From a0e68c2bfb038920ae1412470d14269fc66c276c Mon Sep 17 00:00:00 2001 From: Haojun Liao Date: Fri, 26 Apr 2024 11:14:53 +0800 Subject: [PATCH] refactor: do some internal refactor. --- include/libs/stream/tstream.h | 2 +- .../{streamStart.c => streamStartHistory.c} | 244 +---------- source/libs/stream/src/streamTask.c | 387 +----------------- 3 files changed, 7 insertions(+), 626 deletions(-) rename source/libs/stream/src/{streamStart.c => streamStartHistory.c} (69%) diff --git a/include/libs/stream/tstream.h b/include/libs/stream/tstream.h index 565f6b5938..260c1c54d2 100644 --- a/include/libs/stream/tstream.h +++ b/include/libs/stream/tstream.h @@ -850,7 +850,7 @@ int32_t streamTaskSendCheckpointReq(SStreamTask* pTask); int32_t streamTaskAddReqInfo(STaskCheckInfo* pInfo, int64_t reqId, int32_t taskId, const char* id); int32_t streamTaskUpdateCheckInfo(STaskCheckInfo* pInfo, int32_t taskId, int32_t status, int64_t rspTs, int64_t reqId, int32_t* pNotReady, const char* id); -void streamTaskCleanCheckInfo(STaskCheckInfo* pInfo); +void streamTaskCleanupCheckInfo(STaskCheckInfo* pInfo); int32_t streamTaskStartMonitorCheckRsp(SStreamTask* pTask); int32_t streamTaskStopMonitorCheckRsp(STaskCheckInfo* pInfo, const char* id); diff --git a/source/libs/stream/src/streamStart.c b/source/libs/stream/src/streamStartHistory.c similarity index 69% rename from source/libs/stream/src/streamStart.c rename to source/libs/stream/src/streamStartHistory.c index 90d0de5a0e..04a99feab0 100644 --- a/source/libs/stream/src/streamStart.c +++ b/source/libs/stream/src/streamStartHistory.c @@ -32,12 +32,12 @@ typedef struct SLaunchHTaskInfo { static int32_t streamSetParamForScanHistory(SStreamTask* pTask); static void streamTaskSetRangeStreamCalc(SStreamTask* pTask); static int32_t initScanHistoryReq(SStreamTask* pTask, SStreamScanHistoryReq* pReq, int8_t igUntreated); -static SLaunchHTaskInfo* createHTaskLaunchInfo(SStreamMeta* pMeta, STaskId* pTaskId, int64_t hStreamId, int32_t hTaskId); +static SLaunchHTaskInfo* createHTaskLaunchInfo(SStreamMeta* pMeta, STaskId* pTaskId, int64_t hStreamId, + int32_t hTaskId); static void tryLaunchHistoryTask(void* param, void* tmrId); -static void doProcessDownstreamReadyRsp(SStreamTask* pTask); -static void doExecScanhistoryInFuture(void* param, void* tmrId); -static int32_t doStartScanHistoryTask(SStreamTask* pTask); -static int32_t streamTaskStartScanHistory(SStreamTask* pTask); +static void doExecScanhistoryInFuture(void* param, void* tmrId); +static int32_t doStartScanHistoryTask(SStreamTask* pTask); +static int32_t streamTaskStartScanHistory(SStreamTask* pTask); int32_t streamTaskSetReady(SStreamTask* pTask) { int32_t numOfDowns = streamTaskGetNumOfDownstream(pTask); @@ -165,67 +165,6 @@ int32_t streamTaskStartScanHistory(SStreamTask* pTask) { return 0; } -// check status -void streamTaskCheckDownstream(SStreamTask* pTask) { - SDataRange* pRange = &pTask->dataRange; - STimeWindow* pWindow = &pRange->window; - - SStreamTaskCheckReq req = { - .streamId = pTask->id.streamId, - .upstreamTaskId = pTask->id.taskId, - .upstreamNodeId = pTask->info.nodeId, - .childId = pTask->info.selfChildId, - .stage = pTask->pMeta->stage, - }; - - ASSERT(pTask->status.downstreamReady == 0); - - // serialize streamProcessScanHistoryFinishRsp - if (pTask->outputInfo.type == TASK_OUTPUT__FIXED_DISPATCH) { - streamTaskStartMonitorCheckRsp(pTask); - - req.reqId = tGenIdPI64(); - req.downstreamNodeId = pTask->outputInfo.fixedDispatcher.nodeId; - req.downstreamTaskId = pTask->outputInfo.fixedDispatcher.taskId; - - streamTaskAddReqInfo(&pTask->taskCheckInfo, req.reqId, req.downstreamTaskId, pTask->id.idStr); - - stDebug("s-task:%s (vgId:%d) stage:%" PRId64 " check single downstream task:0x%x(vgId:%d) ver:%" PRId64 "-%" PRId64 - " window:%" PRId64 "-%" PRId64 " reqId:0x%" PRIx64, - pTask->id.idStr, pTask->info.nodeId, req.stage, req.downstreamTaskId, req.downstreamNodeId, - pRange->range.minVer, pRange->range.maxVer, pWindow->skey, pWindow->ekey, req.reqId); - - streamSendCheckMsg(pTask, &req, pTask->outputInfo.fixedDispatcher.nodeId, &pTask->outputInfo.fixedDispatcher.epSet); - - } else if (pTask->outputInfo.type == TASK_OUTPUT__SHUFFLE_DISPATCH) { - streamTaskStartMonitorCheckRsp(pTask); - - SArray* vgInfo = pTask->outputInfo.shuffleDispatcher.dbInfo.pVgroupInfos; - - int32_t numOfVgs = taosArrayGetSize(vgInfo); - stDebug("s-task:%s check %d downstream tasks, ver:%" PRId64 "-%" PRId64 " window:%" PRId64 "-%" PRId64, - pTask->id.idStr, numOfVgs, pRange->range.minVer, pRange->range.maxVer, pWindow->skey, pWindow->ekey); - - for (int32_t i = 0; i < numOfVgs; i++) { - SVgroupInfo* pVgInfo = taosArrayGet(vgInfo, i); - req.reqId = tGenIdPI64(); - req.downstreamNodeId = pVgInfo->vgId; - req.downstreamTaskId = pVgInfo->taskId; - - streamTaskAddReqInfo(&pTask->taskCheckInfo, req.reqId, req.downstreamTaskId, pTask->id.idStr); - - stDebug("s-task:%s (vgId:%d) stage:%" PRId64 - " check downstream task:0x%x (vgId:%d) (shuffle), idx:%d, reqId:0x%" PRIx64, - pTask->id.idStr, pTask->info.nodeId, req.stage, req.downstreamTaskId, req.downstreamNodeId, i, req.reqId); - streamSendCheckMsg(pTask, &req, pVgInfo->vgId, &pVgInfo->epSet); - } - } else { // for sink task, set it ready directly. - stDebug("s-task:%s (vgId:%d) set downstream ready, since no downstream", pTask->id.idStr, pTask->info.nodeId); - streamTaskStopMonitorCheckRsp(&pTask->taskCheckInfo, pTask->id.idStr); - doProcessDownstreamReadyRsp(pTask); - } -} - int32_t streamTaskCheckStatus(SStreamTask* pTask, int32_t upstreamTaskId, int32_t vgId, int64_t stage, int64_t* oldStage) { SStreamChildEpInfo* pInfo = streamTaskGetUpstreamTaskEpInfo(pTask, upstreamTaskId); @@ -327,121 +266,6 @@ int32_t streamTaskOnScanhistoryTaskReady(SStreamTask* pTask) { return TSDB_CODE_SUCCESS; } -void doProcessDownstreamReadyRsp(SStreamTask* pTask) { - EStreamTaskEvent event = (pTask->info.fillHistory == 0) ? TASK_EVENT_INIT : TASK_EVENT_INIT_SCANHIST; - streamTaskOnHandleEventSuccess(pTask->status.pSM, event, NULL, NULL); - - int64_t checkTs = pTask->execInfo.checkTs; - int64_t readyTs = pTask->execInfo.readyTs; - streamMetaAddTaskLaunchResult(pTask->pMeta, pTask->id.streamId, pTask->id.taskId, checkTs, readyTs, true); - - if (pTask->status.taskStatus == TASK_STATUS__HALT) { - ASSERT(HAS_RELATED_FILLHISTORY_TASK(pTask) && (pTask->info.fillHistory == 0)); - - // halt it self for count window stream task until the related fill history task completed. - stDebug("s-task:%s level:%d initial status is %s from mnode, set it to be halt", pTask->id.idStr, - pTask->info.taskLevel, streamTaskGetStatusStr(pTask->status.taskStatus)); - streamTaskHandleEvent(pTask->status.pSM, TASK_EVENT_HALT); - } - - // start the related fill-history task, when current task is ready - // not invoke in success callback due to the deadlock. - if (HAS_RELATED_FILLHISTORY_TASK(pTask)) { - stDebug("s-task:%s try to launch related fill-history task", pTask->id.idStr); - streamLaunchFillHistoryTask(pTask); - } -} - -static void addIntoNodeUpdateList(SStreamTask* pTask, int32_t nodeId) { - int32_t vgId = pTask->pMeta->vgId; - - taosThreadMutexLock(&pTask->lock); - int32_t num = taosArrayGetSize(pTask->outputInfo.pNodeEpsetUpdateList); - bool existed = false; - for (int i = 0; i < num; ++i) { - SDownstreamTaskEpset* p = taosArrayGet(pTask->outputInfo.pNodeEpsetUpdateList, i); - if (p->nodeId == nodeId) { - existed = true; - break; - } - } - - if (!existed) { - SDownstreamTaskEpset t = {.nodeId = nodeId}; - taosArrayPush(pTask->outputInfo.pNodeEpsetUpdateList, &t); - - stInfo("s-task:%s vgId:%d downstream nodeId:%d needs to be updated, total needs updated:%d", pTask->id.idStr, vgId, - t.nodeId, (num + 1)); - } - - taosThreadMutexUnlock(&pTask->lock); -} - -int32_t streamProcessCheckRsp(SStreamTask* pTask, const SStreamTaskCheckRsp* pRsp) { - ASSERT(pTask->id.taskId == pRsp->upstreamTaskId); - - int64_t now = taosGetTimestampMs(); - const char* id = pTask->id.idStr; - STaskCheckInfo* pInfo = &pTask->taskCheckInfo; - int32_t total = streamTaskGetNumOfDownstream(pTask); - int32_t left = -1; - - if (streamTaskShouldStop(pTask)) { - stDebug("s-task:%s should stop, do not do check downstream again", id); - return TSDB_CODE_SUCCESS; - } - - if (pRsp->status == TASK_DOWNSTREAM_READY) { - int32_t code = streamTaskUpdateCheckInfo(pInfo, pRsp->downstreamTaskId, pRsp->status, now, pRsp->reqId, &left, id); - if (code != TSDB_CODE_SUCCESS) { - return TSDB_CODE_SUCCESS; - } - - if (left == 0) { - doProcessDownstreamReadyRsp(pTask); // all downstream tasks are ready, set the complete check downstream flag - streamTaskStopMonitorCheckRsp(pInfo, id); - } else { - stDebug("s-task:%s (vgId:%d) recv check rsp from task:0x%x (vgId:%d) status:%d, total:%d not ready:%d", id, - pRsp->upstreamNodeId, pRsp->downstreamTaskId, pRsp->downstreamNodeId, pRsp->status, total, left); - } - } else { // not ready, wait for 100ms and retry - int32_t code = streamTaskUpdateCheckInfo(pInfo, pRsp->downstreamTaskId, pRsp->status, now, pRsp->reqId, &left, id); - if (code != TSDB_CODE_SUCCESS) { - return TSDB_CODE_SUCCESS; // return success in any cases. - } - - if (pRsp->status == TASK_UPSTREAM_NEW_STAGE || pRsp->status == TASK_DOWNSTREAM_NOT_LEADER) { - if (pRsp->status == TASK_UPSTREAM_NEW_STAGE) { - stError("s-task:%s vgId:%d self vnode-transfer/leader-change/restart detected, old stage:%" PRId64 - ", current stage:%" PRId64 ", not check wait for downstream task nodeUpdate, and all tasks restart", - id, pRsp->upstreamNodeId, pRsp->oldStage, pTask->pMeta->stage); - addIntoNodeUpdateList(pTask, pRsp->upstreamNodeId); - } else { - stError( - "s-task:%s downstream taskId:0x%x (vgId:%d) not leader, self dispatch epset needs to be updated, not check " - "downstream again, nodeUpdate needed", - id, pRsp->downstreamTaskId, pRsp->downstreamNodeId); - addIntoNodeUpdateList(pTask, pRsp->downstreamNodeId); - } - - int32_t startTs = pTask->execInfo.checkTs; - streamMetaAddTaskLaunchResult(pTask->pMeta, pTask->id.streamId, pTask->id.taskId, startTs, now, false); - - // automatically set the related fill-history task to be failed. - if (HAS_RELATED_FILLHISTORY_TASK(pTask)) { - STaskId* pId = &pTask->hTaskInfo.id; - streamMetaAddTaskLaunchResult(pTask->pMeta, pId->streamId, pId->taskId, startTs, now, false); - } - } else { // TASK_DOWNSTREAM_NOT_READY, let's retry in 100ms - ASSERT(left > 0); - stDebug("s-task:%s (vgId:%d) recv check rsp from task:0x%x (vgId:%d) status:%d, total:%d not ready:%d", id, - pRsp->upstreamNodeId, pRsp->downstreamTaskId, pRsp->downstreamNodeId, pRsp->status, total, left); - } - } - - return 0; -} - int32_t streamSendCheckRsp(const SStreamMeta* pMeta, const SStreamTaskCheckReq* pReq, SStreamTaskCheckRsp* pRsp, SRpcHandleInfo* pRpcInfo, int32_t taskId) { SEncoder encoder; @@ -797,64 +621,6 @@ bool streamHistoryTaskSetVerRangeStep2(SStreamTask* pTask, int64_t nextProcessVe } } -int32_t tEncodeStreamTaskCheckReq(SEncoder* pEncoder, const SStreamTaskCheckReq* pReq) { - if (tStartEncode(pEncoder) < 0) return -1; - if (tEncodeI64(pEncoder, pReq->reqId) < 0) return -1; - if (tEncodeI64(pEncoder, pReq->streamId) < 0) return -1; - if (tEncodeI32(pEncoder, pReq->upstreamNodeId) < 0) return -1; - if (tEncodeI32(pEncoder, pReq->upstreamTaskId) < 0) return -1; - if (tEncodeI32(pEncoder, pReq->downstreamNodeId) < 0) return -1; - if (tEncodeI32(pEncoder, pReq->downstreamTaskId) < 0) return -1; - if (tEncodeI32(pEncoder, pReq->childId) < 0) return -1; - if (tEncodeI64(pEncoder, pReq->stage) < 0) return -1; - tEndEncode(pEncoder); - return pEncoder->pos; -} - -int32_t tDecodeStreamTaskCheckReq(SDecoder* pDecoder, SStreamTaskCheckReq* pReq) { - if (tStartDecode(pDecoder) < 0) return -1; - if (tDecodeI64(pDecoder, &pReq->reqId) < 0) return -1; - if (tDecodeI64(pDecoder, &pReq->streamId) < 0) return -1; - if (tDecodeI32(pDecoder, &pReq->upstreamNodeId) < 0) return -1; - if (tDecodeI32(pDecoder, &pReq->upstreamTaskId) < 0) return -1; - if (tDecodeI32(pDecoder, &pReq->downstreamNodeId) < 0) return -1; - if (tDecodeI32(pDecoder, &pReq->downstreamTaskId) < 0) return -1; - if (tDecodeI32(pDecoder, &pReq->childId) < 0) return -1; - if (tDecodeI64(pDecoder, &pReq->stage) < 0) return -1; - tEndDecode(pDecoder); - return 0; -} - -int32_t tEncodeStreamTaskCheckRsp(SEncoder* pEncoder, const SStreamTaskCheckRsp* pRsp) { - if (tStartEncode(pEncoder) < 0) return -1; - if (tEncodeI64(pEncoder, pRsp->reqId) < 0) return -1; - if (tEncodeI64(pEncoder, pRsp->streamId) < 0) return -1; - if (tEncodeI32(pEncoder, pRsp->upstreamNodeId) < 0) return -1; - if (tEncodeI32(pEncoder, pRsp->upstreamTaskId) < 0) return -1; - if (tEncodeI32(pEncoder, pRsp->downstreamNodeId) < 0) return -1; - if (tEncodeI32(pEncoder, pRsp->downstreamTaskId) < 0) return -1; - if (tEncodeI32(pEncoder, pRsp->childId) < 0) return -1; - if (tEncodeI64(pEncoder, pRsp->oldStage) < 0) return -1; - if (tEncodeI8(pEncoder, pRsp->status) < 0) return -1; - tEndEncode(pEncoder); - return pEncoder->pos; -} - -int32_t tDecodeStreamTaskCheckRsp(SDecoder* pDecoder, SStreamTaskCheckRsp* pRsp) { - if (tStartDecode(pDecoder) < 0) return -1; - if (tDecodeI64(pDecoder, &pRsp->reqId) < 0) return -1; - if (tDecodeI64(pDecoder, &pRsp->streamId) < 0) return -1; - if (tDecodeI32(pDecoder, &pRsp->upstreamNodeId) < 0) return -1; - if (tDecodeI32(pDecoder, &pRsp->upstreamTaskId) < 0) return -1; - if (tDecodeI32(pDecoder, &pRsp->downstreamNodeId) < 0) return -1; - if (tDecodeI32(pDecoder, &pRsp->downstreamTaskId) < 0) return -1; - if (tDecodeI32(pDecoder, &pRsp->childId) < 0) return -1; - if (tDecodeI64(pDecoder, &pRsp->oldStage) < 0) return -1; - if (tDecodeI8(pDecoder, &pRsp->status) < 0) return -1; - tEndDecode(pDecoder); - return 0; -} - void streamTaskSetRangeStreamCalc(SStreamTask* pTask) { SDataRange* pRange = &pTask->dataRange; diff --git a/source/libs/stream/src/streamTask.c b/source/libs/stream/src/streamTask.c index 962cadfbd6..9f4e6aaea1 100644 --- a/source/libs/stream/src/streamTask.c +++ b/source/libs/stream/src/streamTask.c @@ -21,8 +21,6 @@ #include "ttimer.h" #include "wal.h" -#define CHECK_NOT_RSP_DURATION 10*1000 // 10 sec - static void streamTaskDestroyUpstreamInfo(SUpstreamInfo* pUpstreamInfo); static void streamTaskUpdateUpstreamInfo(SStreamTask* pTask, int32_t nodeId, const SEpSet* pEpSet, bool* pUpdated); static void streamTaskUpdateDownstreamInfo(SStreamTask* pTask, int32_t nodeId, const SEpSet* pEpSet, bool* pUpdate); @@ -442,7 +440,7 @@ void tFreeStreamTask(SStreamTask* pTask) { taosArrayDestroy(pTask->outputInfo.shuffleDispatcher.dbInfo.pVgroupInfos); } - streamTaskCleanCheckInfo(&pTask->taskCheckInfo); + streamTaskCleanupCheckInfo(&pTask->taskCheckInfo); if (pTask->pState) { stDebug("s-task:0x%x start to free task state", taskId); @@ -994,386 +992,3 @@ int32_t streamTaskSendCheckpointReq(SStreamTask* pTask) { tmsgSendReq(&pTask->info.mnodeEpset, &msg); return 0; } - -static int32_t streamTaskInitTaskCheckInfo(STaskCheckInfo* pInfo, STaskOutputInfo* pOutputInfo, int64_t startTs) { - taosArrayClear(pInfo->pList); - - if (pOutputInfo->type == TASK_OUTPUT__FIXED_DISPATCH) { - pInfo->notReadyTasks = 1; - } else if (pOutputInfo->type == TASK_OUTPUT__SHUFFLE_DISPATCH) { - pInfo->notReadyTasks = taosArrayGetSize(pOutputInfo->shuffleDispatcher.dbInfo.pVgroupInfos); - ASSERT(pInfo->notReadyTasks == pOutputInfo->shuffleDispatcher.dbInfo.vgNum); - } - - pInfo->startTs = startTs; - return TSDB_CODE_SUCCESS; -} - -static SDownstreamStatusInfo* findCheckRspStatus(STaskCheckInfo* pInfo, int32_t taskId) { - for (int32_t j = 0; j < taosArrayGetSize(pInfo->pList); ++j) { - SDownstreamStatusInfo* p = taosArrayGet(pInfo->pList, j); - if (p->taskId == taskId) { - return p; - } - } - - return NULL; -} - -int32_t streamTaskAddReqInfo(STaskCheckInfo* pInfo, int64_t reqId, int32_t taskId, const char* id) { - SDownstreamStatusInfo info = {.taskId = taskId, .status = -1, .reqId = reqId, .rspTs = 0}; - - taosThreadMutexLock(&pInfo->checkInfoLock); - - SDownstreamStatusInfo* p = findCheckRspStatus(pInfo, taskId); - if (p != NULL) { - stDebug("s-task:%s check info to task:0x%x already sent", id, taskId); - taosThreadMutexUnlock(&pInfo->checkInfoLock); - return TSDB_CODE_SUCCESS; - } - - taosArrayPush(pInfo->pList, &info); - - taosThreadMutexUnlock(&pInfo->checkInfoLock); - return TSDB_CODE_SUCCESS; -} - -int32_t streamTaskUpdateCheckInfo(STaskCheckInfo* pInfo, int32_t taskId, int32_t status, int64_t rspTs, int64_t reqId, - int32_t* pNotReady, const char* id) { - taosThreadMutexLock(&pInfo->checkInfoLock); - - SDownstreamStatusInfo* p = findCheckRspStatus(pInfo, taskId); - if (p != NULL) { - - if (reqId != p->reqId) { - stError("s-task:%s reqId:%" PRIx64 " expected:%" PRIx64 - " expired check-rsp recv from downstream task:0x%x, discarded", - id, reqId, p->reqId, taskId); - taosThreadMutexUnlock(&pInfo->checkInfoLock); - return TSDB_CODE_FAILED; - } - - // subtract one not-ready-task, since it is ready now - if ((p->status != TASK_DOWNSTREAM_READY) && (status == TASK_DOWNSTREAM_READY)) { - *pNotReady = atomic_sub_fetch_32(&pInfo->notReadyTasks, 1); - } else { - *pNotReady = pInfo->notReadyTasks; - } - - p->status = status; - p->rspTs = rspTs; - - taosThreadMutexUnlock(&pInfo->checkInfoLock); - return TSDB_CODE_SUCCESS; - } - - taosThreadMutexUnlock(&pInfo->checkInfoLock); - stError("s-task:%s unexpected check rsp msg, invalid downstream task:0x%x, reqId:%" PRIx64 " discarded", id, taskId, - reqId); - return TSDB_CODE_FAILED; -} - -static int32_t streamTaskStartCheckDownstream(STaskCheckInfo* pInfo, const char* id) { - if (pInfo->inCheckProcess == 0) { - pInfo->inCheckProcess = 1; - } else { - ASSERT(pInfo->startTs > 0); - stError("s-task:%s already in check procedure, checkTs:%"PRId64", start monitor check rsp failed", id, pInfo->startTs); - return TSDB_CODE_FAILED; - } - - stDebug("s-task:%s set the in-check-procedure flag", id); - return 0; -} - -static int32_t streamTaskCompleteCheckRsp(STaskCheckInfo* pInfo, const char* id) { - if (!pInfo->inCheckProcess) { - stWarn("s-task:%s already not in-check-procedure", id); - } - - int64_t el = (pInfo->startTs != 0) ? (taosGetTimestampMs() - pInfo->startTs) : 0; - stDebug("s-task:%s clear the in-check-procedure flag, not in-check-procedure elapsed time:%" PRId64 " ms", id, el); - - pInfo->startTs = 0; - pInfo->notReadyTasks = 0; - pInfo->inCheckProcess = 0; - pInfo->stopCheckProcess = 0; - - pInfo->notReadyRetryCount = 0; - pInfo->timeoutRetryCount = 0; - - taosArrayClear(pInfo->pList); - return 0; -} - -static void doSendCheckMsg(SStreamTask* pTask, SDownstreamStatusInfo* p) { - SStreamTaskCheckReq req = { - .streamId = pTask->id.streamId, - .upstreamTaskId = pTask->id.taskId, - .upstreamNodeId = pTask->info.nodeId, - .childId = pTask->info.selfChildId, - .stage = pTask->pMeta->stage, - }; - - STaskOutputInfo* pOutputInfo = &pTask->outputInfo; - if (pOutputInfo->type == TASK_OUTPUT__FIXED_DISPATCH) { - req.reqId = p->reqId; - req.downstreamNodeId = pOutputInfo->fixedDispatcher.nodeId; - req.downstreamTaskId = pOutputInfo->fixedDispatcher.taskId; - stDebug("s-task:%s (vgId:%d) stage:%" PRId64 " re-send check downstream task:0x%x(vgId:%d) reqId:0x%" PRIx64, - pTask->id.idStr, pTask->info.nodeId, req.stage, req.downstreamTaskId, req.downstreamNodeId, req.reqId); - - streamSendCheckMsg(pTask, &req, pOutputInfo->fixedDispatcher.nodeId, &pOutputInfo->fixedDispatcher.epSet); - } else if (pOutputInfo->type == TASK_OUTPUT__SHUFFLE_DISPATCH) { - SArray* vgInfo = pOutputInfo->shuffleDispatcher.dbInfo.pVgroupInfos; - int32_t numOfVgs = taosArrayGetSize(vgInfo); - - for (int32_t i = 0; i < numOfVgs; i++) { - SVgroupInfo* pVgInfo = taosArrayGet(vgInfo, i); - - if (p->taskId == pVgInfo->taskId) { - req.reqId = p->reqId; - req.downstreamNodeId = pVgInfo->vgId; - req.downstreamTaskId = pVgInfo->taskId; - - stDebug("s-task:%s (vgId:%d) stage:%" PRId64 - " re-send check downstream task:0x%x(vgId:%d) (shuffle), idx:%d reqId:0x%" PRIx64, - pTask->id.idStr, pTask->info.nodeId, req.stage, req.downstreamTaskId, req.downstreamNodeId, i, - p->reqId); - streamSendCheckMsg(pTask, &req, pVgInfo->vgId, &pVgInfo->epSet); - break; - } - } - } else { - ASSERT(0); - } -} - -static void getCheckRspStatus(STaskCheckInfo* pInfo, int64_t el, int32_t* numOfReady, int32_t* numOfFault, - int32_t* numOfNotRsp, SArray* pTimeoutList, SArray* pNotReadyList, const char* id) { - for (int32_t i = 0; i < taosArrayGetSize(pInfo->pList); ++i) { - SDownstreamStatusInfo* p = taosArrayGet(pInfo->pList, i); - if (p->status == TASK_DOWNSTREAM_READY) { - (*numOfReady) += 1; - } else if (p->status == TASK_UPSTREAM_NEW_STAGE || p->status == TASK_DOWNSTREAM_NOT_LEADER) { - stDebug("s-task:%s recv status:NEW_STAGE/NOT_LEADER from downstream, task:0x%x, quit from check downstream", id, - p->taskId); - (*numOfFault) += 1; - } else { // TASK_DOWNSTREAM_NOT_READY - if (p->rspTs == 0) { // not response yet - ASSERT(p->status == -1); - if (el >= CHECK_NOT_RSP_DURATION) { // not receive info for 10 sec. - taosArrayPush(pTimeoutList, &p->taskId); - } else { // el < CHECK_NOT_RSP_DURATION - (*numOfNotRsp) += 1; // do nothing and continue waiting for their rsp - } - } else { - taosArrayPush(pNotReadyList, &p->taskId); - } - } - } -} - -static void rspMonitorFn(void* param, void* tmrId) { - SStreamTask* pTask = param; - SStreamTaskState* pStat = streamTaskGetStatus(pTask); - STaskCheckInfo* pInfo = &pTask->taskCheckInfo; - int32_t vgId = pTask->pMeta->vgId; - int64_t now = taosGetTimestampMs(); - int64_t el = now - pInfo->startTs; - ETaskStatus state = pStat->state; - const char* id = pTask->id.idStr; - int32_t numOfReady = 0; - int32_t numOfFault = 0; - int32_t numOfNotRsp = 0; - int32_t numOfNotReady = 0; - int32_t numOfTimeout = 0; - - stDebug("s-task:%s start to do check-downstream-rsp check in tmr", id); - - if (state == TASK_STATUS__STOP) { - int32_t ref = atomic_sub_fetch_32(&pTask->status.timerActive, 1); - stDebug("s-task:%s status:%s vgId:%d quit from monitor check-rsp tmr, ref:%d", id, pStat->name, vgId, ref); - - taosThreadMutexLock(&pInfo->checkInfoLock); - streamTaskCompleteCheckRsp(pInfo, id); - taosThreadMutexUnlock(&pInfo->checkInfoLock); - - streamMetaAddTaskLaunchResult(pTask->pMeta, pTask->id.streamId, pTask->id.taskId, pInfo->startTs, now, false); - if (HAS_RELATED_FILLHISTORY_TASK(pTask)) { - STaskId* pHId = &pTask->hTaskInfo.id; - streamMetaAddTaskLaunchResult(pTask->pMeta, pHId->streamId, pHId->taskId, pInfo->startTs, now, false); - } - return; - } - - if (state == TASK_STATUS__DROPPING || state == TASK_STATUS__READY || state == TASK_STATUS__PAUSE) { - int32_t ref = atomic_sub_fetch_32(&pTask->status.timerActive, 1); - stDebug("s-task:%s status:%s vgId:%d quit from monitor check-rsp tmr, ref:%d", id, pStat->name, vgId, ref); - - taosThreadMutexLock(&pInfo->checkInfoLock); - streamTaskCompleteCheckRsp(pInfo, id); - taosThreadMutexUnlock(&pInfo->checkInfoLock); - return; - } - - taosThreadMutexLock(&pInfo->checkInfoLock); - if (pInfo->notReadyTasks == 0) { - int32_t ref = atomic_sub_fetch_32(&pTask->status.timerActive, 1); - stDebug("s-task:%s status:%s vgId:%d all downstream ready, quit from monitor rsp tmr, ref:%d", id, pStat->name, - vgId, ref); - - streamTaskCompleteCheckRsp(pInfo, id); - taosThreadMutexUnlock(&pInfo->checkInfoLock); - return; - } - - SArray* pNotReadyList = taosArrayInit(4, sizeof(int64_t)); - SArray* pTimeoutList = taosArrayInit(4, sizeof(int64_t)); - - if (pStat->state == TASK_STATUS__UNINIT) { - getCheckRspStatus(pInfo, el, &numOfReady, &numOfFault, &numOfNotRsp, pTimeoutList, pNotReadyList, id); - } else { // unexpected status - stError("s-task:%s unexpected task status:%s during waiting for check rsp", id, pStat->name); - } - - numOfNotReady = (int32_t)taosArrayGetSize(pNotReadyList); - numOfTimeout = (int32_t)taosArrayGetSize(pTimeoutList); - - // fault tasks detected, not try anymore - ASSERT((numOfReady + numOfFault + numOfNotReady + numOfTimeout + numOfNotRsp) == taosArrayGetSize(pInfo->pList)); - if (numOfFault > 0) { - int32_t ref = atomic_sub_fetch_32(&pTask->status.timerActive, 1); - stDebug( - "s-task:%s status:%s vgId:%d all rsp. quit from monitor rsp tmr, since vnode-transfer/leader-change/restart " - "detected, notRsp:%d, notReady:%d, fault:%d, timeout:%d, ready:%d ref:%d", - id, pStat->name, vgId, numOfNotRsp, numOfNotReady, numOfFault, numOfTimeout, numOfReady, ref); - - streamTaskCompleteCheckRsp(pInfo, id); - taosThreadMutexUnlock(&pInfo->checkInfoLock); - - taosArrayDestroy(pNotReadyList); - taosArrayDestroy(pTimeoutList); - return; - } - - // checking of downstream tasks has been stopped by other threads - if (pInfo->stopCheckProcess == 1) { - int32_t ref = atomic_sub_fetch_32(&pTask->status.timerActive, 1); - stDebug( - "s-task:%s status:%s vgId:%d stopped by other threads to check downstream process, notRsp:%d, notReady:%d, " - "fault:%d, timeout:%d, ready:%d ref:%d", - id, pStat->name, vgId, numOfNotRsp, numOfNotReady, numOfFault, numOfTimeout, numOfReady, ref); - - streamTaskCompleteCheckRsp(pInfo, id); - taosThreadMutexUnlock(&pInfo->checkInfoLock); - - // add the not-ready tasks into the final task status result buf, along with related fill-history task if exists. - streamMetaAddTaskLaunchResult(pTask->pMeta, pTask->id.streamId, pTask->id.taskId, pInfo->startTs, now, false); - if (HAS_RELATED_FILLHISTORY_TASK(pTask)) { - STaskId* pHId = &pTask->hTaskInfo.id; - streamMetaAddTaskLaunchResult(pTask->pMeta, pHId->streamId, pHId->taskId, pInfo->startTs, now, false); - } - - taosArrayDestroy(pNotReadyList); - taosArrayDestroy(pTimeoutList); - return; - } - - if (numOfNotReady > 0) { // check to make sure not in recheck timer - ASSERT(pTask->status.downstreamReady == 0); - - // reset the info, and send the check msg to failure downstream again - for (int32_t i = 0; i < numOfNotReady; ++i) { - int32_t taskId = *(int32_t*)taosArrayGet(pNotReadyList, i); - - SDownstreamStatusInfo* p = findCheckRspStatus(pInfo, taskId); - if (p != NULL) { - p->rspTs = 0; - p->status = -1; - doSendCheckMsg(pTask, p); - } - } - - pInfo->notReadyRetryCount += 1; - stDebug("s-task:%s %d downstream task(s) not ready, send check msg again, retry:%d start time:%" PRId64, id, - numOfNotReady, pInfo->notReadyRetryCount, pInfo->startTs); - } - - // todo add into node update list and send to mnode - if (numOfTimeout > 0) { - ASSERT(pTask->status.downstreamReady == 0); - - for (int32_t i = 0; i < numOfTimeout; ++i) { - int32_t taskId = *(int32_t*)taosArrayGet(pTimeoutList, i); - - SDownstreamStatusInfo* p = findCheckRspStatus(pInfo, taskId); - if (p != NULL) { - ASSERT(p->status == -1 && p->rspTs == 0); - doSendCheckMsg(pTask, p); - } - } - - pInfo->timeoutRetryCount += 1; - stDebug("s-task:%s %d downstream task(s) timeout, send check msg again, retry:%d start time:%" PRId64, id, - numOfTimeout, pInfo->timeoutRetryCount, pInfo->startTs); - } - - taosTmrReset(rspMonitorFn, CHECK_RSP_INTERVAL, pTask, streamTimer, &pInfo->checkRspTmr); - taosThreadMutexUnlock(&pInfo->checkInfoLock); - - stDebug("s-task:%s continue checking rsp in 300ms, notRsp:%d, notReady:%d, fault:%d, timeout:%d, ready:%d", id, - numOfNotRsp, numOfNotReady, numOfFault, numOfTimeout, numOfReady); - - taosArrayDestroy(pNotReadyList); - taosArrayDestroy(pTimeoutList); -} - -int32_t streamTaskStartMonitorCheckRsp(SStreamTask* pTask) { - STaskCheckInfo* pInfo = &pTask->taskCheckInfo; - - taosThreadMutexLock(&pInfo->checkInfoLock); - int32_t code = streamTaskStartCheckDownstream(pInfo, pTask->id.idStr); - if (code != TSDB_CODE_SUCCESS) { - - taosThreadMutexUnlock(&pInfo->checkInfoLock); - return TSDB_CODE_FAILED; - } - - streamTaskInitTaskCheckInfo(pInfo, &pTask->outputInfo, taosGetTimestampMs()); - - int32_t ref = atomic_add_fetch_32(&pTask->status.timerActive, 1); - stDebug("s-task:%s start check rsp monit, ref:%d ", pTask->id.idStr, ref); - - if (pInfo->checkRspTmr == NULL) { - pInfo->checkRspTmr = taosTmrStart(rspMonitorFn, CHECK_RSP_INTERVAL, pTask, streamTimer); - } else { - taosTmrReset(rspMonitorFn, CHECK_RSP_INTERVAL, pTask, streamTimer, &pInfo->checkRspTmr); - } - - taosThreadMutexUnlock(&pInfo->checkInfoLock); - return 0; -} - -int32_t streamTaskStopMonitorCheckRsp(STaskCheckInfo* pInfo, const char* id) { - taosThreadMutexLock(&pInfo->checkInfoLock); - streamTaskCompleteCheckRsp(pInfo, id); - - pInfo->stopCheckProcess = 1; - taosThreadMutexUnlock(&pInfo->checkInfoLock); - - stDebug("s-task:%s set stop check rsp mon", id); - return TSDB_CODE_SUCCESS; -} - -void streamTaskCleanCheckInfo(STaskCheckInfo* pInfo) { - ASSERT(pInfo->inCheckProcess == 0); - - pInfo->pList = taosArrayDestroy(pInfo->pList); - if (pInfo->checkRspTmr != NULL) { - /*bool ret = */ taosTmrStop(pInfo->checkRspTmr); - pInfo->checkRspTmr = NULL; - } - - taosThreadMutexDestroy(&pInfo->checkInfoLock); -} \ No newline at end of file