->type].handleRspFp)(pTask, cbParam->reqType, pMsg, rspCode)); @@ -263,7 +263,7 @@ int32_t ctgAsyncSendMsg(SCatalog* pCtg, SRequestConnInfo *pConn, SCtgTask* pTask CTG_ERR_JRET(code); } - ctgDebug("req msg sent, reqId:0x%" PRIx64 ", msg type:%d, %s", pTask->pJob->queryId, msgType, TMSG_INFO(msgType)); + ctgDebug("ctg req msg sent, reqId:0x%" PRIx64 ", msg type:%d, %s", pTask->pJob->queryId, msgType, TMSG_INFO(msgType)); return TSDB_CODE_SUCCESS; _return: diff --git a/source/libs/catalog/src/ctgUtil.c b/source/libs/catalog/src/ctgUtil.c index e97c34dc26..476eb371b0 100644 --- a/source/libs/catalog/src/ctgUtil.c +++ b/source/libs/catalog/src/ctgUtil.c @@ -434,7 +434,7 @@ void ctgFreeJob(void* job) { taosMemoryFree(job); - qDebug("QID:%" PRIx64 ", job %" PRIx64 " freed", qid, rid); + qDebug("QID:0x%" PRIx64 ", ctg job 0x%" PRIx64 " freed", qid, rid); } int32_t ctgUpdateMsgCtx(SCtgMsgCtx* pCtx, int32_t reqType, void* out, char* target) { diff --git a/source/libs/command/src/command.c b/source/libs/command/src/command.c index f330b7ce16..f06664b60b 100644 --- a/source/libs/command/src/command.c +++ b/source/libs/command/src/command.c @@ -14,6 +14,7 @@ */ #include "command.h" +#include "catalog.h" #include "tdatablock.h" static int32_t getSchemaBytes(const SSchema* pSchema) { @@ -120,8 +121,7 @@ static int32_t execDescribe(SNode* pStmt, SRetrieveTableRsp** pRsp) { } static int32_t execResetQueryCache() { - // todo - return TSDB_CODE_SUCCESS; + return catalogClearCache(); } int32_t qExecCommand(SNode* pStmt, SRetrieveTableRsp** pRsp) { diff --git a/source/libs/command/src/explain.c b/source/libs/command/src/explain.c index 79861dfa05..8f91282480 100644 --- a/source/libs/command/src/explain.c +++ b/source/libs/command/src/explain.c @@ -194,6 +194,11 @@ int32_t qExplainGenerateResChildren(SPhysiNode *pNode, SExplainGroup *group, SNo pPhysiChildren = fillPhysiNode->node.pChildren; break; } + case QUERY_NODE_PHYSICAL_PLAN_TABLE_MERGE_SCAN: { + STableMergeScanPhysiNode *mergePhysiNode = (STableMergeScanPhysiNode *)pNode; + pPhysiChildren = mergePhysiNode->scan.node.pChildren; + break; + } default: qError("not supported physical node type %d", pNode->type); QRY_ERR_RET(TSDB_CODE_QRY_APP_ERROR); @@ -398,6 +403,7 @@ int32_t qExplainResNodeToRowsImpl(SExplainResNode *pResNode, SExplainCtx *ctx, i break; } case QUERY_NODE_PHYSICAL_PLAN_TABLE_SEQ_SCAN: + case QUERY_NODE_PHYSICAL_PLAN_TABLE_MERGE_SCAN: case QUERY_NODE_PHYSICAL_PLAN_TABLE_SCAN: { STableScanPhysiNode *pTblScanNode = (STableScanPhysiNode *)pNode; EXPLAIN_ROW_NEW(level, EXPLAIN_TBL_SCAN_FORMAT, pTblScanNode->scan.tableName.tname); diff --git a/source/libs/executor/inc/executorimpl.h b/source/libs/executor/inc/executorimpl.h index 034e2893df..d9a7dc8718 100644 --- a/source/libs/executor/inc/executorimpl.h +++ b/source/libs/executor/inc/executorimpl.h @@ -845,8 +845,10 @@ SOperatorInfo* createTimeSliceOperatorInfo(SOperatorInfo* downstream, SExprInfo* SOperatorInfo* createMergeJoinOperatorInfo(SOperatorInfo** pDownstream, int32_t numOfDownstream, SExprInfo* pExprInfo, int32_t numOfCols, SSDataBlock* pResBlock, SNode* pOnCondition, SExecTaskInfo* pTaskInfo); -SOperatorInfo* createStreamSessionAggOperatorInfo(SOperatorInfo* downstream, SExprInfo* pExprInfo, int32_t numOfCols, - SSDataBlock* pResBlock, int64_t gap, int32_t tsSlotId, STimeWindowAggSupp* pTwAggSupp, SExecTaskInfo* pTaskInfo); +SOperatorInfo* createStreamSessionAggOperatorInfo(SOperatorInfo* downstream, + SPhysiNode* pPhyNode, SExecTaskInfo* pTaskInfo); +SOperatorInfo* createStreamFinalSessionAggOperatorInfo(SOperatorInfo* downstream, + SPhysiNode* pPhyNode, SExecTaskInfo* pTaskInfo, int32_t numOfChild); SOperatorInfo* createStreamStateAggOperatorInfo(SOperatorInfo* downstream, SPhysiNode* pPhyNode, SExecTaskInfo* pTaskInfo); diff --git a/source/libs/executor/src/executorimpl.c b/source/libs/executor/src/executorimpl.c index 40b019eb5d..b8516a3056 100644 --- a/source/libs/executor/src/executorimpl.c +++ b/source/libs/executor/src/executorimpl.c @@ -4728,18 +4728,13 @@ SOperatorInfo* createOperatorTree(SPhysiNode* pPhyNode, SExecTaskInfo* pTaskInfo pOptr = createSessionAggOperatorInfo(ops[0], pExprInfo, num, pResBlock, pSessionNode->gap, tsSlotId, &as, pTaskInfo); } else if (QUERY_NODE_PHYSICAL_PLAN_STREAM_SESSION == type) { - SSessionWinodwPhysiNode* pSessionNode = (SSessionWinodwPhysiNode*)pPhyNode; - - STimeWindowAggSupp as = {.waterMark = pSessionNode->window.watermark, - .calTrigger = pSessionNode->window.triggerType}; - - SExprInfo* pExprInfo = createExprInfo(pSessionNode->window.pFuncs, NULL, &num); - SSDataBlock* pResBlock = createResDataBlock(pPhyNode->pOutputDataBlockDesc); - int32_t tsSlotId = ((SColumnNode*)pSessionNode->window.pTspk)->slotId; - - pOptr = createStreamSessionAggOperatorInfo(ops[0], pExprInfo, num, pResBlock, pSessionNode->gap, tsSlotId, &as, - pTaskInfo); - + pOptr = createStreamSessionAggOperatorInfo(ops[0], pPhyNode, pTaskInfo); + } else if (QUERY_NODE_PHYSICAL_PLAN_STREAM_SEMI_SESSION == type) { + int32_t children = 0; + pOptr = createStreamFinalSessionAggOperatorInfo(ops[0], pPhyNode, pTaskInfo, children); + } else if (QUERY_NODE_PHYSICAL_PLAN_STREAM_FINAL_SESSION == type) { + int32_t children = 1; + pOptr = createStreamFinalSessionAggOperatorInfo(ops[0], pPhyNode, pTaskInfo, children); } else if (QUERY_NODE_PHYSICAL_PLAN_PARTITION == type) { pOptr = createPartitionOperatorInfo(ops[0], (SPartitionPhysiNode*)pPhyNode, pTaskInfo); } else if (QUERY_NODE_PHYSICAL_PLAN_MERGE_STATE == type) { diff --git a/source/libs/executor/src/scanoperator.c b/source/libs/executor/src/scanoperator.c index 80276007de..2871145dcc 100644 --- a/source/libs/executor/src/scanoperator.c +++ b/source/libs/executor/src/scanoperator.c @@ -537,6 +537,8 @@ SOperatorInfo* createTableScanOperatorInfo(STableScanPhysiNode* pTableScanNode, goto _error; } + //taosSsleep(20); + SDataBlockDescNode* pDescNode = pTableScanNode->scan.node.pOutputDataBlockDesc; int32_t numOfCols = 0; diff --git a/source/libs/executor/src/sortoperator.c b/source/libs/executor/src/sortoperator.c index 81899b68cd..77eeb24210 100644 --- a/source/libs/executor/src/sortoperator.c +++ b/source/libs/executor/src/sortoperator.c @@ -392,10 +392,8 @@ SOperatorInfo* createMultiwaySortMergeOperatorInfo(SOperatorInfo** downStreams, pInfo->bufPageSize = getProperSortPageSize(rowSize); - uint32_t numOfSources = taosArrayGetSize(pSortInfo); - numOfSources = TMAX(4, numOfSources); - - pInfo->sortBufSize = numOfSources * pInfo->bufPageSize; + // one additional is reserved for merged result. + pInfo->sortBufSize = pInfo->bufPageSize * (numStreams + 1); pOperator->fpSet = createOperatorFpSet(doOpenMultiwaySortMergeOperator, doMultiwaySortMerge, NULL, NULL, diff --git a/source/libs/executor/src/timewindowoperator.c b/source/libs/executor/src/timewindowoperator.c index d7ae823522..d827067f1c 100644 --- a/source/libs/executor/src/timewindowoperator.c +++ b/source/libs/executor/src/timewindowoperator.c @@ -2425,12 +2425,15 @@ int32_t initSessionAggSupporter(SStreamAggSupporter* pSup, const char* pKey, Sql return initStreamAggSupporter(pSup, pKey, pCtx, numOfOutput, sizeof(SResultWindowInfo)); } -SOperatorInfo* createStreamSessionAggOperatorInfo(SOperatorInfo* downstream, SExprInfo* pExprInfo, int32_t numOfCols, - SSDataBlock* pResBlock, int64_t gap, int32_t tsSlotId, - STimeWindowAggSupp* pTwAggSupp, SExecTaskInfo* pTaskInfo) { - int32_t code = TSDB_CODE_OUT_OF_MEMORY; +SOperatorInfo* createStreamSessionAggOperatorInfo(SOperatorInfo* downstream, SPhysiNode* pPhyNode, SExecTaskInfo* pTaskInfo) { + SSessionWinodwPhysiNode* pSessionNode = (SSessionWinodwPhysiNode*)pPhyNode; + int32_t numOfCols = 0; + SExprInfo* pExprInfo = createExprInfo(pSessionNode->window.pFuncs, NULL, &numOfCols); + SSDataBlock* pResBlock = createResDataBlock(pPhyNode->pOutputDataBlockDesc); + int32_t tsSlotId = ((SColumnNode*)pSessionNode->window.pTspk)->slotId; + int32_t code = TSDB_CODE_OUT_OF_MEMORY; SStreamSessionAggOperatorInfo* pInfo = taosMemoryCalloc(1, sizeof(SStreamSessionAggOperatorInfo)); - SOperatorInfo* pOperator = taosMemoryCalloc(1, sizeof(SOperatorInfo)); + SOperatorInfo* pOperator = taosMemoryCalloc(1, sizeof(SOperatorInfo)); if (pInfo == NULL || pOperator == NULL) { goto _error; } @@ -2453,12 +2456,14 @@ SOperatorInfo* createStreamSessionAggOperatorInfo(SOperatorInfo* downstream, SEx } initDummyFunction(pInfo->pDummyCtx, pInfo->binfo.pCtx, numOfCols); - pInfo->twAggSup = *pTwAggSupp; + pInfo->twAggSup = (STimeWindowAggSupp) {.waterMark = pSessionNode->window.watermark, + .calTrigger = pSessionNode->window.triggerType, + .maxTs = INT64_MIN}; initResultRowInfo(&pInfo->binfo.resultRowInfo, 8); initExecTimeWindowInfo(&pInfo->twAggSup.timeWindowData, &pTaskInfo->window); pInfo->primaryTsIndex = tsSlotId; - pInfo->gap = gap; + pInfo->gap = pSessionNode->gap; pInfo->binfo.pRes = pResBlock; pInfo->order = TSDB_ORDER_ASC; _hash_fn_t hashFn = taosGetDefaultHashFunction(TSDB_DATA_TYPE_BINARY); @@ -2960,25 +2965,20 @@ static SSDataBlock* doStreamSessionAgg(SOperatorInfo* pOperator) { return pBInfo->pRes->info.rows == 0 ? NULL : pBInfo->pRes; } -SOperatorInfo* createStreamFinalSessionAggOperatorInfo(SOperatorInfo* downstream, SExprInfo* pExprInfo, - int32_t numOfCols, SSDataBlock* pResBlock, int64_t gap, - int32_t tsSlotId, STimeWindowAggSupp* pTwAggSupp, - SExecTaskInfo* pTaskInfo) { - int32_t code = TSDB_CODE_OUT_OF_MEMORY; - SStreamSessionAggOperatorInfo* pInfo = NULL; - SOperatorInfo* pOperator = createStreamSessionAggOperatorInfo(downstream, pExprInfo, numOfCols, pResBlock, gap, - tsSlotId, pTwAggSupp, pTaskInfo); +SOperatorInfo* createStreamFinalSessionAggOperatorInfo(SOperatorInfo* downstream, + SPhysiNode* pPhyNode, SExecTaskInfo* pTaskInfo, int32_t numOfChild) { + int32_t code = TSDB_CODE_OUT_OF_MEMORY; + SOperatorInfo* pOperator = createStreamSessionAggOperatorInfo(downstream, pPhyNode, pTaskInfo); if (pOperator == NULL) { goto _error; } pOperator->name = "StreamFinalSessionWindowAggOperator"; pOperator->operatorType = QUERY_NODE_PHYSICAL_PLAN_STREAM_FINAL_SESSION; - int32_t numOfChild = 1; // Todo(liuyao) get it from phy plan - pInfo = pOperator->info; + SStreamSessionAggOperatorInfo* pInfo = pOperator->info; pInfo->pChildren = taosArrayInit(8, sizeof(void*)); for (int32_t i = 0; i < numOfChild; i++) { SOperatorInfo* pChild = - createStreamSessionAggOperatorInfo(NULL, pExprInfo, numOfCols, NULL, gap, tsSlotId, pTwAggSupp, pTaskInfo); + createStreamSessionAggOperatorInfo(NULL, pPhyNode, pTaskInfo); if (pChild == NULL) { goto _error; } @@ -2988,7 +2988,7 @@ SOperatorInfo* createStreamFinalSessionAggOperatorInfo(SOperatorInfo* downstream _error: if (pInfo != NULL) { - destroyStreamSessionAggOperatorInfo(pInfo, numOfCols); + destroyStreamSessionAggOperatorInfo(pInfo, pOperator->numOfExprs); } taosMemoryFreeClear(pInfo); diff --git a/source/libs/executor/src/tsort.c b/source/libs/executor/src/tsort.c index 1502387360..00e07c7199 100644 --- a/source/libs/executor/src/tsort.c +++ b/source/libs/executor/src/tsort.c @@ -195,6 +195,11 @@ static int32_t doAddToBuf(SSDataBlock* pDataBlock, SSortHandle* pHandle) { return doAddNewExternalMemSource(pHandle->pBuf, pHandle->pOrderedSource, pBlock, &pHandle->sourceId); } +static void setCurrentSourceIsDone(SSortSource* pSource, SSortHandle* pHandle) { + pSource->src.rowIndex = -1; + ++pHandle->numOfCompletedSources; +} + static int32_t sortComparInit(SMsortComparParam* cmpParam, SArray* pSources, int32_t startIndex, int32_t endIndex, SSortHandle* pHandle) { cmpParam->pSources = taosArrayGet(pSources, startIndex); cmpParam->numOfSources = (endIndex - startIndex + 1); @@ -205,8 +210,10 @@ static int32_t sortComparInit(SMsortComparParam* cmpParam, SArray* pSources, int for (int32_t i = 0; i < cmpParam->numOfSources; ++i) { SSortSource* pSource = cmpParam->pSources[i]; + // set current source is done if (taosArrayGetSize(pSource->pageIdList) == 0) { - return TSDB_CODE_SUCCESS; + setCurrentSourceIsDone(pSource, pHandle); + continue; } SPageInfo* pPgInfo = *(SPageInfo**)taosArrayGet(pSource->pageIdList, pSource->pageIndex); @@ -233,10 +240,9 @@ static int32_t sortComparInit(SMsortComparParam* cmpParam, SArray* pSources, int SSortSource* pSource = cmpParam->pSources[i]; pSource->src.pBlock = pHandle->fetchfp(pSource->param); - // set current source id done + // set current source is done if (pSource->src.pBlock == NULL) { - pSource->src.rowIndex = -1; - ++pHandle->numOfCompletedSources; + setCurrentSourceIsDone(pSource, pHandle); } } } diff --git a/source/libs/nodes/src/nodesCloneFuncs.c b/source/libs/nodes/src/nodesCloneFuncs.c index 2e30b01357..d0eb1548c1 100644 --- a/source/libs/nodes/src/nodesCloneFuncs.c +++ b/source/libs/nodes/src/nodesCloneFuncs.c @@ -427,7 +427,7 @@ static SNode* logicWindowCopy(const SWindowLogicNode* pSrc, SWindowLogicNode* pD COPY_SCALAR_FIELD(triggerType); COPY_SCALAR_FIELD(watermark); COPY_SCALAR_FIELD(filesFactor); - COPY_SCALAR_FIELD(intervalAlgo); + COPY_SCALAR_FIELD(windowAlgo); return (SNode*)pDst; } @@ -538,6 +538,12 @@ static SNode* physiIntervalCopy(const SIntervalPhysiNode* pSrc, SIntervalPhysiNo return (SNode*)pDst; } +static SNode* physiSessionCopy(const SSessionWinodwPhysiNode* pSrc, SSessionWinodwPhysiNode* pDst) { + COPY_BASE_OBJECT_FIELD(window, physiWindowCopy); + COPY_SCALAR_FIELD(gap); + return (SNode*)pDst; +} + static SNode* dataBlockDescCopy(const SDataBlockDescNode* pSrc, SDataBlockDescNode* pDst) { COPY_SCALAR_FIELD(dataBlockId); CLONE_NODE_LIST_FIELD(pSlots); @@ -678,6 +684,9 @@ SNode* nodesCloneNode(const SNode* pNode) { case QUERY_NODE_PHYSICAL_PLAN_STREAM_FINAL_INTERVAL: case QUERY_NODE_PHYSICAL_PLAN_STREAM_SEMI_INTERVAL: return physiIntervalCopy((const SIntervalPhysiNode*)pNode, (SIntervalPhysiNode*)pDst); + case QUERY_NODE_PHYSICAL_PLAN_STREAM_SEMI_SESSION: + case QUERY_NODE_PHYSICAL_PLAN_STREAM_FINAL_SESSION: + return physiSessionCopy((const SSessionWinodwPhysiNode*)pNode, (SSessionWinodwPhysiNode*)pDst); default: break; } diff --git a/source/libs/nodes/src/nodesCodeFuncs.c b/source/libs/nodes/src/nodesCodeFuncs.c index fb6a428d3c..1b2eadd8e8 100644 --- a/source/libs/nodes/src/nodesCodeFuncs.c +++ b/source/libs/nodes/src/nodesCodeFuncs.c @@ -172,8 +172,6 @@ const char* nodesNodeName(ENodeType type) { return "ShowSubscribesStmt"; case QUERY_NODE_SHOW_SMAS_STMT: return "ShowSmasStmt"; - case QUERY_NODE_SHOW_CONFIGS_STMT: - return "ShowConfigsStmt"; case QUERY_NODE_SHOW_QUERIES_STMT: return "ShowQueriesStmt"; case QUERY_NODE_SHOW_VNODES_STMT: @@ -246,6 +244,10 @@ const char* nodesNodeName(ENodeType type) { return "PhysiSessionWindow"; case QUERY_NODE_PHYSICAL_PLAN_STREAM_SESSION: return "PhysiStreamSessionWindow"; + case QUERY_NODE_PHYSICAL_PLAN_STREAM_SEMI_SESSION: + return "PhysiStreamSemiSessionWindow"; + case QUERY_NODE_PHYSICAL_PLAN_STREAM_FINAL_SESSION: + return "PhysiStreamFinalSessionWindow"; case QUERY_NODE_PHYSICAL_PLAN_MERGE_STATE: return "PhysiStateWindow"; case QUERY_NODE_PHYSICAL_PLAN_STREAM_STATE: @@ -3998,6 +4000,8 @@ static int32_t specificNodeToJson(const void* pObj, SJson* pJson) { return physiFillNodeToJson(pObj, pJson); case QUERY_NODE_PHYSICAL_PLAN_MERGE_SESSION: case QUERY_NODE_PHYSICAL_PLAN_STREAM_SESSION: + case QUERY_NODE_PHYSICAL_PLAN_STREAM_SEMI_SESSION: + case QUERY_NODE_PHYSICAL_PLAN_STREAM_FINAL_SESSION: return physiSessionWindowNodeToJson(pObj, pJson); case QUERY_NODE_PHYSICAL_PLAN_MERGE_STATE: case QUERY_NODE_PHYSICAL_PLAN_STREAM_STATE: @@ -4131,6 +4135,8 @@ static int32_t jsonToSpecificNode(const SJson* pJson, void* pObj) { return jsonToPhysiFillNode(pJson, pObj); case QUERY_NODE_PHYSICAL_PLAN_MERGE_SESSION: case QUERY_NODE_PHYSICAL_PLAN_STREAM_SESSION: + case QUERY_NODE_PHYSICAL_PLAN_STREAM_SEMI_SESSION: + case QUERY_NODE_PHYSICAL_PLAN_STREAM_FINAL_SESSION: return jsonToPhysiSessionWindowNode(pJson, pObj); case QUERY_NODE_PHYSICAL_PLAN_MERGE_STATE: case QUERY_NODE_PHYSICAL_PLAN_STREAM_STATE: diff --git a/source/libs/nodes/src/nodesUtilFuncs.c b/source/libs/nodes/src/nodesUtilFuncs.c index 6d04e7a959..f8b44793a6 100644 --- a/source/libs/nodes/src/nodesUtilFuncs.c +++ b/source/libs/nodes/src/nodesUtilFuncs.c @@ -200,7 +200,6 @@ SNode* nodesMakeNode(ENodeType type) { case QUERY_NODE_SHOW_CONSUMERS_STMT: case QUERY_NODE_SHOW_SUBSCRIBES_STMT: case QUERY_NODE_SHOW_SMAS_STMT: - case QUERY_NODE_SHOW_CONFIGS_STMT: case QUERY_NODE_SHOW_CONNECTIONS_STMT: case QUERY_NODE_SHOW_QUERIES_STMT: case QUERY_NODE_SHOW_VNODES_STMT: @@ -289,6 +288,10 @@ SNode* nodesMakeNode(ENodeType type) { return makeNode(type, sizeof(SSessionWinodwPhysiNode)); case QUERY_NODE_PHYSICAL_PLAN_STREAM_SESSION: return makeNode(type, sizeof(SStreamSessionWinodwPhysiNode)); + case QUERY_NODE_PHYSICAL_PLAN_STREAM_SEMI_SESSION: + return makeNode(type, sizeof(SStreamSemiSessionWinodwPhysiNode)); + case QUERY_NODE_PHYSICAL_PLAN_STREAM_FINAL_SESSION: + return makeNode(type, sizeof(SStreamFinalSessionWinodwPhysiNode)); case QUERY_NODE_PHYSICAL_PLAN_MERGE_STATE: return makeNode(type, sizeof(SStateWinodwPhysiNode)); case QUERY_NODE_PHYSICAL_PLAN_STREAM_STATE: @@ -615,7 +618,6 @@ void nodesDestroyNode(SNode* pNode) { case QUERY_NODE_SHOW_CONSUMERS_STMT: case QUERY_NODE_SHOW_SUBSCRIBES_STMT: case QUERY_NODE_SHOW_SMAS_STMT: - case QUERY_NODE_SHOW_CONFIGS_STMT: case QUERY_NODE_SHOW_CONNECTIONS_STMT: case QUERY_NODE_SHOW_QUERIES_STMT: case QUERY_NODE_SHOW_VNODES_STMT: @@ -806,6 +808,7 @@ void nodesDestroyNode(SNode* pNode) { } case QUERY_NODE_PHYSICAL_PLAN_MERGE_SESSION: case QUERY_NODE_PHYSICAL_PLAN_STREAM_SESSION: + case QUERY_NODE_PHYSICAL_PLAN_STREAM_SEMI_SESSION: case QUERY_NODE_PHYSICAL_PLAN_STREAM_FINAL_SESSION: destroyWinodwPhysiNode((SWinodwPhysiNode*)pNode); break; diff --git a/source/libs/parser/src/parAstCreater.c b/source/libs/parser/src/parAstCreater.c index 4a7cdd628f..4f696e6522 100644 --- a/source/libs/parser/src/parAstCreater.c +++ b/source/libs/parser/src/parAstCreater.c @@ -1455,7 +1455,7 @@ SNode* createKillQueryStmt(SAstCreateContext* pCxt, const SToken* pQueryId) { CHECK_PARSER_STATUS(pCxt); SKillQueryStmt* pStmt = (SKillQueryStmt*)nodesMakeNode(QUERY_NODE_KILL_QUERY_STMT); CHECK_OUT_OF_MEM(pStmt); - strncpy(pStmt->queryId, pQueryId->z, TMIN(pQueryId->n, sizeof(pStmt->queryId) - 1)); + trimString(pQueryId->z, pQueryId->n, pStmt->queryId, sizeof(pStmt->queryId) - 1); return (SNode*)pStmt; } diff --git a/source/libs/parser/src/parAstParser.c b/source/libs/parser/src/parAstParser.c index 1aa1320a3c..1e968fe1fa 100644 --- a/source/libs/parser/src/parAstParser.c +++ b/source/libs/parser/src/parAstParser.c @@ -387,6 +387,21 @@ static int32_t collectMetaKeyFromShowQueries(SCollectMetaKeyCxt* pCxt, SShowStmt pCxt->pMetaCache); } +static int32_t collectMetaKeyFromShowConfigs(SCollectMetaKeyCxt* pCxt, SShowStmt* pStmt) { + return reserveTableMetaInCache(pCxt->pParseCxt->acctId, TSDB_INFORMATION_SCHEMA_DB, TSDB_INS_TABLE_CONFIGS, + pCxt->pMetaCache); +} + +static int32_t collectMetaKeyFromShowVariables(SCollectMetaKeyCxt* pCxt, SShowStmt* pStmt) { + return reserveTableMetaInCache(pCxt->pParseCxt->acctId, TSDB_INFORMATION_SCHEMA_DB, TSDB_INS_TABLE_CONFIGS, + pCxt->pMetaCache); +} + +static int32_t collectMetaKeyFromShowApps(SCollectMetaKeyCxt* pCxt, SShowStmt* pStmt) { + return reserveTableMetaInCache(pCxt->pParseCxt->acctId, TSDB_PERFORMANCE_SCHEMA_DB, TSDB_PERFS_TABLE_APPS, + pCxt->pMetaCache); +} + static int32_t collectMetaKeyFromShowTransactions(SCollectMetaKeyCxt* pCxt, SShowStmt* pStmt) { return reserveTableMetaInCache(pCxt->pParseCxt->acctId, TSDB_PERFORMANCE_SCHEMA_DB, TSDB_PERFS_TABLE_TRANS, pCxt->pMetaCache); @@ -461,6 +476,10 @@ static int32_t collectMetaKeyFromQuery(SCollectMetaKeyCxt* pCxt, SNode* pStmt) { return collectMetaKeyFromShowConnections(pCxt, (SShowStmt*)pStmt); case QUERY_NODE_SHOW_QUERIES_STMT: return collectMetaKeyFromShowQueries(pCxt, (SShowStmt*)pStmt); + case QUERY_NODE_SHOW_VARIABLE_STMT: + return collectMetaKeyFromShowVariables(pCxt, (SShowStmt*)pStmt); + case QUERY_NODE_SHOW_APPS_STMT: + return collectMetaKeyFromShowApps(pCxt, (SShowStmt*)pStmt); case QUERY_NODE_SHOW_TRANSACTIONS_STMT: return collectMetaKeyFromShowTransactions(pCxt, (SShowStmt*)pStmt); case QUERY_NODE_DELETE_STMT: diff --git a/source/libs/parser/src/parTranslater.c b/source/libs/parser/src/parTranslater.c index f4ffc34d8e..6f6c6dd7e8 100644 --- a/source/libs/parser/src/parTranslater.c +++ b/source/libs/parser/src/parTranslater.c @@ -3931,7 +3931,7 @@ static int32_t translateDescribe(STranslateContext* pCxt, SDescribeStmt* pStmt) static int32_t translateKillConnection(STranslateContext* pCxt, SKillStmt* pStmt) { SKillConnReq killReq = {0}; killReq.connId = pStmt->targetId; - return buildCmdMsg(pCxt, TDMT_MND_KILL_CONN, (FSerializeFunc)tSerializeSKillQueryReq, &killReq); + return buildCmdMsg(pCxt, TDMT_MND_KILL_CONN, (FSerializeFunc)tSerializeSKillConnReq, &killReq); } static int32_t translateKillQuery(STranslateContext* pCxt, SKillQueryStmt* pStmt) { diff --git a/source/libs/parser/test/mockCatalogService.cpp b/source/libs/parser/test/mockCatalogService.cpp index 68f9e9d36d..9758da7899 100644 --- a/source/libs/parser/test/mockCatalogService.cpp +++ b/source/libs/parser/test/mockCatalogService.cpp @@ -285,7 +285,7 @@ class MockCatalogServiceImpl { } void createSmaIndex(const SMCreateSmaReq* pReq) { - STableIndexInfo info; + STableIndexInfo info = {0}; info.intervalUnit = pReq->intervalUnit; info.slidingUnit = pReq->slidingUnit; info.interval = pReq->interval; diff --git a/source/libs/planner/src/planLogicCreater.c b/source/libs/planner/src/planLogicCreater.c index cbc74a2711..3cd5eeb655 100644 --- a/source/libs/planner/src/planLogicCreater.c +++ b/source/libs/planner/src/planLogicCreater.c @@ -548,6 +548,7 @@ static int32_t createWindowLogicNodeBySession(SLogicPlanContext* pCxt, SSessionW pWindow->winType = WINDOW_TYPE_SESSION; pWindow->sessionGap = ((SValueNode*)pSession->pGap)->datum.i; + pWindow->windowAlgo = pCxt->pPlanCxt->streamQuery ? SESSION_ALGO_STREAM_SINGLE : SESSION_ALGO_MERGE; pWindow->pTspk = nodesCloneNode((SNode*)pSession->pCol); if (NULL == pWindow->pTspk) { @@ -572,7 +573,7 @@ static int32_t createWindowLogicNodeByInterval(SLogicPlanContext* pCxt, SInterva pWindow->sliding = (NULL != pInterval->pSliding ? ((SValueNode*)pInterval->pSliding)->datum.i : pWindow->interval); pWindow->slidingUnit = (NULL != pInterval->pSliding ? ((SValueNode*)pInterval->pSliding)->unit : pWindow->intervalUnit); - pWindow->intervalAlgo = pCxt->pPlanCxt->streamQuery ? INTERVAL_ALGO_STREAM_SINGLE : INTERVAL_ALGO_HASH; + pWindow->windowAlgo = pCxt->pPlanCxt->streamQuery ? INTERVAL_ALGO_STREAM_SINGLE : INTERVAL_ALGO_HASH; pWindow->pTspk = nodesCloneNode(pInterval->pCol); if (NULL == pWindow->pTspk) { diff --git a/source/libs/planner/src/planOptimizer.c b/source/libs/planner/src/planOptimizer.c index 9d7cd0cf27..a7c25162b7 100644 --- a/source/libs/planner/src/planOptimizer.c +++ b/source/libs/planner/src/planOptimizer.c @@ -731,6 +731,7 @@ static int32_t opkDoOptimized(SOptimizeContext* pCxt, SSortLogicNode* pSort, SNo FOREACH(pNode, pSort->node.pParent->pChildren) { if (nodesEqualNode(pNode, (SNode*)pSort)) { REPLACE_NODE(pDownNode); + ((SLogicNode*)pDownNode)->pParent = pSort->node.pParent; break; } } diff --git a/source/libs/planner/src/planPhysiCreater.c b/source/libs/planner/src/planPhysiCreater.c index 2974b3ef8c..99476c4cbd 100644 --- a/source/libs/planner/src/planPhysiCreater.c +++ b/source/libs/planner/src/planPhysiCreater.c @@ -977,8 +977,8 @@ static int32_t createWindowPhysiNodeFinalize(SPhysiPlanContext* pCxt, SNodeList* return code; } -static ENodeType getIntervalOperatorType(EIntervalAlgorithm intervalAlgo) { - switch (intervalAlgo) { +static ENodeType getIntervalOperatorType(EWindowAlgorithm windowAlgo) { + switch (windowAlgo) { case INTERVAL_ALGO_HASH: return QUERY_NODE_PHYSICAL_PLAN_HASH_INTERVAL; case INTERVAL_ALGO_MERGE: @@ -989,6 +989,14 @@ static ENodeType getIntervalOperatorType(EIntervalAlgorithm intervalAlgo) { return QUERY_NODE_PHYSICAL_PLAN_STREAM_SEMI_INTERVAL; case INTERVAL_ALGO_STREAM_SINGLE: return QUERY_NODE_PHYSICAL_PLAN_STREAM_INTERVAL; + case SESSION_ALGO_STREAM_FINAL: + return QUERY_NODE_PHYSICAL_PLAN_STREAM_FINAL_SESSION; + case SESSION_ALGO_STREAM_SEMI: + return QUERY_NODE_PHYSICAL_PLAN_STREAM_SEMI_SESSION; + case SESSION_ALGO_STREAM_SINGLE: + return QUERY_NODE_PHYSICAL_PLAN_STREAM_SESSION; + case SESSION_ALGO_MERGE: + return QUERY_NODE_PHYSICAL_PLAN_MERGE_SESSION; default: break; } @@ -998,7 +1006,7 @@ static ENodeType getIntervalOperatorType(EIntervalAlgorithm intervalAlgo) { static int32_t createIntervalPhysiNode(SPhysiPlanContext* pCxt, SNodeList* pChildren, SWindowLogicNode* pWindowLogicNode, SPhysiNode** pPhyNode) { SIntervalPhysiNode* pInterval = (SIntervalPhysiNode*)makePhysiNode( - pCxt, (SLogicNode*)pWindowLogicNode, getIntervalOperatorType(pWindowLogicNode->intervalAlgo)); + pCxt, (SLogicNode*)pWindowLogicNode, getIntervalOperatorType(pWindowLogicNode->windowAlgo)); if (NULL == pInterval) { return TSDB_CODE_OUT_OF_MEMORY; } @@ -1015,8 +1023,7 @@ static int32_t createIntervalPhysiNode(SPhysiPlanContext* pCxt, SNodeList* pChil static int32_t createSessionWindowPhysiNode(SPhysiPlanContext* pCxt, SNodeList* pChildren, SWindowLogicNode* pWindowLogicNode, SPhysiNode** pPhyNode) { SSessionWinodwPhysiNode* pSession = (SSessionWinodwPhysiNode*)makePhysiNode( - pCxt, (SLogicNode*)pWindowLogicNode, - (pCxt->pPlanCxt->streamQuery ? QUERY_NODE_PHYSICAL_PLAN_STREAM_SESSION : QUERY_NODE_PHYSICAL_PLAN_MERGE_SESSION)); + pCxt, (SLogicNode*)pWindowLogicNode, getIntervalOperatorType(pWindowLogicNode->windowAlgo)); if (NULL == pSession) { return TSDB_CODE_OUT_OF_MEMORY; } diff --git a/source/libs/planner/src/planSpliter.c b/source/libs/planner/src/planSpliter.c index 3a9e1d7405..fe02dd6425 100644 --- a/source/libs/planner/src/planSpliter.c +++ b/source/libs/planner/src/planSpliter.c @@ -376,8 +376,8 @@ static int32_t stbSplSplitIntervalForBatch(SSplitContext* pCxt, SStableSplitInfo SLogicNode* pPartWindow = NULL; int32_t code = stbSplCreatePartWindowNode((SWindowLogicNode*)pInfo->pSplitNode, &pPartWindow); if (TSDB_CODE_SUCCESS == code) { - ((SWindowLogicNode*)pPartWindow)->intervalAlgo = INTERVAL_ALGO_HASH; - ((SWindowLogicNode*)pInfo->pSplitNode)->intervalAlgo = INTERVAL_ALGO_MERGE; + ((SWindowLogicNode*)pPartWindow)->windowAlgo = INTERVAL_ALGO_HASH; + ((SWindowLogicNode*)pInfo->pSplitNode)->windowAlgo = INTERVAL_ALGO_MERGE; SNodeList* pMergeKeys = NULL; code = stbSplCreateMergeKeysByPrimaryKey(((SWindowLogicNode*)pInfo->pSplitNode)->pTspk, &pMergeKeys); if (TSDB_CODE_SUCCESS == code) { @@ -400,8 +400,8 @@ static int32_t stbSplSplitIntervalForStream(SSplitContext* pCxt, SStableSplitInf SLogicNode* pPartWindow = NULL; int32_t code = stbSplCreatePartWindowNode((SWindowLogicNode*)pInfo->pSplitNode, &pPartWindow); if (TSDB_CODE_SUCCESS == code) { - ((SWindowLogicNode*)pPartWindow)->intervalAlgo = INTERVAL_ALGO_STREAM_SEMI; - ((SWindowLogicNode*)pInfo->pSplitNode)->intervalAlgo = INTERVAL_ALGO_STREAM_FINAL; + ((SWindowLogicNode*)pPartWindow)->windowAlgo = INTERVAL_ALGO_STREAM_SEMI; + ((SWindowLogicNode*)pInfo->pSplitNode)->windowAlgo = INTERVAL_ALGO_STREAM_FINAL; code = stbSplCreateExchangeNode(pCxt, pInfo->pSplitNode, pPartWindow); } if (TSDB_CODE_SUCCESS == code) { @@ -421,8 +421,29 @@ static int32_t stbSplSplitInterval(SSplitContext* pCxt, SStableSplitInfo* pInfo) } } +static int32_t stbSplSplitSessionForStream(SSplitContext* pCxt, SStableSplitInfo* pInfo) { + SLogicNode* pPartWindow = NULL; + int32_t code = stbSplCreatePartWindowNode((SWindowLogicNode*)pInfo->pSplitNode, &pPartWindow); + if (TSDB_CODE_SUCCESS == code) { + ((SWindowLogicNode*)pPartWindow)->windowAlgo = SESSION_ALGO_STREAM_SEMI; + ((SWindowLogicNode*)pInfo->pSplitNode)->windowAlgo = SESSION_ALGO_STREAM_FINAL; + code = stbSplCreateExchangeNode(pCxt, pInfo->pSplitNode, pPartWindow); + } + if (TSDB_CODE_SUCCESS == code) { + code = nodesListMakeStrictAppend(&pInfo->pSubplan->pChildren, + (SNode*)splCreateScanSubplan(pCxt, pPartWindow, SPLIT_FLAG_STABLE_SPLIT)); + } + pInfo->pSubplan->subplanType = SUBPLAN_TYPE_MERGE; + ++(pCxt->groupId); + return code; +} + static int32_t stbSplSplitSession(SSplitContext* pCxt, SStableSplitInfo* pInfo) { - return TSDB_CODE_PLAN_INTERNAL_ERROR; + if (pCxt->pPlanCxt->streamQuery) { + return stbSplSplitSessionForStream(pCxt, pInfo); + } else { + return TSDB_CODE_PLAN_INTERNAL_ERROR; + } } static int32_t stbSplSplitWindowNode(SSplitContext* pCxt, SStableSplitInfo* pInfo) { @@ -537,10 +558,12 @@ static int32_t stbSplCreateMergeKeys(SNodeList* pSortKeys, SNodeList* pTargets, SNode* pNode = NULL; FOREACH(pNode, pSortKeys) { SOrderByExprNode* pSortKey = (SOrderByExprNode*)pNode; + SExprNode* pSortExpr = (SExprNode*)pSortKey->pExpr; SNode* pTarget = NULL; bool found = false; FOREACH(pTarget, pTargets) { - if (0 == strcmp(((SExprNode*)pSortKey->pExpr)->aliasName, ((SColumnNode*)pTarget)->colName)) { + if ((QUERY_NODE_COLUMN == nodeType(pSortExpr) && nodesEqualNode((SNode*)pSortExpr, pTarget)) || + (0 == strcmp(pSortExpr->aliasName, ((SColumnNode*)pTarget)->colName))) { code = nodesListMakeStrictAppend(&pMergeKeys, stbSplCreateOrderByExpr(pSortKey, pTarget)); if (TSDB_CODE_SUCCESS != code) { break; @@ -549,7 +572,7 @@ static int32_t stbSplCreateMergeKeys(SNodeList* pSortKeys, SNodeList* pTargets, } } if (TSDB_CODE_SUCCESS == code && !found) { - SNode* pCol = stbSplCreateColumnNode((SExprNode*)pSortKey->pExpr); + SNode* pCol = stbSplCreateColumnNode(pSortExpr); code = nodesListMakeStrictAppend(&pMergeKeys, stbSplCreateOrderByExpr(pSortKey, pCol)); if (TSDB_CODE_SUCCESS == code) { code = nodesListStrictAppend(pTargets, pCol); diff --git a/source/libs/planner/test/planOrderByTest.cpp b/source/libs/planner/test/planOrderByTest.cpp index b065ad0a59..2ca662cf86 100644 --- a/source/libs/planner/test/planOrderByTest.cpp +++ b/source/libs/planner/test/planOrderByTest.cpp @@ -28,6 +28,8 @@ TEST_F(PlanOrderByTest, basic) { // ORDER BY key is not in the projection list run("SELECT c1 FROM t1 ORDER BY c2"); + run("SELECT c1 AS a FROM t1 ORDER BY a"); + run("SELECT c1 + 10 AS a FROM t1 ORDER BY a"); } @@ -59,4 +61,6 @@ TEST_F(PlanOrderByTest, stable) { run("SELECT c2 FROM st1 ORDER BY c1"); run("SELECT c2 FROM st1 PARTITION BY c2 ORDER BY c1"); + + run("SELECT c1 AS a FROM st1 ORDER BY a"); } diff --git a/source/libs/qcom/src/queryUtil.c b/source/libs/qcom/src/queryUtil.c index a5a499aaf5..9322d7e594 100644 --- a/source/libs/qcom/src/queryUtil.c +++ b/source/libs/qcom/src/queryUtil.c @@ -177,10 +177,6 @@ char* jobTaskStatusStr(int32_t status) { return "SUCCEED"; case JOB_TASK_STATUS_FAILED: return "FAILED"; - case JOB_TASK_STATUS_CANCELLING: - return "CANCELLING"; - case JOB_TASK_STATUS_CANCELLED: - return "CANCELLED"; case JOB_TASK_STATUS_DROPPING: return "DROPPING"; default: diff --git a/source/libs/qworker/src/qwDbg.c b/source/libs/qworker/src/qwDbg.c index 368c3bb517..add9700a3a 100644 --- a/source/libs/qworker/src/qwDbg.c +++ b/source/libs/qworker/src/qwDbg.c @@ -44,40 +44,30 @@ int32_t qwDbgValidateStatus(QW_FPARAMS_DEF, int8_t oriStatus, int8_t newStatus, break; case JOB_TASK_STATUS_EXECUTING: if (newStatus != JOB_TASK_STATUS_PARTIAL_SUCCEED && newStatus != JOB_TASK_STATUS_SUCCEED && - newStatus != JOB_TASK_STATUS_FAILED && newStatus != JOB_TASK_STATUS_CANCELLING && - newStatus != JOB_TASK_STATUS_CANCELLED && newStatus != JOB_TASK_STATUS_DROPPING) { + newStatus != JOB_TASK_STATUS_FAILED && newStatus != JOB_TASK_STATUS_DROPPING) { QW_ERR_JRET(TSDB_CODE_QRY_APP_ERROR); } break; case JOB_TASK_STATUS_PARTIAL_SUCCEED: if (newStatus != JOB_TASK_STATUS_EXECUTING && newStatus != JOB_TASK_STATUS_SUCCEED && - newStatus != JOB_TASK_STATUS_CANCELLED && newStatus != JOB_TASK_STATUS_FAILED && - newStatus != JOB_TASK_STATUS_DROPPING) { + newStatus != JOB_TASK_STATUS_FAILED && newStatus != JOB_TASK_STATUS_DROPPING) { QW_ERR_JRET(TSDB_CODE_QRY_APP_ERROR); } break; case JOB_TASK_STATUS_SUCCEED: - if (newStatus != JOB_TASK_STATUS_CANCELLED && newStatus != JOB_TASK_STATUS_DROPPING && - newStatus != JOB_TASK_STATUS_FAILED) { + if (newStatus != JOB_TASK_STATUS_DROPPING && newStatus != JOB_TASK_STATUS_FAILED) { QW_ERR_JRET(TSDB_CODE_QRY_APP_ERROR); } break; case JOB_TASK_STATUS_FAILED: - if (newStatus != JOB_TASK_STATUS_CANCELLED && newStatus != JOB_TASK_STATUS_DROPPING) { + if (newStatus != JOB_TASK_STATUS_DROPPING) { QW_ERR_JRET(TSDB_CODE_QRY_APP_ERROR); } break; - case JOB_TASK_STATUS_CANCELLING: - if (newStatus != JOB_TASK_STATUS_CANCELLED) { - QW_ERR_JRET(TSDB_CODE_QRY_APP_ERROR); - } - - break; - case JOB_TASK_STATUS_CANCELLED: case JOB_TASK_STATUS_DROPPING: if (newStatus != JOB_TASK_STATUS_FAILED && newStatus != JOB_TASK_STATUS_PARTIAL_SUCCEED) { QW_ERR_JRET(TSDB_CODE_QRY_APP_ERROR); diff --git a/source/libs/qworker/src/qworker.c b/source/libs/qworker/src/qworker.c index 57ebb89ed2..800cc4c6e5 100644 --- a/source/libs/qworker/src/qworker.c +++ b/source/libs/qworker/src/qworker.c @@ -614,6 +614,8 @@ int32_t qwProcessCQuery(QW_FPARAMS_DEF, SQWMsg *qwMsg) { QW_SET_EVENT_PROCESSED(ctx, QW_EVENT_FETCH); qwBuildAndSendFetchRsp(&qwMsg->connInfo, rsp, dataLen, code); + rsp = NULL; + QW_TASK_DLOG("fetch rsp send, handle:%p, code:%x - %s, dataLen:%d", qwMsg->connInfo.handle, code, tstrerror(code), dataLen); } else { @@ -633,7 +635,7 @@ int32_t qwProcessCQuery(QW_FPARAMS_DEF, SQWMsg *qwMsg) { rsp = NULL; qwMsg->connInfo = ctx->dataConnInfo; - qwBuildAndSendFetchRsp(&qwMsg->connInfo, rsp, 0, code); + qwBuildAndSendFetchRsp(&qwMsg->connInfo, NULL, 0, code); QW_TASK_DLOG("fetch rsp send, handle:%p, code:%x - %s, dataLen:%d", qwMsg->connInfo.handle, code, tstrerror(code), 0); } diff --git a/source/libs/scheduler/inc/schedulerInt.h b/source/libs/scheduler/inc/schedulerInt.h index 8875ebdf73..545bb6c45a 100644 --- a/source/libs/scheduler/inc/schedulerInt.h +++ b/source/libs/scheduler/inc/schedulerInt.h @@ -48,6 +48,12 @@ enum { SCH_FETCH_CB, }; +typedef enum { + SCH_OP_NULL = 0, + SCH_OP_EXEC, + SCH_OP_FETCH, +} SCH_OP_TYPE; + typedef struct SSchTrans { void *pTrans; void *pHandle; @@ -188,11 +194,15 @@ typedef struct SSchTask { typedef struct SSchJobAttr { EExplainMode explainMode; - bool syncSchedule; bool queryJob; bool needFlowCtrl; } SSchJobAttr; +typedef struct { + int32_t op; + bool sync; +} SSchOpStatus; + typedef struct SSchJob { int64_t refId; uint64_t queryId; @@ -217,8 +227,8 @@ typedef struct SSchJob { int8_t status; SQueryNodeAddr resNode; tsem_t rspSem; - int8_t userFetch; - int32_t remoteFetch; + SSchOpStatus opStatus; + bool *reqKilled; SSchTask *fetchTask; int32_t errCode; SRWLatch resLock; @@ -227,7 +237,6 @@ typedef struct SSchJob { int32_t resNumOfRows; SSchResInfo userRes; const char *sql; - int32_t userCb; SQueryProfileSummary summary; } SSchJob; @@ -285,6 +294,10 @@ extern SSchedulerMgmt schMgmt; #define SCH_GET_JOB_STATUS(job) atomic_load_8(&(job)->status) #define SCH_GET_JOB_STATUS_STR(job) jobTaskStatusStr(SCH_GET_JOB_STATUS(job)) +#define SCH_JOB_IN_SYNC_OP(job) ((job)->opStatus.op && (job)->opStatus.sync) +#define SCH_JOB_IN_ASYNC_EXEC_OP(job) (((job)->opStatus.op == SCH_OP_EXEC) && (!(job)->opStatus.sync)) +#define SCH_JOB_IN_ASYNC_FETCH_OP(job) (((job)->opStatus.op == SCH_OP_FETCH) && (!(job)->opStatus.sync)) + #define SCH_SET_JOB_NEED_FLOW_CTRL(_job) (_job)->attr.needFlowCtrl = true #define SCH_JOB_NEED_FLOW_CTRL(_job) ((_job)->attr.needFlowCtrl) #define SCH_TASK_NEED_FLOW_CTRL(_job, _task) (SCH_IS_DATA_SRC_QRY_TASK(_task) && SCH_JOB_NEED_FLOW_CTRL(_job) && SCH_IS_LEVEL_UNFINISHED((_task)->level)) @@ -356,7 +369,7 @@ int32_t schMakeBrokenLinkVal(SSchJob *pJob, SSchTask *pTask, SRpcBrokenlinkVal * int32_t schAppendTaskExecNode(SSchJob *pJob, SSchTask *pTask, SQueryNodeAddr *addr, int32_t execIdx); int32_t schExecStaticExplainJob(SSchedulerReq *pReq, int64_t *job, bool sync); int32_t schExecJobImpl(SSchedulerReq *pReq, int64_t *job, SQueryResult* pRes, bool sync); -int32_t schChkUpdateJobStatus(SSchJob *pJob, int8_t newStatus); +int32_t schUpdateJobStatus(SSchJob *pJob, int8_t newStatus); int32_t schCancelJob(SSchJob *pJob); int32_t schProcessOnJobDropped(SSchJob *pJob, int32_t errCode); uint64_t schGenTaskId(void); @@ -368,6 +381,8 @@ int32_t schAsyncFetchRows(SSchJob *pJob); int32_t schUpdateTaskHandle(SSchJob *pJob, SSchTask *pTask, bool dropExecNode, void *handle, int32_t execIdx); int32_t schProcessOnTaskStatusRsp(SQueryNodeEpId* pEpId, SArray* pStatusList); void schFreeSMsgSendInfo(SMsgSendInfo *msgSendInfo); +char* schGetOpStr(SCH_OP_TYPE type); +int32_t schBeginOperation(SSchJob *pJob, SCH_OP_TYPE type, bool sync); #ifdef __cplusplus diff --git a/source/libs/scheduler/src/schJob.c b/source/libs/scheduler/src/schJob.c index 31536f413d..733f8694cc 100644 --- a/source/libs/scheduler/src/schJob.c +++ b/source/libs/scheduler/src/schJob.c @@ -21,9 +21,9 @@ #include "tref.h" #include "trpc.h" -FORCE_INLINE SSchJob *schAcquireJob(int64_t refId) { qDebug("acquire jobId:0x%"PRIx64, refId); return (SSchJob *)taosAcquireRef(schMgmt.jobRef, refId); } +FORCE_INLINE SSchJob *schAcquireJob(int64_t refId) { qDebug("sch acquire jobId:0x%"PRIx64, refId); return (SSchJob *)taosAcquireRef(schMgmt.jobRef, refId); } -FORCE_INLINE int32_t schReleaseJob(int64_t refId) { qDebug("release jobId:0x%"PRIx64, refId); return taosReleaseRef(schMgmt.jobRef, refId); } +FORCE_INLINE int32_t schReleaseJob(int64_t refId) { qDebug("sch release jobId:0x%"PRIx64, refId); return taosReleaseRef(schMgmt.jobRef, refId); } int32_t schInitTask(SSchJob *pJob, SSchTask *pTask, SSubplan *pPlan, SSchLevel *pLevel) { pTask->plan = pPlan; @@ -47,14 +47,14 @@ int32_t schInitJob(SSchedulerReq *pReq, SSchJob **pSchJob, SQueryResult* pRes, b int64_t refId = -1; SSchJob *pJob = taosMemoryCalloc(1, sizeof(SSchJob)); if (NULL == pJob) { - qError("QID:%" PRIx64 " calloc %d failed", pReq->pDag->queryId, (int32_t)sizeof(SSchJob)); + qError("QID:0x%" PRIx64 " calloc %d failed", pReq->pDag->queryId, (int32_t)sizeof(SSchJob)); SCH_ERR_RET(TSDB_CODE_QRY_OUT_OF_MEMORY); } pJob->attr.explainMode = pReq->pDag->explainInfo.mode; - pJob->attr.syncSchedule = syncSchedule; pJob->conn = *pReq->pConn; pJob->sql = pReq->sql; + pJob->reqKilled = pReq->reqKilled; pJob->userRes.queryRes = pRes; pJob->userRes.execFp = pReq->fp; pJob->userRes.userParam = pReq->cbParam; @@ -108,7 +108,7 @@ int32_t schInitJob(SSchedulerReq *pReq, SSchJob **pSchJob, SQueryResult* pRes, b atomic_add_fetch_32(&schMgmt.jobNum, 1); if (NULL == schAcquireJob(refId)) { - SCH_JOB_ELOG("schAcquireJob job failed, refId:%" PRIx64, refId); + SCH_JOB_ELOG("schAcquireJob job failed, refId:0x%" PRIx64, refId); SCH_ERR_JRET(TSDB_CODE_SCH_STATUS_ERROR); } @@ -116,7 +116,7 @@ int32_t schInitJob(SSchedulerReq *pReq, SSchJob **pSchJob, SQueryResult* pRes, b SCH_JOB_DLOG("job refId:0x%" PRIx64" created", pJob->refId); - pJob->status = JOB_TASK_STATUS_NOT_START; + schUpdateJobStatus(pJob, JOB_TASK_STATUS_NOT_START); *pSchJob = pJob; @@ -155,18 +155,57 @@ void schFreeTask(SSchJob *pJob, SSchTask *pTask) { } } + +void schUpdateJobErrCode(SSchJob *pJob, int32_t errCode) { + if (TSDB_CODE_SUCCESS == errCode) { + return; + } + + int32_t origCode = atomic_load_32(&pJob->errCode); + if (TSDB_CODE_SUCCESS == origCode) { + if (origCode == atomic_val_compare_exchange_32(&pJob->errCode, origCode, errCode)) { + goto _return; + } + + origCode = atomic_load_32(&pJob->errCode); + } + + if (NEED_CLIENT_HANDLE_ERROR(origCode)) { + return; + } + + if (NEED_CLIENT_HANDLE_ERROR(errCode)) { + atomic_store_32(&pJob->errCode, errCode); + goto _return; + } + + return; + +_return: + + SCH_JOB_DLOG("job errCode updated to %x - %s", errCode, tstrerror(errCode)); +} + + + FORCE_INLINE bool schJobNeedToStop(SSchJob *pJob, int8_t *pStatus) { int8_t status = SCH_GET_JOB_STATUS(pJob); if (pStatus) { *pStatus = status; } - return (status == JOB_TASK_STATUS_FAILED || status == JOB_TASK_STATUS_CANCELLED || - status == JOB_TASK_STATUS_CANCELLING || status == JOB_TASK_STATUS_DROPPING || + if (*pJob->reqKilled) { + schUpdateJobStatus(pJob, JOB_TASK_STATUS_DROPPING); + schUpdateJobErrCode(pJob, TSDB_CODE_TSC_QUERY_KILLED); + + return true; + } + + return (status == JOB_TASK_STATUS_FAILED || status == JOB_TASK_STATUS_DROPPING || status == JOB_TASK_STATUS_SUCCEED); } -int32_t schChkUpdateJobStatus(SSchJob *pJob, int8_t newStatus) { +int32_t schUpdateJobStatus(SSchJob *pJob, int8_t newStatus) { int32_t code = 0; int8_t oriStatus = 0; @@ -175,7 +214,11 @@ int32_t schChkUpdateJobStatus(SSchJob *pJob, int8_t newStatus) { oriStatus = SCH_GET_JOB_STATUS(pJob); if (oriStatus == newStatus) { - SCH_ERR_JRET(TSDB_CODE_QRY_APP_ERROR); + if (newStatus == JOB_TASK_STATUS_DROPPING) { + SCH_ERR_JRET(TSDB_CODE_SCH_JOB_IS_DROPPING); + } + + SCH_ERR_JRET(TSDB_CODE_SCH_IGNORE_ERROR); } switch (oriStatus) { @@ -186,14 +229,13 @@ int32_t schChkUpdateJobStatus(SSchJob *pJob, int8_t newStatus) { break; case JOB_TASK_STATUS_NOT_START: - if (newStatus != JOB_TASK_STATUS_EXECUTING) { + if (newStatus != JOB_TASK_STATUS_EXECUTING && newStatus != JOB_TASK_STATUS_DROPPING) { SCH_ERR_JRET(TSDB_CODE_QRY_APP_ERROR); } break; case JOB_TASK_STATUS_EXECUTING: if (newStatus != JOB_TASK_STATUS_PARTIAL_SUCCEED && newStatus != JOB_TASK_STATUS_FAILED && - newStatus != JOB_TASK_STATUS_CANCELLING && newStatus != JOB_TASK_STATUS_CANCELLED && newStatus != JOB_TASK_STATUS_DROPPING) { SCH_ERR_JRET(TSDB_CODE_QRY_APP_ERROR); } @@ -208,13 +250,11 @@ int32_t schChkUpdateJobStatus(SSchJob *pJob, int8_t newStatus) { break; case JOB_TASK_STATUS_SUCCEED: case JOB_TASK_STATUS_FAILED: - case JOB_TASK_STATUS_CANCELLING: if (newStatus != JOB_TASK_STATUS_DROPPING) { SCH_ERR_JRET(TSDB_CODE_QRY_APP_ERROR); } break; - case JOB_TASK_STATUS_CANCELLED: case JOB_TASK_STATUS_DROPPING: SCH_ERR_JRET(TSDB_CODE_QRY_JOB_FREED); break; @@ -238,8 +278,67 @@ int32_t schChkUpdateJobStatus(SSchJob *pJob, int8_t newStatus) { _return: SCH_JOB_ELOG("invalid job status update, from %s to %s", jobTaskStatusStr(oriStatus), jobTaskStatusStr(newStatus)); - SCH_ERR_RET(code); + SCH_RET(code); +} + + +void schEndOperation(SSchJob *pJob) { + int32_t op = atomic_load_32(&pJob->opStatus.op); + if (SCH_OP_NULL == op) { + SCH_JOB_DLOG("job already not in any operation, status:%s", jobTaskStatusStr(pJob->status)); + return; + } + + atomic_store_32(&pJob->opStatus.op, SCH_OP_NULL); + + SCH_JOB_DLOG("job end %s operation", schGetOpStr(op)); +} + +int32_t schBeginOperation(SSchJob *pJob, SCH_OP_TYPE type, bool sync) { + int32_t code = 0; + int8_t status = 0; + + if (schJobNeedToStop(pJob, &status)) { + SCH_JOB_ELOG("abort op %s cause of job need to stop", schGetOpStr(type)); + SCH_ERR_JRET(pJob->errCode); + } + + if (SCH_OP_NULL != atomic_val_compare_exchange_32(&pJob->opStatus.op, SCH_OP_NULL, type)) { + SCH_JOB_ELOG("job already in %s operation", schGetOpStr(pJob->opStatus.op)); + SCH_ERR_JRET(TSDB_CODE_TSC_APP_ERROR); + } + + SCH_JOB_DLOG("job start %s operation", schGetOpStr(pJob->opStatus.op)); + + pJob->opStatus.sync = sync; + + switch (type) { + case SCH_OP_EXEC: + SCH_ERR_JRET(schUpdateJobStatus(pJob, JOB_TASK_STATUS_EXECUTING)); + break; + case SCH_OP_FETCH: + if (!SCH_JOB_NEED_FETCH(pJob)) { + SCH_JOB_ELOG("no need to fetch data, status:%s", SCH_GET_JOB_STATUS_STR(pJob)); + SCH_ERR_JRET(TSDB_CODE_QRY_APP_ERROR); + } + + if (status != JOB_TASK_STATUS_PARTIAL_SUCCEED) { + SCH_JOB_ELOG("job status error for fetch, status:%s", jobTaskStatusStr(status)); + SCH_ERR_JRET(TSDB_CODE_SCH_STATUS_ERROR); + } + break; + default: + SCH_JOB_ELOG("unknown operation type %d", type); + SCH_ERR_JRET(TSDB_CODE_TSC_APP_ERROR); + } + return TSDB_CODE_SUCCESS; + +_return: + + schEndOperation(pJob); + + SCH_RET(code); } int32_t schBuildTaskRalation(SSchJob *pJob, SHashObj *planToTask) { @@ -278,7 +377,7 @@ int32_t schBuildTaskRalation(SSchJob *pJob, SHashObj *planToTask) { SCH_ERR_RET(TSDB_CODE_QRY_OUT_OF_MEMORY); } - SCH_TASK_DLOG("children info, the %d child TID %" PRIx64, n, (*childTask)->taskId); + SCH_TASK_DLOG("children info, the %d child TID 0x%" PRIx64, n, (*childTask)->taskId); } if (parentNum > 0) { @@ -312,7 +411,7 @@ int32_t schBuildTaskRalation(SSchJob *pJob, SHashObj *planToTask) { SCH_ERR_RET(TSDB_CODE_QRY_OUT_OF_MEMORY); } - SCH_TASK_DLOG("parents info, the %d parent TID %" PRIx64, n, (*parentTask)->taskId); + SCH_TASK_DLOG("parents info, the %d parent TID 0x%" PRIx64, n, (*parentTask)->taskId); } SCH_TASK_DLOG("level:%d, parentNum:%d, childNum:%d", i, parentNum, childNum); @@ -785,37 +884,6 @@ int32_t schHandleTaskRetry(SSchJob *pJob, SSchTask *pTask) { return TSDB_CODE_SUCCESS; } -void schUpdateJobErrCode(SSchJob *pJob, int32_t errCode) { - if (TSDB_CODE_SUCCESS == errCode) { - return; - } - - int32_t origCode = atomic_load_32(&pJob->errCode); - if (TSDB_CODE_SUCCESS == origCode) { - if (origCode == atomic_val_compare_exchange_32(&pJob->errCode, origCode, errCode)) { - goto _return; - } - - origCode = atomic_load_32(&pJob->errCode); - } - - if (NEED_CLIENT_HANDLE_ERROR(origCode)) { - return; - } - - if (NEED_CLIENT_HANDLE_ERROR(errCode)) { - atomic_store_32(&pJob->errCode, errCode); - goto _return; - } - - return; - -_return: - - SCH_JOB_DLOG("job errCode updated to %x - %s", errCode, tstrerror(errCode)); -} - - int32_t schSetJobQueryRes(SSchJob* pJob, SQueryResult* pRes) { pRes->code = atomic_load_32(&pJob->errCode); pRes->numOfRows = pJob->resNumOfRows; @@ -828,7 +896,7 @@ int32_t schSetJobQueryRes(SSchJob* pJob, SQueryResult* pRes) { int32_t schSetJobFetchRes(SSchJob* pJob, void** pData) { int32_t code = 0; if (pJob->resData && ((SRetrieveTableRsp *)pJob->resData)->completed) { - SCH_ERR_RET(schChkUpdateJobStatus(pJob, JOB_TASK_STATUS_SUCCEED)); + SCH_ERR_RET(schUpdateJobStatus(pJob, JOB_TASK_STATUS_SUCCEED)); } while (true) { @@ -855,15 +923,17 @@ int32_t schSetJobFetchRes(SSchJob* pJob, void** pData) { return TSDB_CODE_SUCCESS; } -int32_t schNotifyUserQueryRes(SSchJob* pJob) { - pJob->userRes.queryRes = taosMemoryCalloc(1, sizeof(*pJob->userRes.queryRes)); - if (pJob->userRes.queryRes) { - schSetJobQueryRes(pJob, pJob->userRes.queryRes); +int32_t schNotifyUserExecRes(SSchJob* pJob) { + SQueryResult* pRes = taosMemoryCalloc(1, sizeof(SQueryResult)); + if (pRes) { + schSetJobQueryRes(pJob, pRes); } - (*pJob->userRes.execFp)(pJob->userRes.queryRes, pJob->userRes.userParam, atomic_load_32(&pJob->errCode)); + schEndOperation(pJob); - pJob->userRes.queryRes = NULL; + SCH_JOB_DLOG("sch start to invoke exec cb, code: %s", tstrerror(pJob->errCode)); + (*pJob->userRes.execFp)(pRes, pJob->userRes.userParam, atomic_load_32(&pJob->errCode)); + SCH_JOB_DLOG("sch end from query cb, code: %s", tstrerror(pJob->errCode)); return TSDB_CODE_SUCCESS; } @@ -871,36 +941,52 @@ int32_t schNotifyUserQueryRes(SSchJob* pJob) { int32_t schNotifyUserFetchRes(SSchJob* pJob) { void* pRes = NULL; - SCH_ERR_RET(schSetJobFetchRes(pJob, &pRes)); + schSetJobFetchRes(pJob, &pRes); + schEndOperation(pJob); + + SCH_JOB_DLOG("sch start to invoke fetch cb, code: %s", tstrerror(pJob->errCode)); (*pJob->userRes.fetchFp)(pRes, pJob->userRes.userParam, atomic_load_32(&pJob->errCode)); + SCH_JOB_DLOG("sch end from fetch cb, code: %s", tstrerror(pJob->errCode)); return TSDB_CODE_SUCCESS; } +void schPostJobRes(SSchJob *pJob, SCH_OP_TYPE op) { + if (SCH_OP_NULL == pJob->opStatus.op) { + SCH_JOB_DLOG("job not in any op, no need to post job res, status:%s", jobTaskStatusStr(pJob->status)); + return; + } + + if (op && pJob->opStatus.op != op) { + SCH_JOB_ELOG("job in op %s mis-match with expected %s", schGetOpStr(pJob->opStatus.op), schGetOpStr(op)); + return; + } + + if (SCH_JOB_IN_SYNC_OP(pJob)) { + tsem_post(&pJob->rspSem); + } else if (SCH_JOB_IN_ASYNC_EXEC_OP(pJob)) { + schNotifyUserExecRes(pJob); + } else if (SCH_JOB_IN_ASYNC_FETCH_OP(pJob)) { + schNotifyUserFetchRes(pJob); + } else { + SCH_JOB_ELOG("job not in any operation, status:%s", jobTaskStatusStr(pJob->status)); + } +} + int32_t schProcessOnJobFailureImpl(SSchJob *pJob, int32_t status, int32_t errCode) { // if already FAILED, no more processing - SCH_ERR_RET(schChkUpdateJobStatus(pJob, status)); + SCH_ERR_RET(schUpdateJobStatus(pJob, status)); schUpdateJobErrCode(pJob, errCode); - - if (atomic_load_8(&pJob->userFetch) || pJob->attr.syncSchedule) { - tsem_post(&pJob->rspSem); - } - + int32_t code = atomic_load_32(&pJob->errCode); - - SCH_JOB_DLOG("job failed with error: %s", tstrerror(code)); - - if (!pJob->attr.syncSchedule) { - if (SCH_EXEC_CB == atomic_val_compare_exchange_32(&pJob->userCb, SCH_EXEC_CB, 0)) { - schNotifyUserQueryRes(pJob); - } else if (SCH_FETCH_CB == atomic_val_compare_exchange_32(&pJob->userCb, SCH_FETCH_CB, 0)) { - atomic_val_compare_exchange_8(&pJob->userFetch, 1, 0); - schNotifyUserFetchRes(pJob); - } + if (code) { + SCH_JOB_DLOG("job failed with error: %s", tstrerror(code)); } + schPostJobRes(pJob, 0); + SCH_RET(code); } @@ -918,20 +1004,9 @@ int32_t schProcessOnJobDropped(SSchJob *pJob, int32_t errCode) { int32_t schProcessOnJobPartialSuccess(SSchJob *pJob) { int32_t code = 0; - SCH_ERR_RET(schChkUpdateJobStatus(pJob, JOB_TASK_STATUS_PARTIAL_SUCCEED)); + SCH_ERR_RET(schUpdateJobStatus(pJob, JOB_TASK_STATUS_PARTIAL_SUCCEED)); - if (pJob->attr.syncSchedule) { - tsem_post(&pJob->rspSem); - } else if (SCH_EXEC_CB == atomic_val_compare_exchange_32(&pJob->userCb, SCH_EXEC_CB, 0)) { - schNotifyUserQueryRes(pJob); - } else if (SCH_FETCH_CB == atomic_val_compare_exchange_32(&pJob->userCb, SCH_FETCH_CB, 0)) { - atomic_val_compare_exchange_8(&pJob->userFetch, 1, 0); - schNotifyUserFetchRes(pJob); - } - - if (atomic_load_8(&pJob->userFetch)) { - SCH_ERR_JRET(schFetchFromRemote(pJob)); - } + schPostJobRes(pJob, SCH_OP_EXEC); return TSDB_CODE_SUCCESS; @@ -940,16 +1015,8 @@ _return: SCH_RET(schProcessOnJobFailure(pJob, code)); } -void schProcessOnDataFetched(SSchJob *job) { - atomic_val_compare_exchange_32(&job->remoteFetch, 1, 0); - - if (job->attr.syncSchedule) { - tsem_post(&job->rspSem); - } else if (SCH_FETCH_CB == atomic_val_compare_exchange_32(&job->userCb, SCH_FETCH_CB, 0)) { - atomic_val_compare_exchange_8(&job->userFetch, 1, 0); - - schNotifyUserFetchRes(job); - } +void schProcessOnDataFetched(SSchJob *pJob) { + schPostJobRes(pJob, SCH_OP_FETCH); } // Note: no more task error processing, handled in function internal @@ -1109,7 +1176,7 @@ int32_t schProcessOnTaskSuccess(SSchJob *pJob, SSchTask *pTask) { SCH_UNLOCK(SCH_WRITE, &parent->lock); if (SCH_TASK_READY_FOR_LAUNCH(readyNum, parent)) { - SCH_TASK_DLOG("all %d children task done, start to launch parent task %" PRIx64, readyNum, parent->taskId); + SCH_TASK_DLOG("all %d children task done, start to launch parent task 0x%" PRIx64, readyNum, parent->taskId); SCH_ERR_RET(schLaunchTask(pJob, parent)); } } @@ -1127,15 +1194,8 @@ _return: int32_t schFetchFromRemote(SSchJob *pJob) { int32_t code = 0; - if (atomic_val_compare_exchange_32(&pJob->remoteFetch, 0, 1) != 0) { - SCH_JOB_ELOG("prior fetching not finished, remoteFetch:%d", atomic_load_32(&pJob->remoteFetch)); - return TSDB_CODE_SUCCESS; - } - void *resData = atomic_load_ptr(&pJob->resData); if (resData) { - atomic_val_compare_exchange_32(&pJob->remoteFetch, 1, 0); - SCH_JOB_DLOG("res already fetched, res:%p", resData); return TSDB_CODE_SUCCESS; } @@ -1146,8 +1206,6 @@ int32_t schFetchFromRemote(SSchJob *pJob) { _return: - atomic_val_compare_exchange_32(&pJob->remoteFetch, 1, 0); - SCH_RET(schProcessOnTaskFailure(pJob, pJob->fetchTask, code)); } @@ -1291,7 +1349,7 @@ int32_t schGetTaskFromList(SHashObj *pTaskList, uint64_t taskId, SSchTask **pTas int32_t schGetTaskInJob(SSchJob *pJob, uint64_t taskId, SSchTask **pTask) { schGetTaskFromList(pJob->taskList, taskId, pTask); if (NULL == *pTask) { - SCH_JOB_ELOG("task not found in job task list, taskId:%" PRIx64, taskId); + SCH_JOB_ELOG("task not found in job task list, taskId:0x%" PRIx64, taskId); SCH_ERR_RET(TSDB_CODE_SCH_INTERNAL_ERROR); } @@ -1382,8 +1440,6 @@ int32_t schLaunchLevelTasks(SSchJob *pJob, SSchLevel *level) { int32_t schLaunchJob(SSchJob *pJob) { SSchLevel *level = taosArrayGet(pJob->levels, pJob->levelIdx); - SCH_ERR_RET(schChkUpdateJobStatus(pJob, JOB_TASK_STATUS_EXECUTING)); - SCH_ERR_RET(schChkJobNeedFlowCtrl(pJob, level)); SCH_ERR_RET(schLaunchLevelTasks(pJob, level)); @@ -1466,9 +1522,9 @@ void schFreeJobImpl(void *job) { taosMemoryFreeClear(pJob->userRes.queryRes); taosMemoryFreeClear(pJob->resData); - taosMemoryFreeClear(pJob); + taosMemoryFree(pJob); - qDebug("QID:0x%" PRIx64 " job freed, refId:%" PRIx64 ", pointer:%p", queryId, refId, pJob); + qDebug("QID:0x%" PRIx64 " sch job freed, refId:0x%" PRIx64 ", pointer:%p", queryId, refId, pJob); int32_t jobNum = atomic_sub_fetch_32(&schMgmt.jobNum, 1); if (jobNum == 0) { @@ -1483,26 +1539,36 @@ int32_t schExecJobImpl(SSchedulerReq *pReq, int64_t *job, SQueryResult* pRes, bo int32_t code = 0; SSchJob *pJob = NULL; - SCH_ERR_RET(schInitJob(pReq, &pJob, pRes, sync)); + SCH_ERR_JRET(schInitJob(pReq, &pJob, pRes, sync)); - qDebug("QID:0x%" PRIx64 " job refId 0x%"PRIx64 " started", pReq->pDag->queryId, pJob->refId); + qDebug("QID:0x%" PRIx64 " sch job refId 0x%"PRIx64 " started", pReq->pDag->queryId, pJob->refId); *job = pJob->refId; - if (!sync) { - pJob->userCb = SCH_EXEC_CB; - } + SCH_ERR_JRET(schBeginOperation(pJob, SCH_OP_EXEC, sync)); - SCH_ERR_JRET(schLaunchJob(pJob)); + code = schLaunchJob(pJob); if (sync) { SCH_JOB_DLOG("will wait for rsp now, job status:%s", SCH_GET_JOB_STATUS_STR(pJob)); tsem_wait(&pJob->rspSem); + + schEndOperation(pJob); + } else if (code) { + schPostJobRes(pJob, SCH_OP_EXEC); } - SCH_JOB_DLOG("job exec done, job status:%s, jobId:0x%"PRIx64, SCH_GET_JOB_STATUS_STR(pJob), pJob->refId); + SCH_JOB_DLOG("job exec done, job status:%s, jobId:0x%" PRIx64, SCH_GET_JOB_STATUS_STR(pJob), pJob->refId); + + schReleaseJob(pJob->refId); + + SCH_RET(code); _return: + if (!sync) { + pReq->fp(NULL, pReq->cbParam, code); + } + schReleaseJob(pJob->refId); SCH_RET(code); @@ -1536,10 +1602,10 @@ int32_t schAsyncExecJob(SSchedulerReq *pReq, int64_t *pJob) { *pJob = 0; if (EXPLAIN_MODE_STATIC == pReq->pDag->explainInfo.mode) { - SCH_ERR_RET(schExecStaticExplainJob(pReq, pJob, false)); - } else { - SCH_ERR_RET(schExecJobImpl(pReq, pJob, NULL, false)); + SCH_RET(schExecStaticExplainJob(pReq, pJob, false)); } + + SCH_ERR_RET(schExecJobImpl(pReq, pJob, NULL, false)); return code; } @@ -1550,19 +1616,29 @@ int32_t schExecStaticExplainJob(SSchedulerReq *pReq, int64_t *job, bool sync) { int32_t code = 0; SSchJob *pJob = taosMemoryCalloc(1, sizeof(SSchJob)); if (NULL == pJob) { - qError("QID:%" PRIx64 " calloc %d failed", pReq->pDag->queryId, (int32_t)sizeof(SSchJob)); - SCH_ERR_RET(TSDB_CODE_QRY_OUT_OF_MEMORY); + qError("QID:0x%" PRIx64 " calloc %d failed", pReq->pDag->queryId, (int32_t)sizeof(SSchJob)); + code = TSDB_CODE_QRY_OUT_OF_MEMORY; + pReq->fp(NULL, pReq->cbParam, code); + SCH_ERR_RET(code); } pJob->sql = pReq->sql; + pJob->reqKilled = pReq->reqKilled; pJob->attr.queryJob = true; - pJob->attr.syncSchedule = sync; pJob->attr.explainMode = pReq->pDag->explainInfo.mode; pJob->queryId = pReq->pDag->queryId; pJob->subPlans = pReq->pDag->pSubplans; pJob->userRes.execFp = pReq->fp; pJob->userRes.userParam = pReq->cbParam; + schUpdateJobStatus(pJob, JOB_TASK_STATUS_NOT_START); + + code = schBeginOperation(pJob, SCH_OP_EXEC, sync); + if (code) { + pReq->fp(NULL, pReq->cbParam, code); + SCH_ERR_RET(code); + } + SCH_ERR_JRET(qExecStaticExplain(pReq->pDag, (SRetrieveTableRsp **)&pJob->resData)); int64_t refId = taosAddRef(schMgmt.jobRef, pJob); @@ -1572,21 +1648,23 @@ int32_t schExecStaticExplainJob(SSchedulerReq *pReq, int64_t *job, bool sync) { } if (NULL == schAcquireJob(refId)) { - SCH_JOB_ELOG("schAcquireJob job failed, refId:%" PRIx64, refId); - SCH_ERR_RET(TSDB_CODE_SCH_STATUS_ERROR); + SCH_JOB_ELOG("schAcquireJob job failed, refId:0x%" PRIx64, refId); + SCH_ERR_JRET(TSDB_CODE_SCH_STATUS_ERROR); } pJob->refId = refId; - SCH_JOB_DLOG("job refId:%" PRIx64, pJob->refId); + SCH_JOB_DLOG("job refId:0x%" PRIx64, pJob->refId); pJob->status = JOB_TASK_STATUS_PARTIAL_SUCCEED; *job = pJob->refId; SCH_JOB_DLOG("job exec done, job status:%s", SCH_GET_JOB_STATUS_STR(pJob)); - - if (!pJob->attr.syncSchedule) { - code = schNotifyUserQueryRes(pJob); + + if (!sync) { + schPostJobRes(pJob, SCH_OP_EXEC); + } else { + schEndOperation(pJob); } schReleaseJob(pJob->refId); @@ -1595,56 +1673,29 @@ int32_t schExecStaticExplainJob(SSchedulerReq *pReq, int64_t *job, bool sync) { _return: + schEndOperation(pJob); + if (!sync) { + pReq->fp(NULL, pReq->cbParam, code); + } + schFreeJobImpl(pJob); + SCH_RET(code); } int32_t schFetchRows(SSchJob *pJob) { int32_t code = 0; - int8_t status = SCH_GET_JOB_STATUS(pJob); - if (status == JOB_TASK_STATUS_DROPPING) { - SCH_JOB_ELOG("job is dropping, status:%s", jobTaskStatusStr(status)); - SCH_ERR_RET(TSDB_CODE_SCH_STATUS_ERROR); - } - - if (!SCH_JOB_NEED_FETCH(pJob)) { - SCH_JOB_ELOG("no need to fetch data, status:%s", SCH_GET_JOB_STATUS_STR(pJob)); - SCH_ERR_RET(TSDB_CODE_QRY_APP_ERROR); - } - - if (atomic_val_compare_exchange_8(&pJob->userFetch, 0, 1) != 0) { - SCH_JOB_ELOG("prior fetching not finished, userFetch:%d", atomic_load_8(&pJob->userFetch)); - SCH_ERR_RET(TSDB_CODE_QRY_APP_ERROR); - } - - if (JOB_TASK_STATUS_FAILED == status || JOB_TASK_STATUS_DROPPING == status) { - SCH_JOB_ELOG("job failed or dropping, status:%s", jobTaskStatusStr(status)); - SCH_ERR_JRET(atomic_load_32(&pJob->errCode)); - } else if (status == JOB_TASK_STATUS_SUCCEED) { - SCH_JOB_DLOG("job already succeed, status:%s", jobTaskStatusStr(status)); - goto _return; - } else if (status != JOB_TASK_STATUS_PARTIAL_SUCCEED) { - SCH_JOB_ELOG("job status error for fetch, status:%s", jobTaskStatusStr(status)); - SCH_ERR_JRET(TSDB_CODE_SCH_STATUS_ERROR); - } - if (!(pJob->attr.explainMode == EXPLAIN_MODE_STATIC)) { SCH_ERR_JRET(schFetchFromRemote(pJob)); tsem_wait(&pJob->rspSem); - - status = SCH_GET_JOB_STATUS(pJob); - if (JOB_TASK_STATUS_FAILED == status || JOB_TASK_STATUS_DROPPING == status) { - SCH_JOB_ELOG("job failed or dropping, status:%s", jobTaskStatusStr(status)); - SCH_ERR_JRET(atomic_load_32(&pJob->errCode)); - } } SCH_ERR_JRET(schSetJobFetchRes(pJob, pJob->userRes.fetchRes)); _return: - atomic_val_compare_exchange_8(&pJob->userFetch, 1, 0); + schEndOperation(pJob); SCH_RET(code); } @@ -1652,50 +1703,14 @@ _return: int32_t schAsyncFetchRows(SSchJob *pJob) { int32_t code = 0; - int8_t status = SCH_GET_JOB_STATUS(pJob); - if (status == JOB_TASK_STATUS_DROPPING) { - SCH_JOB_ELOG("job is dropping, status:%s", jobTaskStatusStr(status)); - SCH_ERR_RET(TSDB_CODE_SCH_STATUS_ERROR); - } - - if (!SCH_JOB_NEED_FETCH(pJob)) { - SCH_JOB_ELOG("no need to fetch data, status:%s", SCH_GET_JOB_STATUS_STR(pJob)); - SCH_ERR_RET(TSDB_CODE_QRY_APP_ERROR); - } - - if (atomic_val_compare_exchange_8(&pJob->userFetch, 0, 1) != 0) { - SCH_JOB_ELOG("prior fetching not finished, userFetch:%d", atomic_load_8(&pJob->userFetch)); - SCH_ERR_RET(TSDB_CODE_QRY_APP_ERROR); - } - - if (JOB_TASK_STATUS_FAILED == status || JOB_TASK_STATUS_DROPPING == status) { - SCH_JOB_ELOG("job failed or dropping, status:%s", jobTaskStatusStr(status)); - SCH_ERR_JRET(atomic_load_32(&pJob->errCode)); - } else if (status == JOB_TASK_STATUS_SUCCEED) { - SCH_JOB_DLOG("job already succeed, status:%s", jobTaskStatusStr(status)); - goto _return; - } else if (status != JOB_TASK_STATUS_PARTIAL_SUCCEED) { - SCH_JOB_ELOG("job status error for fetch, status:%s", jobTaskStatusStr(status)); - SCH_ERR_JRET(TSDB_CODE_SCH_STATUS_ERROR); - } - if (pJob->attr.explainMode == EXPLAIN_MODE_STATIC) { - atomic_val_compare_exchange_8(&pJob->userFetch, 1, 0); - - SCH_ERR_JRET(schNotifyUserFetchRes(pJob)); - } else { - pJob->userCb = SCH_FETCH_CB; - - SCH_ERR_JRET(schFetchFromRemote(pJob)); + schPostJobRes(pJob, SCH_OP_FETCH); + return TSDB_CODE_SUCCESS; } + + SCH_ERR_RET(schFetchFromRemote(pJob)); return TSDB_CODE_SUCCESS; - -_return: - - atomic_val_compare_exchange_8(&pJob->userFetch, 1, 0); - - SCH_RET(code); } diff --git a/source/libs/scheduler/src/schRemote.c b/source/libs/scheduler/src/schRemote.c index ddcb53d62f..0bd747785c 100644 --- a/source/libs/scheduler/src/schRemote.c +++ b/source/libs/scheduler/src/schRemote.c @@ -315,8 +315,6 @@ int32_t schHandleResponseMsg(SSchJob *pJob, SSchTask *pTask, int32_t msgType, ch return TSDB_CODE_SUCCESS; } - atomic_val_compare_exchange_32(&pJob->remoteFetch, 1, 0); - SCH_ERR_JRET(schFetchFromRemote(pJob)); taosMemoryFreeClear(msg); @@ -346,7 +344,7 @@ int32_t schHandleResponseMsg(SSchJob *pJob, SSchTask *pTask, int32_t msgType, ch } case TDMT_VND_DROP_TASK_RSP: { // SHOULD NEVER REACH HERE - SCH_TASK_ELOG("invalid status to handle drop task rsp, refId:%" PRIx64, pJob->refId); + SCH_TASK_ELOG("invalid status to handle drop task rsp, refId:0x%" PRIx64, pJob->refId); SCH_ERR_JRET(TSDB_CODE_SCH_INTERNAL_ERROR); break; } @@ -376,7 +374,7 @@ int32_t schHandleCallback(void *param, const SDataBuf *pMsg, int32_t msgType, in SSchJob *pJob = schAcquireJob(pParam->refId); if (NULL == pJob) { - qWarn("QID:0x%" PRIx64 ",TID:0x%" PRIx64 "taosAcquireRef job failed, may be dropped, refId:%" PRIx64, + qWarn("QID:0x%" PRIx64 ",TID:0x%" PRIx64 "taosAcquireRef job failed, may be dropped, refId:0x%" PRIx64, pParam->queryId, pParam->taskId, pParam->refId); SCH_ERR_JRET(TSDB_CODE_QRY_JOB_FREED); } @@ -445,7 +443,7 @@ int32_t schHandleExplainCallback(void *param, const SDataBuf *pMsg, int32_t code int32_t schHandleDropCallback(void *param, const SDataBuf *pMsg, int32_t code) { SSchTaskCallbackParam *pParam = (SSchTaskCallbackParam *)param; - qDebug("QID:%" PRIx64 ",TID:%" PRIx64 " drop task rsp received, code:%x", pParam->queryId, pParam->taskId, code); + qDebug("QID:0x%" PRIx64 ",TID:0x%" PRIx64 " drop task rsp received, code:0x%x", pParam->queryId, pParam->taskId, code); taosMemoryFreeClear(param); return TSDB_CODE_SUCCESS; } diff --git a/source/libs/scheduler/src/schUtil.c b/source/libs/scheduler/src/schUtil.c index 66483187da..73077cbf0f 100644 --- a/source/libs/scheduler/src/schUtil.c +++ b/source/libs/scheduler/src/schUtil.c @@ -21,6 +21,18 @@ #include "tref.h" #include "trpc.h" +char* schGetOpStr(SCH_OP_TYPE type) { + switch (type) { + case SCH_OP_NULL: + return "NULL"; + case SCH_OP_EXEC: + return "EXEC"; + case SCH_OP_FETCH: + return "FETCH"; + default: + return "UNKNOWN"; + } +} void schCleanClusterHb(void* pTrans) { SCH_LOCK(SCH_WRITE, &schMgmt.hbLock); @@ -188,7 +200,7 @@ int32_t schUpdateHbConnection(SQueryNodeEpId *epId, SSchTrans *trans) { SCH_UNLOCK(SCH_WRITE, &hb->lock); SCH_UNLOCK(SCH_READ, &schMgmt.hbLock); - qDebug("hb connection updated, sId:%" PRIx64 ", nodeId:%d, fqdn:%s, port:%d, pTrans:%p, pHandle:%p", schMgmt.sId, + qDebug("hb connection updated, sId:0x%" PRIx64 ", nodeId:%d, fqdn:%s, port:%d, pTrans:%p, pHandle:%p", schMgmt.sId, epId->nodeId, epId->ep.fqdn, epId->ep.port, trans->pTrans, trans->pHandle); return TSDB_CODE_SUCCESS; diff --git a/source/libs/scheduler/src/scheduler.c b/source/libs/scheduler/src/scheduler.c index 0eaeeae9cb..57a405ffa3 100644 --- a/source/libs/scheduler/src/scheduler.c +++ b/source/libs/scheduler/src/scheduler.c @@ -62,12 +62,14 @@ int32_t schedulerInit(SSchedulerCfg *cfg) { SCH_ERR_RET(TSDB_CODE_QRY_SYS_ERROR); } - qInfo("scheduler %" PRIx64 " initizlized, maxJob:%u", schMgmt.sId, schMgmt.cfg.maxJobNum); + qInfo("scheduler 0x%" PRIx64 " initizlized, maxJob:%u", schMgmt.sId, schMgmt.cfg.maxJobNum); return TSDB_CODE_SUCCESS; } int32_t schedulerExecJob(SSchedulerReq *pReq, int64_t *pJob, SQueryResult *pRes) { + qDebug("scheduler sync exec job start"); + if (NULL == pReq || NULL == pJob || NULL == pRes) { SCH_ERR_RET(TSDB_CODE_QRY_INVALID_INPUT); } @@ -76,21 +78,27 @@ int32_t schedulerExecJob(SSchedulerReq *pReq, int64_t *pJob, SQueryResult *pRes) } int32_t schedulerAsyncExecJob(SSchedulerReq *pReq, int64_t *pJob) { + qDebug("scheduler async exec job start"); + int32_t code = 0; if (NULL == pReq || NULL == pJob) { - code = TSDB_CODE_QRY_INVALID_INPUT; - } else { - code = schAsyncExecJob(pReq, pJob); + SCH_ERR_JRET(TSDB_CODE_QRY_INVALID_INPUT); } + + schAsyncExecJob(pReq, pJob); + +_return: if (code != TSDB_CODE_SUCCESS) { pReq->fp(NULL, pReq->cbParam, code); } - return code; + SCH_RET(code); } int32_t schedulerFetchRows(int64_t job, void **pData) { + qDebug("scheduler sync fetch rows start"); + if (NULL == pData) { SCH_ERR_RET(TSDB_CODE_QRY_INVALID_INPUT); } @@ -102,7 +110,8 @@ int32_t schedulerFetchRows(int64_t job, void **pData) { SCH_ERR_RET(TSDB_CODE_SCH_STATUS_ERROR); } - pJob->attr.syncSchedule = true; + SCH_ERR_RET(schBeginOperation(pJob, SCH_OP_FETCH, true)); + pJob->userRes.fetchRes = pData; code = schFetchRows(pJob); @@ -112,24 +121,32 @@ int32_t schedulerFetchRows(int64_t job, void **pData) { } void schedulerAsyncFetchRows(int64_t job, schedulerFetchCallback fp, void* param) { + qDebug("scheduler async fetch rows start"); + + int32_t code = 0; if (NULL == fp || NULL == param) { - fp(NULL, param, TSDB_CODE_QRY_INVALID_INPUT); - return; + SCH_ERR_JRET(TSDB_CODE_QRY_INVALID_INPUT); } SSchJob *pJob = schAcquireJob(job); if (NULL == pJob) { - qError("acquire job from jobRef list failed, may be dropped, jobId:0x%" PRIx64, job); - fp(NULL, param, TSDB_CODE_SCH_STATUS_ERROR); - return; + qError("acquire sch job from job list failed, may be dropped, jobId:0x%" PRIx64, job); + SCH_ERR_JRET(TSDB_CODE_SCH_STATUS_ERROR); } - pJob->attr.syncSchedule = false; + SCH_ERR_JRET(schBeginOperation(pJob, SCH_OP_FETCH, false)); + pJob->userRes.fetchFp = fp; pJob->userRes.userParam = param; - /*code = */schAsyncFetchRows(pJob); + SCH_ERR_JRET(schAsyncFetchRows(pJob)); +_return: + + if (code) { + fp(NULL, param, code); + } + schReleaseJob(job); } @@ -137,12 +154,12 @@ int32_t schedulerGetTasksStatus(int64_t job, SArray *pSub) { int32_t code = 0; SSchJob *pJob = schAcquireJob(job); if (NULL == pJob) { - qDebug("acquire job from jobRef list failed, may not started or dropped, refId:%" PRIx64, job); + qDebug("acquire job from jobRef list failed, may not started or dropped, refId:0x%" PRIx64, job); SCH_ERR_RET(TSDB_CODE_SCH_STATUS_ERROR); } if (pJob->status < JOB_TASK_STATUS_NOT_START || pJob->levelNum <= 0 || NULL == pJob->levels) { - qDebug("job not initialized or not executable job, refId:%" PRIx64, job); + qDebug("job not initialized or not executable job, refId:0x%" PRIx64, job); SCH_ERR_JRET(TSDB_CODE_SCH_STATUS_ERROR); } @@ -188,21 +205,23 @@ void schedulerStopQueryHb(void *pTrans) { schCleanClusterHb(pTrans); } -void schedulerFreeJob(int64_t job) { +void schedulerFreeJob(int64_t job, int32_t errCode) { SSchJob *pJob = schAcquireJob(job); if (NULL == pJob) { qError("acquire job from jobRef list failed, may be dropped, jobId:0x%" PRIx64, job); return; } - if (atomic_load_8(&pJob->userFetch) > 0) { - schProcessOnJobDropped(pJob, TSDB_CODE_QRY_JOB_FREED); + int32_t code = schProcessOnJobDropped(pJob, errCode); + if (TSDB_CODE_SCH_JOB_IS_DROPPING == code) { + SCH_JOB_DLOG("sch job is already dropping, refId:0x%" PRIx64, job); + return; } - SCH_JOB_DLOG("start to remove job from jobRef list, refId:%" PRIx64, job); + SCH_JOB_DLOG("start to remove job from jobRef list, refId:0x%" PRIx64, job); if (taosRemoveRef(schMgmt.jobRef, job)) { - SCH_JOB_ELOG("remove job from job list failed, refId:%" PRIx64, job); + SCH_JOB_ELOG("remove job from job list failed, refId:0x%" PRIx64, job); } schReleaseJob(job); diff --git a/source/libs/scheduler/test/schedulerTests.cpp b/source/libs/scheduler/test/schedulerTests.cpp index 9b624ee5cd..e5cc3cd481 100644 --- a/source/libs/scheduler/test/schedulerTests.cpp +++ b/source/libs/scheduler/test/schedulerTests.cpp @@ -457,7 +457,7 @@ void schtFreeQueryJob(int32_t freeThread) { int64_t job = queryJobRefId; if (job && atomic_val_compare_exchange_64(&queryJobRefId, job, 0)) { - schedulerFreeJob(job); + schedulerFreeJob(job, 0); if (freeThread) { if (++freeNum % schtTestPrintNum == 0) { printf("FreeNum:%d\n", freeNum); @@ -724,7 +724,7 @@ TEST(queryTest, normalCase) { schReleaseJob(job); - schedulerFreeJob(job); + schedulerFreeJob(job, 0); schtFreeQueryDag(&dag); @@ -828,7 +828,7 @@ TEST(queryTest, readyFirstCase) { schReleaseJob(job); - schedulerFreeJob(job); + schedulerFreeJob(job, 0); schtFreeQueryDag(&dag); @@ -940,7 +940,7 @@ TEST(queryTest, flowCtrlCase) { schReleaseJob(job); - schedulerFreeJob(job); + schedulerFreeJob(job, 0); schtFreeQueryDag(&dag); @@ -994,7 +994,7 @@ TEST(insertTest, normalCase) { ASSERT_EQ(code, 0); ASSERT_EQ(res.numOfRows, 20); - schedulerFreeJob(insertJobRefId); + schedulerFreeJob(insertJobRefId, 0); schedulerDestroy(); } diff --git a/source/libs/sync/src/syncAppendEntries.c b/source/libs/sync/src/syncAppendEntries.c index 7b2f79e24c..a8cb3d0e69 100644 --- a/source/libs/sync/src/syncAppendEntries.c +++ b/source/libs/sync/src/syncAppendEntries.c @@ -713,7 +713,7 @@ static int32_t syncNodeMakeLogSame(SSyncNode* ths, SyncAppendEntries* pMsg) { // delete confict entries code = ths->pLogStore->syncLogTruncate(ths->pLogStore, delBegin); ASSERT(code == 0); - sDebug("vgId:%d sync event %s commitIndex:%ld currentTerm:%lu log truncate, from %ld to %ld", ths->vgId, + sDebug("vgId:%d, sync event %s commitIndex:%ld currentTerm:%lu log truncate, from %ld to %ld", ths->vgId, syncUtilState2String(ths->state), ths->commitIndex, ths->pRaftStore->currentTerm, delBegin, delEnd); logStoreSimpleLog2("after syncNodeMakeLogSame", ths->pLogStore); @@ -1062,7 +1062,7 @@ int32_t syncNodeOnAppendEntriesSnapshotCb(SSyncNode* ths, SyncAppendEntries* pMs ths->commitIndex = snapshot.lastApplyIndex; sDebug( - "vgId:%d sync event %s commitIndex:%ld currentTerm:%lu commit by snapshot from index:%ld to index:%ld", + "vgId:%d, sync event %s commitIndex:%ld currentTerm:%lu commit by snapshot from index:%ld to index:%ld", ths->vgId, syncUtilState2String(ths->state), ths->commitIndex, ths->pRaftStore->currentTerm, commitBegin, commitEnd); } diff --git a/source/libs/sync/src/syncAppendEntriesReply.c b/source/libs/sync/src/syncAppendEntriesReply.c index f934f9a268..eabd14bf28 100644 --- a/source/libs/sync/src/syncAppendEntriesReply.c +++ b/source/libs/sync/src/syncAppendEntriesReply.c @@ -190,7 +190,7 @@ int32_t syncNodeOnAppendEntriesReplySnapshotCb(SSyncNode* ths, SyncAppendEntries if (gRaftDetailLog) { char* s = snapshotSender2Str(pSender); sDebug( - "vgId:%d sync event %s commitIndex:%ld currentTerm:%lu snapshot send to %s:%d start sender first time, " + "vgId:%d, sync event %s commitIndex:%ld currentTerm:%lu snapshot send to %s:%d start sender first time, " "lastApplyIndex:%ld " "lastApplyTerm:%lu " "lastConfigIndex:%ld privateTerm:%lu " @@ -201,7 +201,7 @@ int32_t syncNodeOnAppendEntriesReplySnapshotCb(SSyncNode* ths, SyncAppendEntries taosMemoryFree(s); } else { sDebug( - "vgId:%d sync event %s commitIndex:%ld currentTerm:%lu snapshot send to %s:%d start sender first time, " + "vgId:%d, sync event %s commitIndex:%ld currentTerm:%lu snapshot send to %s:%d start sender first time, " "lastApplyIndex:%ld " "lastApplyTerm:%lu lastConfigIndex:%ld privateTerm:%lu", ths->vgId, syncUtilState2String(ths->state), ths->commitIndex, ths->pRaftStore->currentTerm, host, port, diff --git a/source/libs/sync/src/syncCommit.c b/source/libs/sync/src/syncCommit.c index d010728c78..d563e5aacc 100644 --- a/source/libs/sync/src/syncCommit.c +++ b/source/libs/sync/src/syncCommit.c @@ -56,7 +56,7 @@ void syncMaybeAdvanceCommitIndex(SSyncNode* pSyncNode) { SyncIndex commitEnd = snapshot.lastApplyIndex; pSyncNode->commitIndex = snapshot.lastApplyIndex; - sDebug("vgId:%d sync event %s commitIndex:%ld currentTerm:%lu commit by snapshot from index:%ld to index:%ld", + sDebug("vgId:%d, sync event %s commitIndex:%ld currentTerm:%lu commit by snapshot from index:%ld to index:%ld", pSyncNode->vgId, syncUtilState2String(pSyncNode->state), pSyncNode->commitIndex, pSyncNode->pRaftStore->currentTerm, pSyncNode->commitIndex, snapshot.lastApplyIndex); } diff --git a/source/libs/sync/src/syncMain.c b/source/libs/sync/src/syncMain.c index 3e5dcfea42..2b06673a62 100644 --- a/source/libs/sync/src/syncMain.c +++ b/source/libs/sync/src/syncMain.c @@ -414,7 +414,7 @@ int32_t syncGetSnapshotMeta(int64_t rid, struct SSnapshotMeta* sMeta) { assert(rid == pSyncNode->rid); sMeta->lastConfigIndex = pSyncNode->pRaftCfg->lastConfigIndex; - sTrace("sync get snapshot meta: lastConfigIndex:%ld", pSyncNode->pRaftCfg->lastConfigIndex); + sTrace("vgId:%d, get snapshot meta, lastConfigIndex:%" PRId64, pSyncNode->vgId, pSyncNode->pRaftCfg->lastConfigIndex); taosReleaseRef(tsNodeRefId, pSyncNode->rid); return 0; @@ -437,7 +437,8 @@ int32_t syncGetSnapshotMetaByIndex(int64_t rid, SyncIndex snapshotIndex, struct } } sMeta->lastConfigIndex = lastIndex; - sTrace("sync get snapshot meta by index:%ld lastConfigIndex:%ld", snapshotIndex, sMeta->lastConfigIndex); + sTrace("vgId:%d, get snapshot meta by index:%" PRId64 " lastConfigIndex:%" PRId64, pSyncNode->vgId, snapshotIndex, + sMeta->lastConfigIndex); taosReleaseRef(tsNodeRefId, pSyncNode->rid); return 0; @@ -598,7 +599,7 @@ int32_t syncPropose(int64_t rid, const SRpcMsg* pMsg, bool isWeak) { return -1; } assert(rid == pSyncNode->rid); - sDebug("vgId:%d sync event %s commitIndex:%ld currentTerm:%lu propose msgType:%s,%d", pSyncNode->vgId, + sDebug("vgId:%d, sync event %s commitIndex:%ld currentTerm:%lu propose msgType:%s,%d", pSyncNode->vgId, syncUtilState2String(pSyncNode->state), pSyncNode->commitIndex, pSyncNode->pRaftStore->currentTerm, TMSG_INFO(pMsg->msgType), pMsg->msgType); ret = syncNodePropose(pSyncNode, pMsg, isWeak); @@ -609,7 +610,7 @@ int32_t syncPropose(int64_t rid, const SRpcMsg* pMsg, bool isWeak) { int32_t syncNodePropose(SSyncNode* pSyncNode, const SRpcMsg* pMsg, bool isWeak) { int32_t ret = 0; - sDebug("vgId:%d sync event %s commitIndex:%ld currentTerm:%lu propose msgType:%s,%d", pSyncNode->vgId, + sDebug("vgId:%d, sync event %s commitIndex:%ld currentTerm:%lu propose msgType:%s,%d", pSyncNode->vgId, syncUtilState2String(pSyncNode->state), pSyncNode->commitIndex, pSyncNode->pRaftStore->currentTerm, TMSG_INFO(pMsg->msgType), pMsg->msgType); @@ -857,7 +858,7 @@ SSyncNode* syncNodeOpen(const SSyncInfo* pOldSyncInfo) { // snapshot meta // pSyncNode->sMeta.lastConfigIndex = -1; - sDebug("vgId:%d sync event %s commitIndex:%ld currentTerm:%lu sync open", pSyncNode->vgId, + sDebug("vgId:%d, sync event %s commitIndex:%ld currentTerm:%lu sync open", pSyncNode->vgId, syncUtilState2String(pSyncNode->state), pSyncNode->commitIndex, pSyncNode->pRaftStore->currentTerm); return pSyncNode; @@ -905,7 +906,7 @@ void syncNodeStartStandBy(SSyncNode* pSyncNode) { } void syncNodeClose(SSyncNode* pSyncNode) { - sDebug("vgId:%d sync event %s commitIndex:%ld currentTerm:%lu sync close", pSyncNode->vgId, + sDebug("vgId:%d, sync event %s commitIndex:%ld currentTerm:%lu sync close", pSyncNode->vgId, syncUtilState2String(pSyncNode->state), pSyncNode->commitIndex, pSyncNode->pRaftStore->currentTerm); int32_t ret; @@ -1294,7 +1295,7 @@ char* syncNode2SimpleStr(const SSyncNode* pSyncNode) { int len = 256; char* s = (char*)taosMemoryMalloc(len); snprintf(s, len, - "syncNode: vgId:%d currentTerm:%lu, commitIndex:%ld, state:%d %s, isStandBy:%d, " + "syncNode: vgId:%d, currentTerm:%lu, commitIndex:%ld, state:%d %s, isStandBy:%d, " "electTimerLogicClock:%lu, " "electTimerLogicClockUser:%lu, " "electTimerMS:%d, replicaNum:%d", @@ -1345,7 +1346,7 @@ void syncNodeUpdateConfig(SSyncNode* pSyncNode, SSyncCfg* pNewConfig, SyncIndex SSyncSnapshotSender* oldSenders[TSDB_MAX_REPLICA]; for (int i = 0; i < TSDB_MAX_REPLICA; ++i) { oldSenders[i] = (pSyncNode->senders)[i]; - sDebug("vgId:%d sync event %s commitIndex:%ld currentTerm:%lu save senders %d, %p, privateTerm:%lu", + sDebug("vgId:%d, sync event %s commitIndex:%ld currentTerm:%lu save senders %d, %p, privateTerm:%lu", pSyncNode->vgId, syncUtilState2String(pSyncNode->state), pSyncNode->commitIndex, pSyncNode->pRaftStore->currentTerm, i, oldSenders[i], oldSenders[i]->privateTerm); if (gRaftDetailLog) { @@ -1400,7 +1401,7 @@ void syncNodeUpdateConfig(SSyncNode* pSyncNode, SSyncCfg* pNewConfig, SyncIndex uint16_t port; syncUtilU642Addr((pSyncNode->replicasId)[i].addr, host, sizeof(host), &port); sDebug( - "vgId:%d sync event %s commitIndex:%ld currentTerm:%lu reset sender for %lu, newIndex:%d, %s:%d, %p, " + "vgId:%d, sync event %s commitIndex:%ld currentTerm:%lu reset sender for %lu, newIndex:%d, %s:%d, %p, " "privateTerm:%lu", pSyncNode->vgId, syncUtilState2String(pSyncNode->state), pSyncNode->commitIndex, pSyncNode->pRaftStore->currentTerm, (pSyncNode->replicasId)[i].addr, i, host, port, oldSenders[j], @@ -1413,7 +1414,7 @@ void syncNodeUpdateConfig(SSyncNode* pSyncNode, SSyncCfg* pNewConfig, SyncIndex int32_t oldreplicaIndex = (pSyncNode->senders)[i]->replicaIndex; (pSyncNode->senders)[i]->replicaIndex = i; sDebug( - "vgId:%d sync event %s commitIndex:%ld currentTerm:%lu udpate replicaIndex from %d to %d, %s:%d, %p, " + "vgId:%d, sync event %s commitIndex:%ld currentTerm:%lu udpate replicaIndex from %d to %d, %s:%d, %p, " "reset:%d", pSyncNode->vgId, syncUtilState2String(pSyncNode->state), pSyncNode->commitIndex, pSyncNode->pRaftStore->currentTerm, oldreplicaIndex, i, host, port, (pSyncNode->senders)[i], reset); @@ -1426,7 +1427,7 @@ void syncNodeUpdateConfig(SSyncNode* pSyncNode, SSyncCfg* pNewConfig, SyncIndex if ((pSyncNode->senders)[i] == NULL) { (pSyncNode->senders)[i] = snapshotSenderCreate(pSyncNode, i); sDebug( - "vgId:%d sync event %s commitIndex:%ld currentTerm:%lu create new sender %p replicaIndex:%d, privateTerm:%lu", + "vgId:%d, sync event %s commitIndex:%ld currentTerm:%lu create new sender %p replicaIndex:%d, privateTerm:%lu", pSyncNode->vgId, syncUtilState2String(pSyncNode->state), pSyncNode->commitIndex, pSyncNode->pRaftStore->currentTerm, (pSyncNode->senders)[i], i, (pSyncNode->senders)[i]->privateTerm); } @@ -1436,7 +1437,7 @@ void syncNodeUpdateConfig(SSyncNode* pSyncNode, SSyncCfg* pNewConfig, SyncIndex for (int i = 0; i < TSDB_MAX_REPLICA; ++i) { if (oldSenders[i] != NULL) { snapshotSenderDestroy(oldSenders[i]); - sDebug("vgId:%d sync event %s commitIndex:%ld currentTerm:%lu delete old sender %p replicaIndex:%d", + sDebug("vgId:%d, sync event %s commitIndex:%ld currentTerm:%lu delete old sender %p replicaIndex:%d", pSyncNode->vgId, syncUtilState2String(pSyncNode->state), pSyncNode->commitIndex, pSyncNode->pRaftStore->currentTerm, oldSenders[i], i); oldSenders[i] = NULL; @@ -1509,7 +1510,7 @@ void syncNodeUpdateTerm(SSyncNode* pSyncNode, SyncTerm term) { void syncNodeBecomeFollower(SSyncNode* pSyncNode, const char* debugStr) { sDebug( - "vgId:%d sync event %s commitIndex:%ld currentTerm:%lu become follower, isStandBy:%d, replicaNum:%d, " + "vgId:%d, sync event %s commitIndex:%ld currentTerm:%lu become follower, isStandBy:%d, replicaNum:%d, " "restoreFinish:%d, %s", pSyncNode->vgId, syncUtilState2String(pSyncNode->state), pSyncNode->commitIndex, pSyncNode->pRaftStore->currentTerm, pSyncNode->pRaftCfg->isStandBy, pSyncNode->replicaNum, @@ -1551,7 +1552,7 @@ void syncNodeBecomeLeader(SSyncNode* pSyncNode, const char* debugStr) { pSyncNode->restoreFinish = false; sDebug( - "vgId:%d sync event %s commitIndex:%ld currentTerm:%lu become leader, isStandBy:%d, replicaNum:%d, " + "vgId:%d, sync event %s commitIndex:%ld currentTerm:%lu become leader, isStandBy:%d, replicaNum:%d, " "restoreFinish:%d, %s", pSyncNode->vgId, syncUtilState2String(pSyncNode->state), pSyncNode->commitIndex, pSyncNode->pRaftStore->currentTerm, pSyncNode->pRaftCfg->isStandBy, pSyncNode->replicaNum, @@ -1857,7 +1858,7 @@ static void syncNodeEqPingTimer(void* param, void* tmrId) { if (pSyncNode->FpEqMsg != NULL) { int32_t code = pSyncNode->FpEqMsg(pSyncNode->msgcb, &rpcMsg); if (code != 0) { - sError("vgId:%d sync enqueue ping msg error, code:%d", pSyncNode->vgId, code); + sError("vgId:%d, sync enqueue ping msg error, code:%d", pSyncNode->vgId, code); rpcFreeCont(rpcMsg.pCont); syncTimeoutDestroy(pSyncMsg); return; @@ -1891,7 +1892,7 @@ static void syncNodeEqElectTimer(void* param, void* tmrId) { if (pSyncNode->FpEqMsg != NULL) { int32_t code = pSyncNode->FpEqMsg(pSyncNode->msgcb, &rpcMsg); if (code != 0) { - sError("vgId:%d sync enqueue elect msg error, code:%d", pSyncNode->vgId, code); + sError("vgId:%d, sync enqueue elect msg error, code:%d", pSyncNode->vgId, code); rpcFreeCont(rpcMsg.pCont); syncTimeoutDestroy(pSyncMsg); return; @@ -1929,7 +1930,7 @@ static void syncNodeEqHeartbeatTimer(void* param, void* tmrId) { if (pSyncNode->FpEqMsg != NULL) { int32_t code = pSyncNode->FpEqMsg(pSyncNode->msgcb, &rpcMsg); if (code != 0) { - sError("vgId:%d sync enqueue timer msg error, code:%d", pSyncNode->vgId, code); + sError("vgId:%d, sync enqueue timer msg error, code:%d", pSyncNode->vgId, code); rpcFreeCont(rpcMsg.pCont); syncTimeoutDestroy(pSyncMsg); return; @@ -2129,12 +2130,12 @@ const char* syncStr(ESyncState state) { static int32_t syncDoLeaderTransfer(SSyncNode* ths, SRpcMsg* pRpcMsg, SSyncRaftEntry* pEntry) { SyncLeaderTransfer* pSyncLeaderTransfer = syncLeaderTransferFromRpcMsg2(pRpcMsg); - sDebug("vgId:%d sync event %s commitIndex:%ld currentTerm:%lu begin leader transfer", ths->vgId, + sDebug("vgId:%d, sync event %s commitIndex:%ld currentTerm:%lu begin leader transfer", ths->vgId, syncUtilState2String(ths->state), ths->commitIndex, ths->pRaftStore->currentTerm); if (strcmp(pSyncLeaderTransfer->newNodeInfo.nodeFqdn, ths->myNodeInfo.nodeFqdn) == 0 && pSyncLeaderTransfer->newNodeInfo.nodePort == ths->myNodeInfo.nodePort) { - sDebug("vgId:%d sync event %s commitIndex:%ld currentTerm:%lu maybe leader transfer to %s:%d %lu", ths->vgId, + sDebug("vgId:%d, sync event %s commitIndex:%ld currentTerm:%lu maybe leader transfer to %s:%d %lu", ths->vgId, syncUtilState2String(ths->state), ths->commitIndex, ths->pRaftStore->currentTerm, pSyncLeaderTransfer->newNodeInfo.nodeFqdn, pSyncLeaderTransfer->newNodeInfo.nodePort, pSyncLeaderTransfer->newLeaderId.addr); @@ -2275,7 +2276,7 @@ static int32_t syncNodeConfigChange(SSyncNode* ths, SRpcMsg* pRpcMsg, SSyncRaftE int32_t syncNodeCommit(SSyncNode* ths, SyncIndex beginIndex, SyncIndex endIndex, uint64_t flag) { int32_t code = 0; ESyncState state = flag; - sDebug("vgId:%d sync event %s commitIndex:%ld currentTerm:%lu commit by wal from index:%" PRId64 " to index:%" PRId64 + sDebug("vgId:%d, sync event %s commitIndex:%ld currentTerm:%lu commit by wal from index:%" PRId64 " to index:%" PRId64 ", %s", ths->vgId, syncUtilState2String(ths->state), ths->commitIndex, ths->pRaftStore->currentTerm, beginIndex, endIndex, syncUtilState2String(state)); @@ -2327,7 +2328,7 @@ int32_t syncNodeCommit(SSyncNode* ths, SyncIndex beginIndex, SyncIndex endIndex, ths->pFsm->FpRestoreFinishCb(ths->pFsm); } ths->restoreFinish = true; - sDebug("vgId:%d sync event %s commitIndex:%ld currentTerm:%lu restore finish, %s, index:%ld", ths->vgId, + sDebug("vgId:%d, sync event %s commitIndex:%ld currentTerm:%lu restore finish, %s, index:%ld", ths->vgId, syncUtilState2String(ths->state), ths->commitIndex, ths->pRaftStore->currentTerm, syncUtilState2String(ths->state), pEntry->index); } diff --git a/source/libs/sync/src/syncRaftLog.c b/source/libs/sync/src/syncRaftLog.c index 87dfef5fdd..8fff7ae6de 100644 --- a/source/libs/sync/src/syncRaftLog.c +++ b/source/libs/sync/src/syncRaftLog.c @@ -164,7 +164,7 @@ static int32_t raftLogAppendEntry(struct SSyncLogStore* pLogStore, SSyncRaftEntr walFsync(pWal, true); sDebug( - "vgId:%d sync event %s commitIndex:%ld currentTerm:%lu write index:%ld, isStandBy:%d, msgType:%s,%d, " + "vgId:%d, sync event %s commitIndex:%ld currentTerm:%lu write index:%ld, isStandBy:%d, msgType:%s,%d, " "originalRpcType:%s,%d", pData->pSyncNode->vgId, syncUtilState2String(pData->pSyncNode->state), pData->pSyncNode->commitIndex, pData->pSyncNode->pRaftStore->currentTerm, pEntry->index, pData->pSyncNode->pRaftCfg->isStandBy, @@ -325,7 +325,7 @@ int32_t logStoreAppendEntry(SSyncLogStore* pLogStore, SSyncRaftEntry* pEntry) { walFsync(pWal, true); sDebug( - "vgId:%d sync event %s commitIndex:%ld currentTerm:%lu old write index:%ld, isStandBy:%d, msgType:%s,%d, " + "vgId:%d, sync event %s commitIndex:%ld currentTerm:%lu old write index:%ld, isStandBy:%d, msgType:%s,%d, " "originalRpcType:%s,%d", pData->pSyncNode->vgId, syncUtilState2String(pData->pSyncNode->state), pData->pSyncNode->commitIndex, pData->pSyncNode->pRaftStore->currentTerm, pEntry->index, pData->pSyncNode->pRaftCfg->isStandBy, diff --git a/source/libs/sync/src/syncRespMgr.c b/source/libs/sync/src/syncRespMgr.c index 551c2611b0..831e286d76 100644 --- a/source/libs/sync/src/syncRespMgr.c +++ b/source/libs/sync/src/syncRespMgr.c @@ -47,7 +47,7 @@ int64_t syncRespMgrAdd(SSyncRespMgr *pObj, SRespStub *pStub) { SSyncNode *pSyncNode = pObj->data; sDebug( - "vgId:%d sync event %s commitIndex:%ld currentTerm:%lu resp mgr add, msgType:%s,%d seq:%lu handle:%p ahandle:%p", + "vgId:%d, sync event %s commitIndex:%ld currentTerm:%lu resp mgr add, msgType:%s,%d seq:%lu handle:%p ahandle:%p", pSyncNode->vgId, syncUtilState2String(pSyncNode->state), pSyncNode->commitIndex, pSyncNode->pRaftStore->currentTerm, TMSG_INFO(pStub->rpcMsg.msgType), pStub->rpcMsg.msgType, keyCode, pStub->rpcMsg.info.handle, pStub->rpcMsg.info.ahandle); @@ -74,7 +74,7 @@ int32_t syncRespMgrGet(SSyncRespMgr *pObj, uint64_t index, SRespStub *pStub) { SSyncNode *pSyncNode = pObj->data; sDebug( - "vgId:%d sync event %s commitIndex:%ld currentTerm:%lu resp mgr get, msgType:%s,%d seq:%lu handle:%p " + "vgId:%d, sync event %s commitIndex:%ld currentTerm:%lu resp mgr get, msgType:%s,%d seq:%lu handle:%p " "ahandle:%p", pSyncNode->vgId, syncUtilState2String(pSyncNode->state), pSyncNode->commitIndex, pSyncNode->pRaftStore->currentTerm, TMSG_INFO(pStub->rpcMsg.msgType), pStub->rpcMsg.msgType, index, @@ -96,7 +96,7 @@ int32_t syncRespMgrGetAndDel(SSyncRespMgr *pObj, uint64_t index, SRespStub *pStu SSyncNode *pSyncNode = pObj->data; sDebug( - "vgId:%d sync event %s commitIndex:%ld currentTerm:%lu resp mgr get and del, msgType:%s,%d seq:%lu handle:%p " + "vgId:%d, sync event %s commitIndex:%ld currentTerm:%lu resp mgr get and del, msgType:%s,%d seq:%lu handle:%p " "ahandle:%p", pSyncNode->vgId, syncUtilState2String(pSyncNode->state), pSyncNode->commitIndex, pSyncNode->pRaftStore->currentTerm, TMSG_INFO(pStub->rpcMsg.msgType), pStub->rpcMsg.msgType, index, diff --git a/source/libs/sync/src/syncSnapshot.c b/source/libs/sync/src/syncSnapshot.c index 4973d36295..7c8abfe494 100644 --- a/source/libs/sync/src/syncSnapshot.c +++ b/source/libs/sync/src/syncSnapshot.c @@ -141,7 +141,7 @@ void snapshotSenderStart(SSyncSnapshotSender *pSender) { if (gRaftDetailLog) { char *msgStr = syncSnapshotSend2Str(pMsg); sDebug( - "vgId:%d sync event %s commitIndex:%ld currentTerm:%lu snapshot send to %s:%d begin seq:%d ack:%d " + "vgId:%d, sync event %s commitIndex:%ld currentTerm:%lu snapshot send to %s:%d begin seq:%d ack:%d " "lastApplyIndex:%ld " "lastApplyTerm:%lu " "lastConfigIndex:%ld privateTerm:%lu send " @@ -153,7 +153,7 @@ void snapshotSenderStart(SSyncSnapshotSender *pSender) { taosMemoryFree(msgStr); } else { sDebug( - "vgId:%d sync event %s commitIndex:%ld currentTerm:%lu snapshot send to %s:%d begin seq:%d ack:%d " + "vgId:%d, sync event %s commitIndex:%ld currentTerm:%lu snapshot send to %s:%d begin seq:%d ack:%d " "lastApplyIndex:%ld " "lastApplyTerm:%lu " "lastConfigIndex:%ld privateTerm:%lu", @@ -287,7 +287,7 @@ int32_t snapshotSend(SSyncSnapshotSender *pSender) { if (gRaftDetailLog) { char *msgStr = syncSnapshotSend2Str(pMsg); sDebug( - "vgId:%d sync event %s commitIndex:%ld currentTerm:%lu snapshot send to %s:%d finish seq:%d ack:%d " + "vgId:%d, sync event %s commitIndex:%ld currentTerm:%lu snapshot send to %s:%d finish seq:%d ack:%d " "lastApplyIndex:%ld " "lastApplyTerm:%lu " "lastConfigIndex:%ld privateTerm:%lu send " @@ -299,7 +299,7 @@ int32_t snapshotSend(SSyncSnapshotSender *pSender) { taosMemoryFree(msgStr); } else { sDebug( - "vgId:%d sync event %s commitIndex:%ld currentTerm:%lu snapshot send to %s:%d finish seq:%d ack:%d " + "vgId:%d, sync event %s commitIndex:%ld currentTerm:%lu snapshot send to %s:%d finish seq:%d ack:%d " "lastApplyIndex:%ld " "lastApplyTerm:%lu " "lastConfigIndex:%ld privateTerm:%lu", @@ -310,7 +310,7 @@ int32_t snapshotSend(SSyncSnapshotSender *pSender) { } } else { sDebug( - "vgId:%d sync event %s commitIndex:%ld currentTerm:%lu snapshot send to %s:%d sending seq:%d ack:%d " + "vgId:%d, sync event %s commitIndex:%ld currentTerm:%lu snapshot send to %s:%d sending seq:%d ack:%d " "lastApplyIndex:%ld " "lastApplyTerm:%lu " "lastConfigIndex:%ld privateTerm:%lu", @@ -349,7 +349,7 @@ int32_t snapshotReSend(SSyncSnapshotSender *pSender) { if (gRaftDetailLog) { char *msgStr = syncSnapshotSend2Str(pMsg); sDebug( - "vgId:%d sync event %s commitIndex:%ld currentTerm:%lu snapshot send to %s:%d resend seq:%d ack:%d " + "vgId:%d, sync event %s commitIndex:%ld currentTerm:%lu snapshot send to %s:%d resend seq:%d ack:%d " "privateTerm:%lu send " "msg:%s", pSender->pSyncNode->vgId, syncUtilState2String(pSender->pSyncNode->state), pSender->pSyncNode->commitIndex, @@ -358,7 +358,7 @@ int32_t snapshotReSend(SSyncSnapshotSender *pSender) { taosMemoryFree(msgStr); } else { sDebug( - "vgId:%d sync event %s commitIndex:%ld currentTerm:%lu snapshot send to %s:%d resend seq:%d ack:%d " + "vgId:%d, sync event %s commitIndex:%ld currentTerm:%lu snapshot send to %s:%d resend seq:%d ack:%d " "privateTerm:%lu", pSender->pSyncNode->vgId, syncUtilState2String(pSender->pSyncNode->state), pSender->pSyncNode->commitIndex, pSender->pSyncNode->pRaftStore->currentTerm, host, port, pSender->seq, pSender->ack, pSender->privateTerm); @@ -594,7 +594,7 @@ int32_t syncNodeOnSnapshotSendCb(SSyncNode *pSyncNode, SyncSnapshotSend *pMsg) { if (gRaftDetailLog) { char *msgStr = syncSnapshotSend2Str(pMsg); sDebug( - "vgId:%d sync event %s commitIndex:%ld currentTerm:%lu snapshot recv from %s:%d begin ack:%d, " + "vgId:%d, sync event %s commitIndex:%ld currentTerm:%lu snapshot recv from %s:%d begin ack:%d, " "lastIndex:%ld, " "lastTerm:%lu, " "lastConfigIndex:%ld, privateTerm:%lu, recv msg:%s", @@ -604,7 +604,7 @@ int32_t syncNodeOnSnapshotSendCb(SSyncNode *pSyncNode, SyncSnapshotSend *pMsg) { taosMemoryFree(msgStr); } else { sDebug( - "vgId:%d sync event %s commitIndex:%ld currentTerm:%lu snapshot recv from %s:%d begin ack:%d, " + "vgId:%d, sync event %s commitIndex:%ld currentTerm:%lu snapshot recv from %s:%d begin ack:%d, " "lastIndex:%ld, " "lastTerm:%lu, " "lastConfigIndex:%ld privateTerm:%lu", @@ -648,7 +648,7 @@ int32_t syncNodeOnSnapshotSendCb(SSyncNode *pSyncNode, SyncSnapshotSend *pMsg) { bool isDrop; if (IamInNew) { sDebug( - "vgId:%d sync event %s commitIndex:%ld currentTerm:%lu update config by snapshot, lastIndex:%ld, " + "vgId:%d, sync event %s commitIndex:%ld currentTerm:%lu update config by snapshot, lastIndex:%ld, " "lastTerm:%lu, " "lastConfigIndex:%ld ", pSyncNode->vgId, syncUtilState2String(pSyncNode->state), pSyncNode->commitIndex, @@ -656,7 +656,7 @@ int32_t syncNodeOnSnapshotSendCb(SSyncNode *pSyncNode, SyncSnapshotSend *pMsg) { syncNodeUpdateConfig(pSyncNode, &newSyncCfg, pMsg->lastConfigIndex, &isDrop); } else { sDebug( - "vgId:%d sync event %s commitIndex:%ld currentTerm:%lu do not update config by snapshot, I am not in " + "vgId:%d, sync event %s commitIndex:%ld currentTerm:%lu do not update config by snapshot, I am not in " "newCfg, " "lastIndex:%ld, lastTerm:%lu, " "lastConfigIndex:%ld ", @@ -694,7 +694,7 @@ int32_t syncNodeOnSnapshotSendCb(SSyncNode *pSyncNode, SyncSnapshotSend *pMsg) { if (gRaftDetailLog) { char *logSimpleStr = logStoreSimple2Str(pSyncNode->pLogStore); sDebug( - "vgId:%d sync event %s commitIndex:%ld currentTerm:%lu snapshot recv from %s:%d finish, update log begin " + "vgId:%d, sync event %s commitIndex:%ld currentTerm:%lu snapshot recv from %s:%d finish, update log begin " "index:%ld, " "snapshot.lastApplyIndex:%ld, " "snapshot.lastApplyTerm:%lu, snapshot.lastConfigIndex:%ld, privateTerm:%lu, raft log:%s", @@ -704,7 +704,7 @@ int32_t syncNodeOnSnapshotSendCb(SSyncNode *pSyncNode, SyncSnapshotSend *pMsg) { taosMemoryFree(logSimpleStr); } else { sDebug( - "vgId:%d sync event %s commitIndex:%ld currentTerm:%lu snapshot recv from %s:%d finish, update log begin " + "vgId:%d, sync event %s commitIndex:%ld currentTerm:%lu snapshot recv from %s:%d finish, update log begin " "index:%ld, " "snapshot.lastApplyIndex:%ld, " "snapshot.lastApplyTerm:%lu, snapshot.lastConfigIndex:%ld, privateTerm:%lu", @@ -721,7 +721,7 @@ int32_t syncNodeOnSnapshotSendCb(SSyncNode *pSyncNode, SyncSnapshotSend *pMsg) { if (gRaftDetailLog) { char *msgStr = syncSnapshotSend2Str(pMsg); sDebug( - "vgId:%d sync event %s commitIndex:%ld currentTerm:%lu snapshot recv from %s:%d end ack:%d, " + "vgId:%d, sync event %s commitIndex:%ld currentTerm:%lu snapshot recv from %s:%d end ack:%d, " "lastIndex:%ld, lastTerm:%lu, " "lastConfigIndex:%ld, privateTerm:%lu, recv msg:%s", pReceiver->pSyncNode->vgId, syncUtilState2String(pSyncNode->state), pSyncNode->commitIndex, @@ -730,7 +730,7 @@ int32_t syncNodeOnSnapshotSendCb(SSyncNode *pSyncNode, SyncSnapshotSend *pMsg) { taosMemoryFree(msgStr); } else { sDebug( - "vgId:%d sync event %s commitIndex:%ld currentTerm:%lu snapshot recv from %s:%d end ack:%d, " + "vgId:%d, sync event %s commitIndex:%ld currentTerm:%lu snapshot recv from %s:%d end ack:%d, " "lastIndex:%ld, lastTerm:%lu, " "lastConfigIndex:%ld, privateTerm:%lu", pReceiver->pSyncNode->vgId, syncUtilState2String(pSyncNode->state), pSyncNode->commitIndex, @@ -750,7 +750,7 @@ int32_t syncNodeOnSnapshotSendCb(SSyncNode *pSyncNode, SyncSnapshotSend *pMsg) { if (gRaftDetailLog) { char *msgStr = syncSnapshotSend2Str(pMsg); sDebug( - "vgId:%d sync event %s commitIndex:%ld currentTerm:%lu snapshot recv from %s:%d force close ack:%d, " + "vgId:%d, sync event %s commitIndex:%ld currentTerm:%lu snapshot recv from %s:%d force close ack:%d, " "lastIndex:%ld, " "lastTerm:%lu, " "lastConfigIndex:%ld, privateTerm:%lu, recv " @@ -761,7 +761,7 @@ int32_t syncNodeOnSnapshotSendCb(SSyncNode *pSyncNode, SyncSnapshotSend *pMsg) { taosMemoryFree(msgStr); } else { sDebug( - "vgId:%d sync event %s commitIndex:%ld currentTerm:%lu snapshot recv from %s:%d force close ack:%d, " + "vgId:%d, sync event %s commitIndex:%ld currentTerm:%lu snapshot recv from %s:%d force close ack:%d, " "lastIndex:%ld, " "lastTerm:%lu, " "lastConfigIndex:%ld, privateTerm:%lu", @@ -787,7 +787,7 @@ int32_t syncNodeOnSnapshotSendCb(SSyncNode *pSyncNode, SyncSnapshotSend *pMsg) { if (gRaftDetailLog) { char *msgStr = syncSnapshotSend2Str(pMsg); sDebug( - "vgId:%d sync event %s commitIndex:%ld currentTerm:%lu snapshot recv from %s:%d receiving ack:%d, " + "vgId:%d, sync event %s commitIndex:%ld currentTerm:%lu snapshot recv from %s:%d receiving ack:%d, " "lastIndex:%ld, " "lastTerm:%lu, " "lastConfigIndex:%ld, privateTerm:%lu, recv msg:%s", @@ -797,7 +797,7 @@ int32_t syncNodeOnSnapshotSendCb(SSyncNode *pSyncNode, SyncSnapshotSend *pMsg) { taosMemoryFree(msgStr); } else { sDebug( - "vgId:%d sync event %s commitIndex:%ld currentTerm:%lu snapshot recv from %s:%d receiving ack:%d, " + "vgId:%d, sync event %s commitIndex:%ld currentTerm:%lu snapshot recv from %s:%d receiving ack:%d, " "lastIndex:%ld, " "lastTerm:%lu, " "lastConfigIndex:%ld, privateTerm:%lu", diff --git a/source/os/src/osSocket.c b/source/os/src/osSocket.c index 4a0d9e2866..4d61e7036d 100644 --- a/source/os/src/osSocket.c +++ b/source/os/src/osSocket.c @@ -946,8 +946,7 @@ int32_t taosGetFqdn(char *fqdn) { #endif // __APPLE__ int32_t ret = getaddrinfo(hostname, NULL, &hints, &result); if (!result) { - // printf("failed to get fqdn, code:%d, reason:%s", ret, gai_strerror(ret)); - assert(0); + fprintf(stderr,"failed to get fqdn, code:%d, reason:%s", ret, gai_strerror(ret)); return -1; } diff --git a/source/util/src/tenv.c b/source/util/src/tenv.c index e717e82c5b..4fc0542816 100644 --- a/source/util/src/tenv.c +++ b/source/util/src/tenv.c @@ -72,6 +72,9 @@ int32_t taosEnvToCfg(const char *envStr, char *cfgStr) { if (cfgNameLen > 0) { memcpy(cfgStr, buf, cfgNameLen); memset(&cfgStr[cfgNameLen], ' ', p - cfgStr - cfgNameLen + 1); + } else { + *cfgStr = '\0'; + return -1; } } return strlen(cfgStr); diff --git a/source/util/src/terror.c b/source/util/src/terror.c index 933e3ba92b..7976cb09af 100644 --- a/source/util/src/terror.c +++ b/source/util/src/terror.c @@ -132,6 +132,7 @@ TAOS_DEFINE_ERROR(TSDB_CODE_TSC_INVALID_INPUT, "Invalid tsc input") TAOS_DEFINE_ERROR(TSDB_CODE_TSC_STMT_API_ERROR, "Stmt API usage error") TAOS_DEFINE_ERROR(TSDB_CODE_TSC_STMT_TBNAME_ERROR, "Stmt table name not set") TAOS_DEFINE_ERROR(TSDB_CODE_TSC_STMT_CLAUSE_ERROR, "not supported stmt clause") +TAOS_DEFINE_ERROR(TSDB_CODE_TSC_QUERY_KILLED, "Query killed") // mnode-common TAOS_DEFINE_ERROR(TSDB_CODE_MND_APP_ERROR, "Mnode internal error") @@ -455,6 +456,7 @@ TAOS_DEFINE_ERROR(TSDB_CODE_CTG_VG_META_MISMATCH, "table meta and vgroup TAOS_DEFINE_ERROR(TSDB_CODE_SCH_STATUS_ERROR, "scheduler status error") TAOS_DEFINE_ERROR(TSDB_CODE_SCH_INTERNAL_ERROR, "scheduler internal error") TAOS_DEFINE_ERROR(TSDB_CODE_SCH_TIMEOUT_ERROR, "Task timeout") +TAOS_DEFINE_ERROR(TSDB_CODE_SCH_JOB_IS_DROPPING, "Job is dropping") TAOS_DEFINE_ERROR(TSDB_CODE_QW_MSG_ERROR, "Invalid msg order") // parser @@ -577,9 +579,14 @@ TAOS_DEFINE_ERROR(TSDB_CODE_TSMA_NO_INDEX_IN_CACHE, "No tsma index in ca TAOS_DEFINE_ERROR(TSDB_CODE_RSMA_INVALID_ENV, "Invalid rsma env") TAOS_DEFINE_ERROR(TSDB_CODE_RSMA_INVALID_STAT, "Invalid rsma state") +//tq +TAOS_DEFINE_ERROR(TSDB_CODE_TQ_NO_COMMITTED_OFFSET, "No committed offset") + TAOS_DEFINE_ERROR(TSDB_CODE_INDEX_REBUILDING, "Index is rebuilding") +TAOS_DEFINE_ERROR(TSDB_CODE_TMQ_INVALID_MSG, "Invalid message") + #ifdef TAOS_ERROR_C }; #endif diff --git a/tests/script/sh/copy_udf.bat b/tests/script/sh/copy_udf.bat new file mode 100644 index 0000000000..5144cf8d25 --- /dev/null +++ b/tests/script/sh/copy_udf.bat @@ -0,0 +1,23 @@ +@echo off + +echo Executing copy_udf.bat +set SCRIPT_DIR=%cd% +echo SCRIPT_DIR: %SCRIPT_DIR% + +cd ..\..\.. +set TAOS_DIR=%cd% +echo find udf library in %TAOS_DIR% +set UDF1_DIR=%TAOS_DIR%\debug\build\lib\udf1.dll +set UDF2_DIR=%TAOS_DIR%\debug\build\lib\udf2.dll + +echo %UDF1_DIR% +echo %UDF2_DIR% + +set UDF_TMP=C:\Windows\Temp\udf +rm -rf %UDF_TMP% +mkdir %UDF_TMP% + +echo Copy udf shared library files to %UDF_TMP% + +cp %UDF1_DIR% %UDF_TMP% +cp %UDF2_DIR% %UDF_TMP% diff --git a/tests/script/tsim/db/alter_option.sim b/tests/script/tsim/db/alter_option.sim index 4351ee5cb1..fede960a6a 100644 --- a/tests/script/tsim/db/alter_option.sim +++ b/tests/script/tsim/db/alter_option.sim @@ -92,7 +92,7 @@ endi if $data5_db != no_strict then # strict return -1 endi -if $data6_db != 345600 then # duration +if $data6_db != 345600m then # duration return -1 endi if $data7_db != 1440000m,1440000m,1440000m then # keep diff --git a/tests/script/tsim/db/basic6.sim b/tests/script/tsim/db/basic6.sim index 64103b5dac..7525fe2087 100644 --- a/tests/script/tsim/db/basic6.sim +++ b/tests/script/tsim/db/basic6.sim @@ -34,7 +34,7 @@ endi if $data24 != 1 then return -1 endi -if $data26 != 2880 then +if $data26 != 2880m then return -1 endi if $data27 != 14400m,14400m,14400m then @@ -78,7 +78,7 @@ endi if $data24 != 1 then return -1 endi -if $data26 != 21600 then +if $data26 != 21600m then return -1 endi diff --git a/tests/script/tsim/db/create_all_options.sim b/tests/script/tsim/db/create_all_options.sim index 284875ee08..efe7ff99cf 100644 --- a/tests/script/tsim/db/create_all_options.sim +++ b/tests/script/tsim/db/create_all_options.sim @@ -113,7 +113,7 @@ endi if $data5_db != no_strict then # strict return -1 endi -if $data6_db != 14400 then # duration +if $data6_db != 14400m then # duration return -1 endi if $data7_db != 5256000m,5256000m,5256000m then # keep diff --git a/tests/script/tsim/query/udf.sim b/tests/script/tsim/query/udf.sim index 93cae4e391..d9821a0495 100644 --- a/tests/script/tsim/query/udf.sim +++ b/tests/script/tsim/query/udf.sim @@ -19,8 +19,14 @@ sql show databases; sql create table t (ts timestamp, f int); sql insert into t values(now, 1)(now+1s, 2); -sql create function udf1 as '/tmp/udf/libudf1.so' outputtype int bufSize 8; -sql create aggregate function udf2 as '/tmp/udf/libudf2.so' outputtype double bufSize 8; +system_content printf %OS% +if $system_content == Windows_NT then + sql create function udf1 as 'C:\\Windows\\Temp\\udf1.dll' outputtype int bufSize 8; + sql create aggregate function udf2 as 'C:\\Windows\\Temp\\udf2.dll' outputtype double bufSize 8; +else + sql create function udf1 as '/tmp/udf/libudf1.so' outputtype int bufSize 8; + sql create aggregate function udf2 as '/tmp/udf/libudf2.so' outputtype double bufSize 8; +endi sql show functions; if $rows != 2 then return -1 diff --git a/tests/script/tsim/show/basic.sim b/tests/script/tsim/show/basic.sim index 95201bc48e..f23d75d78b 100644 --- a/tests/script/tsim/show/basic.sim +++ b/tests/script/tsim/show/basic.sim @@ -98,7 +98,7 @@ if $rows != 1 then endi #sql select * from information_schema.`streams` sql select * from information_schema.user_tables -if $rows != 28 then +if $rows != 29 then return -1 endi #sql select * from information_schema.user_table_distributed @@ -196,7 +196,7 @@ if $rows != 1 then endi #sql select * from performance_schema.`streams` sql select * from information_schema.user_tables -if $rows != 28 then +if $rows != 29 then return -1 endi #sql select * from information_schema.user_table_distributed @@ -210,4 +210,4 @@ if $rows != 3 then endi system sh/exec.sh -n dnode1 -s stop -x SIGINT -system sh/exec.sh -n dnode2 -s stop -x SIGINT \ No newline at end of file +system sh/exec.sh -n dnode2 -s stop -x SIGINT diff --git a/tests/system-test/7-tmq/subscribeDb1.py b/tests/system-test/7-tmq/subscribeDb1.py index ed92a429ae..c08c7d3dae 100644 --- a/tests/system-test/7-tmq/subscribeDb1.py +++ b/tests/system-test/7-tmq/subscribeDb1.py @@ -442,7 +442,7 @@ class TDTestCase: showRow = 1 self.startTmqSimProcess(buildPath,cfgPath,pollDelay,parameterDict["dbName"],showMsg, showRow) - time.sleep(3) + time.sleep(6) tdLog.info("pkill consume processor") if (platform.system().lower() == 'windows'): os.system("TASKKILL /F /IM tmq_sim.exe") diff --git a/tests/test/c/tmqDemo.c b/tests/test/c/tmqDemo.c index 5a972e071a..02fd3c1396 100644 --- a/tests/test/c/tmqDemo.c +++ b/tests/test/c/tmqDemo.c @@ -353,8 +353,8 @@ tmq_list_t* build_topic_list() { void sync_consume_loop(tmq_t* tmq, tmq_list_t* topics) { static const int MIN_COMMIT_COUNT = 1000; - int msg_count = 0; - tmq_resp_err_t err; + int msg_count = 0; + int32_t err; if ((err = tmq_subscribe(tmq, topics))) { fprintf(stderr, "%% Failed to start consuming topics: %s\n", tmq_err2str(err)); @@ -379,7 +379,7 @@ void sync_consume_loop(tmq_t* tmq, tmq_list_t* topics) { } void perf_loop(tmq_t* tmq, tmq_list_t* topics, int32_t totalMsgs, int64_t walLogSize) { - tmq_resp_err_t err; + int32_t err; if ((err = tmq_subscribe(tmq, topics))) { fprintf(stderr, "%% Failed to start consuming topics: %s\n", tmq_err2str(err)); diff --git a/tests/test/c/tmqSim.c b/tests/test/c/tmqSim.c index 8455bd9890..0f78a003d6 100644 --- a/tests/test/c/tmqSim.c +++ b/tests/test/c/tmqSim.c @@ -62,7 +62,7 @@ typedef struct { tmq_t* tmq; tmq_list_t* topicList; - + int32_t numOfVgroups; int32_t rowsOfPerVgroups[MAX_VGROUP_CNT][2]; // [i][0]: vgroup id, [i][1]: rows of consume int64_t ts; @@ -74,7 +74,7 @@ typedef struct { char cdbName[32]; char dbName[32]; int32_t showMsgFlag; - int32_t showRowFlag; + int32_t showRowFlag; int32_t saveRowFlag; int32_t consumeDelay; // unit s int32_t numOfThread; @@ -108,26 +108,20 @@ static void printHelp() { } char* getCurrentTimeString(char* timeString) { - time_t tTime = taosGetTimestampSec(); + time_t tTime = taosGetTimestampSec(); struct tm tm = *taosLocalTime(&tTime, NULL); - sprintf(timeString, "%d-%02d-%02d %02d:%02d:%02d", - tm.tm_year + 1900, - tm.tm_mon + 1, - tm.tm_mday, - tm.tm_hour, - tm.tm_min, - tm.tm_sec); + sprintf(timeString, "%d-%02d-%02d %02d:%02d:%02d", tm.tm_year + 1900, tm.tm_mon + 1, tm.tm_mday, tm.tm_hour, + tm.tm_min, tm.tm_sec); return timeString; } - void initLogFile() { char filename[256]; - char tmpString[128]; + char tmpString[128]; - sprintf(filename,"%s/../log/tmqlog_%s.txt", configDir, getCurrentTimeString(tmpString)); - //sprintf(filename, "%s/../log/tmqlog.txt", configDir); + sprintf(filename, "%s/../log/tmqlog_%s.txt", configDir, getCurrentTimeString(tmpString)); + // sprintf(filename, "%s/../log/tmqlog.txt", configDir); #ifdef WINDOWS for (int i = 2; i < sizeof(filename); i++) { if (filename[i] == ':') filename[i] = '-'; @@ -249,17 +243,18 @@ void addRowsToVgroupId(SThreadInfo* pInfo, int32_t vgroupId, int32_t rows) { for (i = 0; i < pInfo->numOfVgroups; i++) { if (vgroupId == pInfo->rowsOfPerVgroups[i][0]) { pInfo->rowsOfPerVgroups[i][1] += rows; - return; - } + return; + } } pInfo->rowsOfPerVgroups[pInfo->numOfVgroups][0] = vgroupId; pInfo->rowsOfPerVgroups[pInfo->numOfVgroups][1] += rows; pInfo->numOfVgroups++; - + taosFprintfFile(g_fp, "consume id %d, add one new vogroup id: %d\n", pInfo->consumerId, vgroupId); if (pInfo->numOfVgroups > MAX_VGROUP_CNT) { - taosFprintfFile(g_fp, "====consume id %d, vgroup num %d over than 32. new vgroupId: %d\n", pInfo->consumerId, pInfo->numOfVgroups, vgroupId); + taosFprintfFile(g_fp, "====consume id %d, vgroup num %d over than 32. new vgroupId: %d\n", pInfo->consumerId, + pInfo->numOfVgroups, vgroupId); taosCloseFile(&g_fp); exit(-1); } @@ -277,7 +272,8 @@ int32_t saveConsumeContentToTbl(SThreadInfo* pInfo, char* buf) { TAOS* pConn = taos_connect(NULL, "root", "taosdata", NULL, 0); assert(pConn != NULL); - sprintf(sqlStr, "insert into %s.content_%d values (%"PRId64", \'%s\')", g_stConfInfo.cdbName, pInfo->consumerId, pInfo->ts++, buf); + sprintf(sqlStr, "insert into %s.content_%d values (%" PRId64 ", \'%s\')", g_stConfInfo.cdbName, pInfo->consumerId, + pInfo->ts++, buf); TAOS_RES* pRes = taos_query(pConn, sqlStr); if (taos_errno(pRes) != 0) { pError("error in insert consume result, reason:%s\n", taos_errstr(pRes)); @@ -295,12 +291,13 @@ int32_t saveConsumeContentToTbl(SThreadInfo* pInfo, char* buf) { static int32_t msg_process(TAOS_RES* msg, SThreadInfo* pInfo, int32_t msgIndex) { char buf[1024]; int32_t totalRows = 0; - + // printf("topic: %s\n", tmq_get_topic_name(msg)); int32_t vgroupId = tmq_get_vgroup_id(msg); - + taosFprintfFile(g_fp, "msg index:%" PRId64 ", consumerId: %d\n", msgIndex, pInfo->consumerId); - //taosFprintfFile(g_fp, "topic: %s, vgroupId: %d, tableName: %s\n", tmq_get_topic_name(msg), vgroupId, tmq_get_table_name(msg)); + // taosFprintfFile(g_fp, "topic: %s, vgroupId: %d, tableName: %s\n", tmq_get_topic_name(msg), vgroupId, + // tmq_get_table_name(msg)); taosFprintfFile(g_fp, "topic: %s, vgroupId: %d\n", tmq_get_topic_name(msg), vgroupId); while (1) { @@ -316,9 +313,9 @@ static int32_t msg_process(TAOS_RES* msg, SThreadInfo* pInfo, int32_t msgIndex) const char* tbName = tmq_get_table_name(msg); if (0 != g_stConfInfo.showRowFlag) { - taosFprintfFile(g_fp, "tbname:%s, rows[%d]: %s\n", (tbName != NULL ? tbName:"null table"), totalRows, buf); - if (0 != g_stConfInfo.saveRowFlag) { - saveConsumeContentToTbl(pInfo, buf); + taosFprintfFile(g_fp, "tbname:%s, rows[%d]: %s\n", (tbName != NULL ? tbName : "null table"), totalRows, buf); + if (0 != g_stConfInfo.saveRowFlag) { + saveConsumeContentToTbl(pInfo, buf); } } @@ -326,7 +323,7 @@ static int32_t msg_process(TAOS_RES* msg, SThreadInfo* pInfo, int32_t msgIndex) } addRowsToVgroupId(pInfo, vgroupId, totalRows); - + return totalRows; } @@ -342,8 +339,8 @@ int queryDB(TAOS* taos, char* command) { return 0; } -static void tmq_commit_cb_print(tmq_t* tmq, tmq_resp_err_t resp, tmq_topic_vgroup_list_t* offsets, void* param) { - pError("tmq_commit_cb_print() commit %d\n", resp); +static void tmq_commit_cb_print(tmq_t* tmq, int32_t code, void* param) { + pError("tmq_commit_cb_print() commit %d\n", code); } void build_consumer(SThreadInfo* pInfo) { @@ -401,16 +398,11 @@ int32_t saveConsumeResult(SThreadInfo* pInfo) { int64_t now = taosGetTimestampMs(); // schema: ts timestamp, consumerid int, consummsgcnt bigint, checkresult int - sprintf(sqlStr, "insert into %s.consumeresult values (%"PRId64", %d, %" PRId64 ", %" PRId64 ", %d)", - g_stConfInfo.cdbName, - now, - pInfo->consumerId, - pInfo->consumeMsgCnt, - pInfo->consumeRowCnt, - pInfo->checkresult); + sprintf(sqlStr, "insert into %s.consumeresult values (%" PRId64 ", %d, %" PRId64 ", %" PRId64 ", %d)", + g_stConfInfo.cdbName, now, pInfo->consumerId, pInfo->consumeMsgCnt, pInfo->consumeRowCnt, pInfo->checkresult); char tmpString[128]; - taosFprintfFile(g_fp, "%s, consume id %d result: %s\n", getCurrentTimeString(tmpString), pInfo->consumerId ,sqlStr); + taosFprintfFile(g_fp, "%s, consume id %d result: %s\n", getCurrentTimeString(tmpString), pInfo->consumerId, sqlStr); TAOS_RES* pRes = taos_query(pConn, sqlStr); if (taos_errno(pRes) != 0) { @@ -421,7 +413,7 @@ int32_t saveConsumeResult(SThreadInfo* pInfo) { taos_free_result(pRes); - #if 0 +#if 0 // vgroups for (i = 0; i < pInfo->numOfVgroups; i++) { // schema: ts timestamp, consumerid int, consummsgcnt bigint, checkresult int @@ -445,19 +437,20 @@ int32_t saveConsumeResult(SThreadInfo* pInfo) { taos_free_result(pRes); } - #endif +#endif return 0; } void loop_consume(SThreadInfo* pInfo) { - tmq_resp_err_t err; + int32_t code; int64_t totalMsgs = 0; int64_t totalRows = 0; char tmpString[128]; - taosFprintfFile(g_fp, "%s consumer id %d start to loop pull msg\n", getCurrentTimeString(tmpString), pInfo->consumerId); + taosFprintfFile(g_fp, "%s consumer id %d start to loop pull msg\n", getCurrentTimeString(tmpString), + pInfo->consumerId); pInfo->ts = taosGetTimestampMs(); @@ -473,7 +466,7 @@ void loop_consume(SThreadInfo* pInfo) { totalMsgs++; if (totalRows >= pInfo->expectMsgCnt) { - char tmpString[128]; + char tmpString[128]; taosFprintfFile(g_fp, "%s over than expect rows, so break consume\n", getCurrentTimeString(tmpString)); break; } @@ -503,8 +496,8 @@ void* consumeThreadFunc(void* param) { return NULL; } - tmq_resp_err_t err = tmq_subscribe(pInfo->tmq, pInfo->topicList); - if (err) { + int32_t err = tmq_subscribe(pInfo->tmq, pInfo->topicList); + if (err != 0) { pError("tmq_subscribe() fail, reason: %s\n", tmq_err2str(err)); exit(-1); } @@ -519,17 +512,19 @@ void* consumeThreadFunc(void* param) { pPrint("tmq_commit() manual commit when consume end.\n"); /*tmq_commit(pInfo->tmq, NULL, 0);*/ tmq_commit_sync(pInfo->tmq, NULL); + taosFprintfFile(g_fp, "tmq_commit() manual commit over.\n"); + pPrint("tmq_commit() manual commit over.\n"); } err = tmq_unsubscribe(pInfo->tmq); - if (err) { + if (err != 0) { pError("tmq_unsubscribe() fail, reason: %s\n", tmq_err2str(err)); /*pInfo->consumeMsgCnt = -1;*/ /*return NULL;*/ } err = tmq_consumer_close(pInfo->tmq); - if (err) { + if (err != 0) { pError("tmq_consumer_close() fail, reason: %s\n", tmq_err2str(err)); /*exit(-1);*/ } diff --git a/tests/tsim/src/simExe.c b/tests/tsim/src/simExe.c index dbc3c2c460..f97a13d2c5 100644 --- a/tests/tsim/src/simExe.c +++ b/tests/tsim/src/simExe.c @@ -458,11 +458,17 @@ bool simExecuteSystemContentCmd(SScript *script, char *option) { char buf[4096] = {0}; char buf1[4096 + 512] = {0}; char filename[400] = {0}; - sprintf(filename, "%s/%s.tmp", simScriptDir, script->fileName); + sprintf(filename, "%s" TD_DIRSEP "%s.tmp", simScriptDir, script->fileName); +#ifdef WINDOWS + sprintf(buf, "cd %s && ", simScriptDir); + simVisuallizeOption(script, option, buf + strlen(buf)); + sprintf(buf1, "%s > %s 2>nul", buf, filename); +#else sprintf(buf, "cd %s; ", simScriptDir); simVisuallizeOption(script, option, buf + strlen(buf)); sprintf(buf1, "%s > %s 2>/dev/null", buf, filename); +#endif sprintf(script->system_exit_code, "%d", system(buf1)); simStoreSystemContentResult(script, filename); diff --git a/tests/tsim/src/simParse.c b/tests/tsim/src/simParse.c index 638c4a1ccb..5b6dda4dae 100644 --- a/tests/tsim/src/simParse.c +++ b/tests/tsim/src/simParse.c @@ -206,7 +206,7 @@ SScript *simParseScript(char *fileName) { for (int32_t i = 0; i < cmdlen; ++i) { if (buffer[i] == '\r' || buffer[i] == '\n') { - buffer[i] = ' '; + buffer[i] = '\0'; } } diff --git a/tools/shell/src/shellEngine.c b/tools/shell/src/shellEngine.c index 8ed0e9ddcf..1f29237d38 100644 --- a/tools/shell/src/shellEngine.c +++ b/tools/shell/src/shellEngine.c @@ -855,8 +855,7 @@ void shellGetGrantInfo() { if (code == TSDB_CODE_OPS_NOT_SUPPORT) { fprintf(stdout, "Server is Community Edition, %s\n\n", sinfo); } else { - fprintf(stderr, "Failed to check Server Edition, Reason:0x%04x:%s\n\n", taos_errno(shell.conn), - taos_errstr(shell.conn)); + fprintf(stderr, "Failed to check Server Edition, Reason:0x%04x:%s\n\n", code, taos_errstr(tres)); } return; }