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

fix: fix task retry and qnode fetch thread num issue

This commit is contained in:
dapan1121 2022-09-14 18:08:29 +08:00
parent 43bdaa6c19
commit c7efe17177
4 changed files with 35 additions and 21 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

14
source/common/src/tglobal.c
View File

@ -63,7 +63,7 @@ int32_t tsNumOfVnodeWriteThreads = 2;
int32_t tsNumOfVnodeSyncThreads = 2;
int32_t tsNumOfVnodeRsmaThreads = 2;
int32_t tsNumOfQnodeQueryThreads = 4;
int32_t tsNumOfQnodeFetchThreads = 4;
int32_t tsNumOfQnodeFetchThreads = 1;
int32_t tsNumOfSnodeSharedThreads = 2;
int32_t tsNumOfSnodeUniqueThreads = 2;
@ -385,9 +385,9 @@ static int32_t taosAddServerCfg(SConfig *pCfg) {
tsNumOfQnodeQueryThreads = TMAX(tsNumOfQnodeQueryThreads, 4);
if (cfgAddInt32(pCfg, "numOfQnodeQueryThreads", tsNumOfQnodeQueryThreads, 1, 1024, 0) != 0) return -1;
tsNumOfQnodeFetchThreads = tsNumOfCores / 2;
tsNumOfQnodeFetchThreads = TMAX(tsNumOfQnodeFetchThreads, 4);
if (cfgAddInt32(pCfg, "numOfQnodeFetchThreads", tsNumOfQnodeFetchThreads, 1, 1024, 0) != 0) return -1;
// tsNumOfQnodeFetchThreads = tsNumOfCores / 2;
// tsNumOfQnodeFetchThreads = TMAX(tsNumOfQnodeFetchThreads, 4);
// if (cfgAddInt32(pCfg, "numOfQnodeFetchThreads", tsNumOfQnodeFetchThreads, 1, 1024, 0) != 0) return -1;
tsNumOfSnodeSharedThreads = tsNumOfCores / 4;
tsNumOfSnodeSharedThreads = TRANGE(tsNumOfSnodeSharedThreads, 2, 4);
@ -527,6 +527,7 @@ static int32_t taosUpdateServerCfg(SConfig *pCfg) {
pItem->stype = stype;
}
/*
pItem = cfgGetItem(tsCfg, "numOfQnodeFetchThreads");
if (pItem != NULL && pItem->stype == CFG_STYPE_DEFAULT) {
tsNumOfQnodeFetchThreads = numOfCores / 2;
@ -534,6 +535,7 @@ static int32_t taosUpdateServerCfg(SConfig *pCfg) {
pItem->i32 = tsNumOfQnodeFetchThreads;
pItem->stype = stype;
}
*/
pItem = cfgGetItem(tsCfg, "numOfSnodeSharedThreads");
if (pItem != NULL && pItem->stype == CFG_STYPE_DEFAULT) {
@ -691,7 +693,7 @@ static int32_t taosSetServerCfg(SConfig *pCfg) {
tsNumOfVnodeSyncThreads = cfgGetItem(pCfg, "numOfVnodeSyncThreads")->i32;
tsNumOfVnodeRsmaThreads = cfgGetItem(pCfg, "numOfVnodeRsmaThreads")->i32;
tsNumOfQnodeQueryThreads = cfgGetItem(pCfg, "numOfQnodeQueryThreads")->i32;
tsNumOfQnodeFetchThreads = cfgGetItem(pCfg, "numOfQnodeFetchThreads")->i32;
// tsNumOfQnodeFetchThreads = cfgGetItem(pCfg, "numOfQnodeFetchThreads")->i32;
tsNumOfSnodeSharedThreads = cfgGetItem(pCfg, "numOfSnodeSharedThreads")->i32;
tsNumOfSnodeUniqueThreads = cfgGetItem(pCfg, "numOfSnodeUniqueThreads")->i32;
tsRpcQueueMemoryAllowed = cfgGetItem(pCfg, "rpcQueueMemoryAllowed")->i64;
@ -939,8 +941,10 @@ int32_t taosSetCfg(SConfig *pCfg, char *name) {
tsNumOfVnodeRsmaThreads = cfgGetItem(pCfg, "numOfVnodeRsmaThreads")->i32;
} else if (strcasecmp("numOfQnodeQueryThreads", name) == 0) {
tsNumOfQnodeQueryThreads = cfgGetItem(pCfg, "numOfQnodeQueryThreads")->i32;
/*
} else if (strcasecmp("numOfQnodeFetchThreads", name) == 0) {
tsNumOfQnodeFetchThreads = cfgGetItem(pCfg, "numOfQnodeFetchThreads")->i32;
*/
} else if (strcasecmp("numOfSnodeSharedThreads", name) == 0) {
tsNumOfSnodeSharedThreads = cfgGetItem(pCfg, "numOfSnodeSharedThreads")->i32;
} else if (strcasecmp("numOfSnodeUniqueThreads", name) == 0) {

1
source/libs/scheduler/inc/schInt.h
View File

@ -285,6 +285,7 @@ typedef struct SSchJob {
typedef struct SSchTaskCtx {
int64_t jobRid;
SSchTask *pTask;
bool asyncLaunch;
} SSchTaskCtx;
extern SSchedulerMgmt schMgmt;

35
source/libs/scheduler/src/schTask.c
View File

@ -138,11 +138,6 @@ int32_t schUpdateTaskExecNode(SSchJob *pJob, SSchTask *pTask, void *handle, int3
return TSDB_CODE_SUCCESS;
}
if ((execId != pTask->execId) || pTask->waitRetry) { // ignore it
SCH_TASK_DLOG("handle not updated since execId %d is already not current execId %d, waitRetry %d", execId, pTask->execId, pTask->waitRetry);
return TSDB_CODE_SUCCESS;
}
SSchNodeInfo *nodeInfo = taosHashGet(pTask->execNodes, &execId, sizeof(execId));
if (NULL == nodeInfo) { // ignore it
SCH_TASK_DLOG("handle not updated since execId %d already not exist, current execId %d, waitRetry %d", execId, pTask->execId, pTask->waitRetry);
@ -161,10 +156,15 @@ int32_t schUpdateTaskHandle(SSchJob *pJob, SSchTask *pTask, bool dropExecNode, v
SCH_RET(schDropTaskExecNode(pJob, pTask, handle, execId));
}
SCH_SET_TASK_HANDLE(pTask, handle);
schUpdateTaskExecNode(pJob, pTask, handle, execId);
if ((execId != pTask->execId) || pTask->waitRetry) { // ignore it
SCH_TASK_DLOG("handle not updated since execId %d is already not current execId %d, waitRetry %d", execId, pTask->execId, pTask->waitRetry);
SCH_ERR_RET(TSDB_CODE_SCH_IGNORE_ERROR);
}
SCH_SET_TASK_HANDLE(pTask, handle);
return TSDB_CODE_SUCCESS;
}
@ -349,7 +349,7 @@ int32_t schDoTaskRedirect(SSchJob *pJob, SSchTask *pTask, SDataBuf *pData, int32
pTask->waitRetry = true;
schDropTaskOnExecNode(pJob, pTask);
taosHashClear(pTask->execNodes);
SCH_ERR_JRET(schRemoveTaskFromExecList(pJob, pTask));
schRemoveTaskFromExecList(pJob, pTask);
schDeregisterTaskHb(pJob, pTask);
atomic_sub_fetch_32(&pTask->level->taskLaunchedNum, 1);
taosMemoryFreeClear(pTask->msg);
@ -593,7 +593,7 @@ int32_t schTaskCheckSetRetry(SSchJob *pJob, SSchTask *pTask, int32_t errCode, bo
int32_t schHandleTaskRetry(SSchJob *pJob, SSchTask *pTask) {
atomic_sub_fetch_32(&pTask->level->taskLaunchedNum, 1);
SCH_ERR_RET(schRemoveTaskFromExecList(pJob, pTask));
schRemoveTaskFromExecList(pJob, pTask);
SCH_SET_TASK_STATUS(pTask, JOB_TASK_STATUS_INIT);
if (SCH_TASK_NEED_FLOW_CTRL(pJob, pTask)) {
@ -739,8 +739,7 @@ _return:
int32_t schRemoveTaskFromExecList(SSchJob *pJob, SSchTask *pTask) {
int32_t code = taosHashRemove(pJob->execTasks, &pTask->taskId, sizeof(pTask->taskId));
if (code) {
SCH_TASK_ELOG("task failed to rm from execTask list, code:%x", code);
SCH_ERR_RET(TSDB_CODE_SCH_INTERNAL_ERROR);
SCH_TASK_WLOG("task already not in execTask list, code:%x", code);
}
return TSDB_CODE_SUCCESS;
@ -829,6 +828,11 @@ int32_t schLaunchTaskImpl(void *param) {
}
SSchTask *pTask = pCtx->pTask;
if (pCtx->asyncLaunch) {
SCH_LOCK_TASK(pTask);
}
int8_t status = 0;
int32_t code = 0;
@ -875,8 +879,6 @@ int32_t schLaunchTaskImpl(void *param) {
_return:
taosMemoryFree(param);
if (pJob->taskNum >= SCH_MIN_AYSNC_EXEC_NUM) {
if (code) {
code = schProcessOnTaskFailure(pJob, pTask, code);
@ -886,8 +888,14 @@ _return:
}
}
if (pCtx->asyncLaunch) {
SCH_UNLOCK_TASK(pTask);
}
schReleaseJob(pJob->refId);
taosMemoryFree(param);
SCH_RET(code);
}
@ -902,6 +910,7 @@ int32_t schAsyncLaunchTaskImpl(SSchJob *pJob, SSchTask *pTask) {
param->pTask = pTask;
if (pJob->taskNum >= SCH_MIN_AYSNC_EXEC_NUM) {
param->asyncLaunch = true;
taosAsyncExec(schLaunchTaskImpl, param, NULL);
} else {
SCH_ERR_RET(schLaunchTaskImpl(param));