From a52c0cb5f23f62afb4e40adf20d0ee8b468e20c1 Mon Sep 17 00:00:00 2001
From: slzhou <shenglian_zhou@163.com>
Date: Mon, 19 Sep 2022 14:01:12 +0800
Subject: [PATCH] fix: improve group by performance by skip sort when group by

---
 src/common/inc/tglobal.h  |  2 ++
 src/common/src/tglobal.c  | 25 +++++++++++++++++++++++++
 src/query/src/qExecutor.c |  4 ++--
 src/util/inc/tconfig.h    |  2 +-
 4 files changed, 30 insertions(+), 3 deletions(-)

diff --git a/src/common/inc/tglobal.h b/src/common/inc/tglobal.h
index e1b7cff8be..650bb6bb4e 100644
--- a/src/common/inc/tglobal.h
+++ b/src/common/inc/tglobal.h
@@ -67,6 +67,8 @@ extern int32_t  tsCompressColData;
 extern int32_t  tsMaxNumOfDistinctResults;
 extern char     tsTempDir[];
 extern int32_t  tsShortcutFlag;
+extern int32_t  tsMaxSqlGroups;
+extern int8_t   tsSortWhenGroupBy;
 
 // query buffer management
 extern int32_t tsQueryBufferSize;  // maximum allowed usage buffer size in MB for each data node during query processing
diff --git a/src/common/src/tglobal.c b/src/common/src/tglobal.c
index 77540cd0b6..d9b254a26a 100644
--- a/src/common/src/tglobal.c
+++ b/src/common/src/tglobal.c
@@ -118,6 +118,12 @@ int32_t tsRetryStreamCompDelay = 30 * 60 * 1000;
 // The delayed computing ration. 10% of the whole computing time window by default.
 float tsStreamComputDelayRatio = 0.1f;
 
+// max supported groups for group by clause / interval clause
+int32_t tsMaxSqlGroups = 1000000;
+
+// order by first group by column when group by
+int8_t tsSortWhenGroupBy = 1;
+
 int32_t tsProjectExecInterval = 10000;   // every 10sec, the projection will be executed once
 int64_t tsMaxRetentWindow = 24 * 3600L;  // maximum time window tolerance
 
@@ -1772,6 +1778,25 @@ static void doInitGlobalConfig(void) {
   cfg.unitType = TAOS_CFG_UTYPE_NONE;
   taosInitConfigOption(cfg);
 
+  cfg.option = "maxSqlGroups";
+  cfg.ptr = &tsMaxSqlGroups;
+  cfg.valType = TAOS_CFG_VTYPE_INT32;
+  cfg.cfgType = TSDB_CFG_CTYPE_B_CONFIG;
+  cfg.minValue = 500000;
+  cfg.maxValue = 10000000;
+  cfg.ptrLength = 0;
+  cfg.unitType = TAOS_CFG_UTYPE_NONE;
+
+  cfg.option = "sortWhenGroupBy";
+  cfg.ptr = &tsSortWhenGroupBy;
+  cfg.valType = TAOS_CFG_VTYPE_INT8;
+  cfg.cfgType = TSDB_CFG_CTYPE_B_CONFIG;
+  cfg.minValue = 0;
+  cfg.maxValue = 1;
+  cfg.ptrLength = 0;
+  cfg.unitType = TAOS_CFG_UTYPE_NONE;
+  taosInitConfigOption(cfg);
+
 #ifdef TD_TSZ
   // lossy compress
   cfg.option = "lossyColumns";
diff --git a/src/query/src/qExecutor.c b/src/query/src/qExecutor.c
index bb9ded8e40..47334ab81e 100644
--- a/src/query/src/qExecutor.c
+++ b/src/query/src/qExecutor.c
@@ -621,7 +621,7 @@ static SResultRow* doSetResultOutBufByKey(SQueryRuntimeEnv* pRuntimeEnv, SResult
   }
 
   // too many time window in query
-  if (pResultRowInfo->size > MAX_INTERVAL_TIME_WINDOW) {
+  if (pResultRowInfo->size > tsMaxSqlGroups) {
     longjmp(pRuntimeEnv->env, TSDB_CODE_QRY_TOO_MANY_TIMEWINDOW);
   }
 
@@ -7607,7 +7607,7 @@ static SSDataBlock* hashGroupbyAggregate(void* param, bool* newgroup) {
   }
 
   initGroupResInfo(&pRuntimeEnv->groupResInfo, &pInfo->binfo.resultRowInfo);
-  if (!pRuntimeEnv->pQueryAttr->stableQuery) {
+  if (!pRuntimeEnv->pQueryAttr->stableQuery && tsSortWhenGroupBy) {
     sortGroupResByOrderList(&pRuntimeEnv->groupResInfo, pRuntimeEnv, pInfo->binfo.pRes, pInfo->binfo.pCtx);
   }
 
diff --git a/src/util/inc/tconfig.h b/src/util/inc/tconfig.h
index 872da82a8e..a2c2937722 100644
--- a/src/util/inc/tconfig.h
+++ b/src/util/inc/tconfig.h
@@ -20,7 +20,7 @@
 extern "C" {
 #endif
 
-#define TSDB_CFG_MAX_NUM    134
+#define TSDB_CFG_MAX_NUM    136
 #define TSDB_CFG_PRINT_LEN  23
 #define TSDB_CFG_OPTION_LEN 24
 #define TSDB_CFG_VALUE_LEN  41
-- 
GitLab