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

fix(stream): wait for the task checkpoint before stop.

This commit is contained in:
Haojun Liao 2023-09-21 15:53:21 +08:00
parent e4073bd9a8
commit 2efd155adf
3 changed files with 24 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
source/dnode/vnode/src/tq/tqUtil.c
View File

@ -37,14 +37,19 @@ int32_t tqInitDataRsp(SMqDataRsp* pRsp, STqOffsetVal pOffset) {
}
void tqUpdateNodeStage(STQ* pTq, bool isLeader) {
SSyncState state = syncGetState(pTq->pVnode->sync);
SSyncState state = syncGetState(pTq->pVnode->sync);
SStreamMeta* pMeta = pTq->pStreamMeta;
tqInfo("vgId:%d update the meta stage:%"PRId64", prev:%"PRId64" leader:%d", pMeta->vgId, state.term, pMeta->stage, isLeader);
int64_t stage = pMeta->stage;
pMeta->stage = state.term;
pMeta->leader = isLeader;
if (isLeader) {
tqInfo("vgId:%d update meta stage:%" PRId64 ", prev:%" PRId64 " leader:%d, start to send Hb", pMeta->vgId,
state.term, stage, isLeader);
streamMetaStartHb(pMeta);
} else {
tqInfo("vgId:%d update meta stage:%" PRId64 " prev:%" PRId64 " leader:%d", pMeta->vgId, state.term, stage,
isLeader);
}
}

2
source/dnode/vnode/src/vnd/vnodeSync.c
View File

@ -612,10 +612,10 @@ static void vnodeBecomeLearner(const SSyncFSM *pFsm) {
static void vnodeBecomeLeader(const SSyncFSM *pFsm) {
SVnode *pVnode = pFsm->data;
vDebug("vgId:%d, become leader", pVnode->config.vgId);
if (pVnode->pTq) {
tqUpdateNodeStage(pVnode->pTq, true);
}
vDebug("vgId:%d, become leader", pVnode->config.vgId);
}
static bool vnodeApplyQueueEmpty(const SSyncFSM *pFsm) {

17
source/libs/stream/src/streamTask.c
View File

@ -543,7 +543,22 @@ int32_t streamTaskStop(SStreamTask* pTask) {
int64_t st = taosGetTimestampMs();
const char* id = pTask->id.idStr;
pTask->status.taskStatus = TASK_STATUS__STOP;
// we should wait for the task complete the checkpoint operation before stop it, otherwise, the operation maybe blocked
// by the unfinished checkpoint operation, even if the leader has become the follower.
while(1) {
taosThreadMutexLock(&pTask->lock);
if (pTask->status.taskStatus == TASK_STATUS__CK) {
stDebug("s-task:%s in checkpoint, wait for it completed for 500ms before stop task", pTask->id.idStr);
taosThreadMutexUnlock(&pTask->lock);
taosMsleep(500);
} else {
pTask->status.taskStatus = TASK_STATUS__STOP;
taosThreadMutexUnlock(&pTask->lock);
break;
}
}
qKillTask(pTask->exec.pExecutor, TSDB_CODE_SUCCESS);
while (/*pTask->status.schedStatus != TASK_SCHED_STATUS__INACTIVE */ !streamTaskIsIdle(pTask)) {