提交 072a73ec 编写于 作者: D dapan1121

enh: support max retry wait time configuration

上级 2fbf082f
......@@ -90,6 +90,10 @@ extern int32_t tsQueryNodeChunkSize;
extern bool tsQueryUseNodeAllocator;
extern bool tsKeepColumnName;
extern bool tsEnableQueryHb;
extern int32_t tsRedirectPeriod;
extern int32_t tsRedirectFactor;
extern int32_t tsRedirectMaxPeriod;
extern int32_t tsMaxRetryWaitTime;
// client
extern int32_t tsMinSlidingTime;
......
......@@ -87,6 +87,10 @@ bool tsQueryPlannerTrace = false;
int32_t tsQueryNodeChunkSize = 32 * 1024;
bool tsQueryUseNodeAllocator = true;
bool tsKeepColumnName = false;
int32_t tsRedirectPeriod = 100;
int32_t tsRedirectFactor = 5;
int32_t tsRedirectMaxPeriod = 10000;
int32_t tsMaxRetryWaitTime = 60000;
/*
* denote if the server needs to compress response message at the application layer to client, including query rsp,
......@@ -301,6 +305,7 @@ static int32_t taosAddClientCfg(SConfig *pCfg) {
if (cfgAddInt32(pCfg, "maxMemUsedByInsert", tsMaxMemUsedByInsert, 1, INT32_MAX, true) != 0) return -1;
if (cfgAddInt32(pCfg, "rpcRetryLimit", tsRpcRetryLimit, 1, 100000, 0) != 0) return -1;
if (cfgAddInt32(pCfg, "rpcRetryInterval", tsRpcRetryInterval, 1, 100000, 0) != 0) return -1;
if (cfgAddInt32(pCfg, "maxRetryWaitTime", tsMaxRetryWaitTime, 0, 86400000, 0) != 0) return -1;
tsNumOfTaskQueueThreads = tsNumOfCores / 2;
tsNumOfTaskQueueThreads = TMAX(tsNumOfTaskQueueThreads, 4);
......@@ -645,6 +650,7 @@ static int32_t taosSetClientCfg(SConfig *pCfg) {
tsRpcRetryLimit = cfgGetItem(pCfg, "rpcRetryLimit")->i32;
tsRpcRetryInterval = cfgGetItem(pCfg, "rpcRetryInterval")->i32;
tsMaxRetryWaitTime = cfgGetItem(pCfg, "maxRetryWaitTime")->i32;
return 0;
}
......@@ -860,6 +866,8 @@ int32_t taosSetCfg(SConfig *pCfg, char *name) {
tsMaxNumOfDistinctResults = cfgGetItem(pCfg, "maxNumOfDistinctRes")->i32;
} else if (strcasecmp("maxMemUsedByInsert", name) == 0) {
tsMaxMemUsedByInsert = cfgGetItem(pCfg, "maxMemUsedByInsert")->i32;
} else if (strcasecmp("maxRetryWaitTime", name) == 0) {
tsMaxRetryWaitTime = cfgGetItem(pCfg, "maxRetryWaitTime")->i32;
}
break;
}
......
......@@ -27,6 +27,7 @@ extern "C" {
#include "tarray.h"
#include "thash.h"
#include "trpc.h"
#include "ttimer.h"
enum {
SCH_READ = 1,
......@@ -507,6 +508,7 @@ extern SSchedulerMgmt schMgmt;
void schDeregisterTaskHb(SSchJob *pJob, SSchTask *pTask);
void schCleanClusterHb(void *pTrans);
int32_t schLaunchTask(SSchJob *job, SSchTask *task);
int32_t schDelayLaunchTask(SSchJob *pJob, SSchTask *pTask);
int32_t schBuildAndSendMsg(SSchJob *job, SSchTask *task, SQueryNodeAddr *addr, int32_t msgType);
SSchJob *schAcquireJob(int64_t refId);
int32_t schReleaseJob(int64_t refId);
......
......@@ -887,8 +887,13 @@ int32_t schAsyncSendMsg(SSchJob *pJob, SSchTask *pTask, SSchTrans *trans, SQuery
SCH_ERR_JRET(schGenerateCallBackInfo(pJob, pTask, msg, msgSize, msgType, trans, isHb, &pMsgSendInfo));
SCH_ERR_JRET(schUpdateSendTargetInfo(pMsgSendInfo, addr, pTask));
if (pJob && pTask) {
SCH_TASK_DLOG("start to send %s msg to node[%d,%s,%d], pTrans:%p, pHandle:%p", TMSG_INFO(msgType), addr->nodeId,
epSet->eps[epSet->inUse].fqdn, epSet->eps[epSet->inUse].port, trans->pTrans, trans->pHandle);
} else {
qDebug("start to send %s msg to node[%d,%s,%d], pTrans:%p, pHandle:%p", TMSG_INFO(msgType), addr->nodeId,
epSet->eps[epSet->inUse].fqdn, epSet->eps[epSet->inUse].port, trans->pTrans, trans->pHandle);
}
if (pTask) {
pTask->lastMsgType = msgType;
......
......@@ -340,7 +340,7 @@ int32_t schRescheduleTask(SSchJob *pJob, SSchTask *pTask) {
return TSDB_CODE_SUCCESS;
}
int32_t schChkUpdateRedirectCtx(SSchTask *pTask, SEpSet *pEpSet) {
int32_t schChkUpdateRedirectCtx(SSchJob *pJob, SSchTask *pTask, SEpSet *pEpSet) {
SSchRedirectCtx *pCtx = &pTask->redirectCtx;
if (!pCtx->inRedirect) {
pCtx->inRedirect = true;
......@@ -363,13 +363,6 @@ int32_t schChkUpdateRedirectCtx(SSchTask *pTask, SEpSet *pEpSet) {
pCtx->totalTimes++;
int64_t nowTs = taosGetTimestampMs();
if ((nowTs - pCtx->startTs) > tsMaxRetryWaitTime) {
SCH_TASK_DLOG("task no more redirect retry since timeout, now:%" PRId64 ", start:%" PRId64 ", max:%d, total:%d",
nowTs, pCtx->startTs, tsMaxRetryWaitTime, pCtx->totalTimes);
SCH_ERR_RET(TSDB_CODE_TIMEOUT_ERROR);
}
if (SCH_IS_DATA_BIND_TASK(pTask) && pEpSet) {
pCtx->roundTotal = pEpSet->numOfEps;
pCtx->roundTimes = 0;
......@@ -382,12 +375,21 @@ int32_t schChkUpdateRedirectCtx(SSchTask *pTask, SEpSet *pEpSet) {
pCtx->roundTimes++;
if (pCtx->roundTimes >= pCtx->roundTotal) {
int64_t nowTs = taosGetTimestampMs();
int64_t lastTime = nowTs - pCtx->startTs;
if (lastTime > tsMaxRetryWaitTime) {
SCH_TASK_DLOG("task no more redirect retry since timeout, now:%" PRId64 ", start:%" PRId64 ", max:%d, total:%d",
nowTs, pCtx->startTs, tsMaxRetryWaitTime, pCtx->totalTimes);
SCH_ERR_RET(TSDB_CODE_TIMEOUT_ERROR);
}
pCtx->periodMs *= tsRedirectFactor;
if (pCtx->periodMs > tsRedirectMaxPeriod) {
pCtx->periodMs = tsRedirectMaxPeriod;
}
pTask->delayExecMs = pCtx->periodMs;
int64_t leftTime = tsMaxRetryWaitTime - lastTime;
pTask->delayExecMs = leftTime < pCtx->periodMs ? leftTime : pCtx->periodMs;
goto _return;
}
......@@ -410,7 +412,7 @@ int32_t schDoTaskRedirect(SSchJob *pJob, SSchTask *pTask, SDataBuf *pData, int32
pTask->retryTimes = 0;
}
SCH_ERR_JRET(schChkUpdateRedirectCtx(pTask, pData ? pData->pEpSet : NULL));
SCH_ERR_JRET(schChkUpdateRedirectCtx(pJob, pTask, pData ? pData->pEpSet : NULL));
pTask->waitRetry = true;
......@@ -431,6 +433,10 @@ int32_t schDoTaskRedirect(SSchJob *pJob, SSchTask *pTask, SDataBuf *pData, int32
SQueryNodeAddr *addr = taosArrayGet(pTask->candidateAddrs, pTask->candidateIdx);
SCH_SWITCH_EPSET(addr);
SCH_TASK_DLOG("switch task target node %d epset to %d/%d", addr->nodeId, addr->epSet.inUse, addr->epSet.numOfEps);
} else {
SQueryNodeAddr *addr = taosArrayGet(pTask->candidateAddrs, pTask->candidateIdx);
SEp *pEp = &addr->epSet.eps[addr->epSet.inUse];
SCH_TASK_DLOG("task retry node %d current ep, idx:%d/%d,%s:%d", addr->nodeId, addr->epSet.inUse, addr->epSet.numOfEps, pEp->fqdn, pEp->port);
}
if (SCH_TASK_NEED_FLOW_CTRL(pJob, pTask)) {
......@@ -1141,15 +1147,13 @@ void schHandleTimerEvent(void *param, void *tmrId) {
SSchJob *pJob = NULL;
int32_t code = 0;
SCH_ERR_RET(schProcessOnCbBegin(&pJob, &pTask, pTimerParam->queryId, pTimerParam->rId, pTimerParam->taskId));
SCH_ERR_JRET(schLaunchTask(pJob, pTask));
if (schProcessOnCbBegin(&pJob, &pTask, pTimerParam->queryId, pTimerParam->rId, pTimerParam->taskId)) {
return;
}
_return:
code = schLaunchTask(pJob, pTask);
schHandleJobFailure(pJob, code);
schProcessOnCbEnd(pJob, pTask, code);
}
int32_t schDelayLaunchTask(SSchJob *pJob, SSchTask *pTask) {
......@@ -1157,7 +1161,7 @@ int32_t schDelayLaunchTask(SSchJob *pJob, SSchTask *pTask) {
SSchTimerParam *param = taosMemoryMalloc(sizeof(SSchTimerParam));
if (NULL == param) {
SCH_TASK_ELOG("taosMemoryMalloc %d failed", sizeof(SSchTimerParam));
QW_ERR_RET(TSDB_CODE_QRY_OUT_OF_MEMORY);
SCH_ERR_RET(TSDB_CODE_QRY_OUT_OF_MEMORY);
}
param->rId = pJob->refId;
......@@ -1167,8 +1171,8 @@ int32_t schDelayLaunchTask(SSchJob *pJob, SSchTask *pTask) {
if (NULL == pTask->delayTimer) {
pTask->delayTimer = taosTmrStart(schHandleTimerEvent, pTask->delayExecMs, (void *)param, schMgmt.timer);
if (NULL == pTask->delayTimer) {
SCH_TASK_ELOG("start delay timer failed");
QW_ERR_RET(TSDB_CODE_QRY_OUT_OF_MEMORY);
SCH_TASK_ELOG("start delay timer failed, handle:%p", schMgmt.timer);
SCH_ERR_RET(TSDB_CODE_QRY_OUT_OF_MEMORY);
}
return TSDB_CODE_SUCCESS;
......@@ -1203,7 +1207,12 @@ void schDropTaskInHashList(SSchJob *pJob, SHashObj *list) {
while (pIter) {
SSchTask *pTask = *(SSchTask **)pIter;
SCH_LOCK_TASK(pTask);
if (pTask->delayTimer) {
taosTmrStopA(&pTask->delayTimer);
}
schDropTaskOnExecNode(pJob, pTask);
SCH_UNLOCK_TASK(pTask);
pIter = taosHashIterate(list, pIter);
}
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册