fix: distinct wrong result issue

source/libs/planner/src/planLogicCreater.c

@@ -544,11 +544,16 @@ static SNode* createGroupingSetNode(SNode* pExpr) {
   return (SNode*)pGroupingSet;
 }
 
-static EGroupAction getGroupAction(SLogicPlanContext* pCxt, SSelectStmt* pSelect) {
+static EGroupAction getDistinctGroupAction(SLogicPlanContext* pCxt, SSelectStmt* pSelect) {
   return (pCxt->pPlanCxt->streamQuery || NULL != pSelect->pLimit || NULL != pSelect->pSlimit) ? GROUP_ACTION_KEEP
                                                                                               : GROUP_ACTION_NONE;
 }
 
+static EGroupAction getGroupAction(SLogicPlanContext* pCxt, SSelectStmt* pSelect) {
+  return ((pCxt->pPlanCxt->streamQuery || NULL != pSelect->pLimit || NULL != pSelect->pSlimit) && !pSelect->isDistinct) ? GROUP_ACTION_KEEP
+                                                                                              : GROUP_ACTION_NONE;
+}
+
 static EDataOrderLevel getRequireDataOrder(bool needTimeline, SSelectStmt* pSelect) {
   return needTimeline ? (NULL != pSelect->pPartitionByList ? DATA_ORDER_LEVEL_IN_GROUP : DATA_ORDER_LEVEL_GLOBAL)
                       : DATA_ORDER_LEVEL_NONE;
@@ -1152,7 +1157,7 @@ static int32_t createDistinctLogicNode(SLogicPlanContext* pCxt, SSelectStmt* pSe
     return TSDB_CODE_OUT_OF_MEMORY;
   }
 
-  pAgg->node.groupAction = GROUP_ACTION_CLEAR;
+  pAgg->node.groupAction = getDistinctGroupAction(pCxt, pSelect);
   pAgg->node.requireDataOrder = DATA_ORDER_LEVEL_NONE;
   pAgg->node.resultDataOrder = DATA_ORDER_LEVEL_NONE;
 

tests/parallel_test/cases.task

@@ -900,6 +900,7 @@
 ,,y,script,./test.sh -f tsim/query/multi_order_by.sim
 ,,y,script,./test.sh -f tsim/query/sys_tbname.sim
 ,,y,script,./test.sh -f tsim/query/groupby.sim
+,,y,script,./test.sh -f tsim/query/groupby_distinct.sim
 ,,y,script,./test.sh -f tsim/query/event.sim
 ,,y,script,./test.sh -f tsim/query/forceFill.sim
 ,,y,script,./test.sh -f tsim/query/emptyTsRange.sim

tests/script/tsim/query/groupby_distinct.sim

+system sh/stop_dnodes.sh
+system sh/deploy.sh -n dnode1 -i 1
+system sh/exec.sh -n dnode1 -s start
+sql connect
+
+
+sql drop database if exists db1;
+sql create database db1;
+sql use db1;
+
+sql create stable sta (ts timestamp, f1 int, f2 binary(200)) tags(t1 int, t2 int, t3 int);
+sql create table tba1 using sta tags(1, 1, 1);
+sql insert into tba1 values ('2022-04-26 15:15:08', 1, "a");
+sql insert into tba1 values ('2022-04-26 15:15:07', 1, "b");
+sql insert into tba1 values ('2022-04-26 15:15:06', 1, "a");
+sql insert into tba1 values ('2022-04-26 15:15:05', 1, "b");
+sql insert into tba1 values ('2022-04-26 15:15:04', 1, "c");
+sql insert into tba1 values ('2022-04-26 15:15:03', 1, "c");
+sql insert into tba1 values ('2022-04-26 15:15:02', 1, "d");
+sql insert into tba1 values ('2022-04-26 15:15:01', 1, "d");
+sql select distinct avg(f1) as avgv from sta group by f2;
+if $rows != 1 then
+  return -1
+endi
+sql select distinct avg(f1) as avgv from sta group by f2 limit 1,10;
+if $rows != 0 then
+  return -1
+endi
+
+system sh/exec.sh -n dnode1 -s stop -x SIGINT