Merge pull request #771 from taosdata/feature/liaohj

Feature/liaohj

src/client/src/tscServer.c

@@ -59,6 +59,22 @@ void tscPrintMgmtIp() {
 }
 #endif
 
+/*
+ * For each management node, try twice at least in case of poor network situation.
+ * If the client start to connect to a non-management node from the client, and the first retry may fail due to
+ * the poor network quality. And then, the second retry get the response with redirection command.
+ * The retry will not be executed since only *two* retry is allowed in case of single management node in the cluster.
+ * Therefore, we need to multiply the retry times by factor of 2 to fix this problem.
+ */
+static int32_t tscGetMgmtConnMaxRetryTimes() {
+  int32_t factor = 2;
+#ifdef CLUSTER
+  return tscMgmtIpList.numOfIps * factor;
+#else
+  return 1*factor;
+#endif
+}
+
 void tscProcessHeartBeatRsp(void *param, TAOS_RES *tres, int code) {
   STscObj *pObj = (STscObj *)param;
   if (pObj == NULL) return;
@@ -134,18 +150,17 @@ void tscProcessActivityTimer(void *handle, void *tmrId) {
   tscProcessSql(pObj->pHb);
 }
 
-//TODO HANDLE error from mgmt
 void tscGetConnToMgmt(SSqlObj *pSql, uint8_t *pCode) {
   STscObj *pTscObj = pSql->pTscObj;
 #ifdef CLUSTER
-  if (pSql->retry < tscMgmtIpList.numOfIps) {
+  if (pSql->retry < tscGetMgmtConnMaxRetryTimes()) {
     *pCode = 0;
     pSql->retry++;
     pSql->index = pSql->index % tscMgmtIpList.numOfIps;
     if (pSql->cmd.command > TSDB_SQL_READ && pSql->index == 0) pSql->index = 1;
     void *thandle = taosGetConnFromCache(tscConnCache, tscMgmtIpList.ip[pSql->index], TSC_MGMT_VNODE, pTscObj->user);
 #else
-  if (pSql->retry < 1) {
+  if (pSql->retry < tscGetMgmtConnMaxRetryTimes()) {
     *pCode = 0;
     pSql->retry++;
     void *thandle = taosGetConnFromCache(tscConnCache, tsServerIp, TSC_MGMT_VNODE, pTscObj->user);
@@ -444,17 +459,14 @@ void *tscProcessMsgFromServer(char *msg, void *ahandle, void *thandle) {
     }
   } else {
     uint16_t rspCode = pMsg->content[0];
-#ifdef CLUSTER
     
+#ifdef CLUSTER
     
     if (rspCode == TSDB_CODE_REDIRECT) {
       tscTrace("%p it shall be redirected!", pSql);
       taosAddConnIntoCache(tscConnCache, thandle, pSql->ip, pSql->vnode, pObj->user);
       pSql->thandle = NULL;
 
-      // reset the retry times for a new mgmt node
-      pSql->retry = 0;
-
       if (pCmd->command > TSDB_SQL_MGMT) {
         tscProcessMgmtRedirect(pSql, pMsg->content + 1);
       } else {

src/system/detail/src/vnodeQueryImpl.c

@@ -256,9 +256,9 @@ static void doUnmapHeaderFileData(SQueryRuntimeEnv* pRuntimeEnv) {
     
     otherVnodeFiles->pHeaderFileData = NULL;
     pRuntimeEnv->mmapedHFileIndex = -1;
-  } else {
-    assert(pRuntimeEnv->mmapedHFileIndex == -1);
   }
+
+  assert(pRuntimeEnv->mmapedHFileIndex == -1);
 }
 
 /**
@@ -286,12 +286,12 @@ char *vnodeGetHeaderFileData(SQueryRuntimeEnv *pRuntimeEnv, int32_t fileIndex) {
     pVnodeFiles->pHeaderFileData = mmap(NULL, size, PROT_READ, MAP_SHARED, pVnodeFiles->headerFd, 0);
     if (pVnodeFiles->pHeaderFileData == MAP_FAILED) {
       pVnodeFiles->pHeaderFileData = NULL;
-      dError("QInfo:%p failed to map header file:%s, size:%lld, %s", pQInfo, pVnodeFiles->headerFilePath, size,
+      dError("QInfo:%p failed to mmap header file:%s, size:%lld, %s", pQInfo, pVnodeFiles->headerFilePath, size,
              strerror(errno));
     } else {
       pRuntimeEnv->mmapedHFileIndex = fileIndex;  // set the value in case of success mmap file
       if (madvise(pVnodeFiles->pHeaderFileData, size, MADV_SEQUENTIAL) == -1) {
-        dError("QInfo:%p failed to advise kernel the usage of header files, reason:%s", pQInfo, strerror(errno));
+        dError("QInfo:%p failed to advise kernel the usage of header file, reason:%s", pQInfo, strerror(errno));
       }
     }
   } else {
@@ -326,9 +326,9 @@ static int vnodeGetCompBlockInfo(SMeterObj *pMeterObj, SQueryRuntimeEnv *pRuntim
   pSummary->numOfSeek++;
 
 #if 1
-  char *data = vnodeGetHeaderFileData(pRuntimeEnv, fileIndex);  // failed to load the header file data into memory
+  char *data = vnodeGetHeaderFileData(pRuntimeEnv, fileIndex);
   if (data == NULL) {
-    return -1;
+    return -1;  // failed to load the header file data into memory
   }
   
 #else
@@ -2842,9 +2842,8 @@ int64_t loadRequiredBlockIntoMem(SQueryRuntimeEnv *pRuntimeEnv, SPositionInfo *p
      * currently opened file is not the start file, reset to the start file
      */
     int32_t fileIdx = vnodeGetVnodeHeaderFileIdx(&pQuery->fileId, pRuntimeEnv, pQuery->order.order);
-    if (fileIdx < 0) {
+    if (fileIdx < 0) { // ignore the files on disk
       dError("QInfo:%p failed to get data file:%d", GET_QINFO_ADDR(pQuery), pQuery->fileId);
-      // ignore the files on disk
       position->fileId = -1;
       return -1;
     }
@@ -5494,11 +5493,15 @@ SMeterDataInfo **vnodeFilterQualifiedMeters(SQInfo *pQInfo, int32_t vid, int32_t
 
   SVnodeObj *pVnode = &vnodeList[vid];
 
-  char *  pHeaderData = vnodeGetHeaderFileData(pRuntimeEnv, fileIndex);
+  char *  pHeaderFileData = vnodeGetHeaderFileData(pRuntimeEnv, fileIndex);
+  if (pHeaderFileData == NULL) { // failed to load header file into buffer
+    return 0;
+  }
+  
   int32_t tmsize = sizeof(SCompHeader) * (pVnode->cfg.maxSessions) + sizeof(TSCKSUM);
 
   // file is corrupted, abort query in current file
-  if (validateHeaderOffsetSegment(pQInfo, pQueryFileInfo->headerFilePath, vid, pHeaderData, tmsize) < 0) {
+  if (validateHeaderOffsetSegment(pQInfo, pQueryFileInfo->headerFilePath, vid, pHeaderFileData, tmsize) < 0) {
     *numOfMeters = 0;
     return 0;
   }
@@ -5549,7 +5552,7 @@ SMeterDataInfo **vnodeFilterQualifiedMeters(SQInfo *pQInfo, int32_t vid, int32_t
 
     int64_t headerOffset = TSDB_FILE_HEADER_LEN + sizeof(SCompHeader) * pMeterObj->sid;
 
-    SCompHeader *compHeader = (SCompHeader *)(pHeaderData + headerOffset);
+    SCompHeader *compHeader = (SCompHeader *)(pHeaderFileData + headerOffset);
 
     if (compHeader->compInfoOffset == 0) {
       continue;

src/system/detail/src/vnodeQueryProcess.c

@@ -290,11 +290,14 @@ static SMeterDataInfo *queryOnMultiDataFiles(SQInfo *pQInfo, SMeterDataInfo *pMe
     pSummary->numOfFiles++;
 
     SQueryFileInfo *pQueryFileInfo = &pRuntimeEnv->pVnodeFiles[fileIdx];
+    char *pHeaderData = vnodeGetHeaderFileData(pRuntimeEnv, fileIdx);
+    if (pHeaderData == NULL) { // failed to mmap header file into buffer, ignore current file, try next
+      continue;
+    }
     
     int32_t          numOfQualifiedMeters = 0;
     SMeterDataInfo **pReqMeterDataInfo = vnodeFilterQualifiedMeters(pQInfo, vnodeId, fileIdx, pSupporter->pSidSet,
         pMeterDataInfo, &numOfQualifiedMeters);
-    dTrace("QInfo:%p file:%s, %d meters qualified", pQInfo, pQueryFileInfo->dataFilePath, numOfQualifiedMeters);
 
     if (pReqMeterDataInfo == NULL) {
       dError("QInfo:%p failed to allocate memory to perform query processing, abort", pQInfo);
@@ -304,6 +307,8 @@ static SMeterDataInfo *queryOnMultiDataFiles(SQInfo *pQInfo, SMeterDataInfo *pMe
       return NULL;
     }
 
+    dTrace("QInfo:%p file:%s, %d meters qualified", pQInfo, pQueryFileInfo->dataFilePath, numOfQualifiedMeters);
+    
     // none of meters in query set have pHeaderData in this file, try next file
     if (numOfQualifiedMeters == 0) {
       fid += step;
@@ -311,11 +316,6 @@ static SMeterDataInfo *queryOnMultiDataFiles(SQInfo *pQInfo, SMeterDataInfo *pMe
       continue;
     }
 
-    char *pHeaderData = vnodeGetHeaderFileData(pRuntimeEnv, fileIdx);
-    if (pHeaderData == NULL) { // failed to mmap header file into buffer
-      continue;
-    }
-    
     uint32_t numOfBlocks = getDataBlocksForMeters(pSupporter, pQuery, pHeaderData, numOfQualifiedMeters, pQueryFileInfo,
                                                   pReqMeterDataInfo);
 

src/util/src/tglobalcfg.c

@@ -510,7 +510,7 @@ static void doInitGlobalConfig() {
                      0, TSDB_MAX_VNODES, 0, TSDB_CFG_UTYPE_NONE);
   tsInitConfigOption(cfg++, "tables", &tsSessionsPerVnode, TSDB_CFG_VTYPE_INT,
                      TSDB_CFG_CTYPE_B_CONFIG | TSDB_CFG_CTYPE_B_SHOW,
-                     4, 220000, 0, TSDB_CFG_UTYPE_NONE);
+                     TSDB_MIN_TABLES_PER_VNODE, TSDB_MAX_TABLES_PER_VNODE, 0, TSDB_CFG_UTYPE_NONE);
   tsInitConfigOption(cfg++, "cache", &tsCacheBlockSize, TSDB_CFG_VTYPE_INT,
                      TSDB_CFG_CTYPE_B_CONFIG | TSDB_CFG_CTYPE_B_SHOW,
                      100, 1048576, 0, TSDB_CFG_UTYPE_BYTE);