Merge pull request #1888 from citusdata/subplan_explain

Show distributed subplan ID in EXPLAIN output

src/backend/distributed/planner/multi_explain.c

@@ -72,7 +72,7 @@ typedef struct RemoteExplainPlan
 
 
 /* Explain functions for distributed queries */
-static void ExplainSubPlans(List *subPlanList, ExplainState *es);
+static void ExplainSubPlans(DistributedPlan *distributedPlan, ExplainState *es);
 static void ExplainJob(Job *job, ExplainState *es);
 static void ExplainMapMergeJob(MapMergeJob *mapMergeJob, ExplainState *es);
 static void ExplainTaskList(List *taskList, ExplainState *es);
@@ -127,7 +127,7 @@ CitusExplainScan(CustomScanState *node, List *ancestors, struct ExplainState *es
 
 	if (distributedPlan->subPlanList != NIL)
 	{
-		ExplainSubPlans(distributedPlan->subPlanList, es);
+		ExplainSubPlans(distributedPlan, es);
 	}
 
 	ExplainJob(distributedPlan->workerJob, es);
@@ -179,20 +179,14 @@ CoordinatorInsertSelectExplainScan(CustomScanState *node, List *ancestors,
  * planning time and set it to 0.
  */
 static void
-ExplainSubPlans(List *subPlanList, ExplainState *es)
+ExplainSubPlans(DistributedPlan *distributedPlan, ExplainState *es)
 {
 	ListCell *subPlanCell = NULL;
+	uint64 planId = distributedPlan->planId;
 
 	ExplainOpenGroup("Subplans", "Subplans", false, es);
 
-	if (es->format == EXPLAIN_FORMAT_TEXT)
+	foreach(subPlanCell, distributedPlan->subPlanList)
-	{
-		appendStringInfoSpaces(es->str, es->indent * 2);
-		appendStringInfo(es->str, "->  Distributed Subplan\n");
-		es->indent += 3;
-	}
-
-	foreach(subPlanCell, subPlanList)
 	{
 		DistributedSubPlan *subPlan = (DistributedSubPlan *) lfirst(subPlanCell);
 		PlannedStmt *plan = subPlan->plan;
@@ -201,6 +195,15 @@ ExplainSubPlans(List *subPlanList, ExplainState *es)
 		char *queryString = NULL;
 		instr_time planduration;
 
+		if (es->format == EXPLAIN_FORMAT_TEXT)
+		{
+			char *resultId = GenerateResultId(planId, subPlan->subPlanId);
+
+			appendStringInfoSpaces(es->str, es->indent * 2);
+			appendStringInfo(es->str, "->  Distributed Subplan %s\n", resultId);
+			es->indent += 3;
+		}
+
 		/* set the planning time to 0 */
 		INSTR_TIME_SET_CURRENT(planduration);
 		INSTR_TIME_SUBTRACT(planduration, planduration);
@@ -210,12 +213,12 @@ ExplainSubPlans(List *subPlanList, ExplainState *es)
 #else
 		ExplainOnePlan(plan, into, es, queryString, params, &planduration);
 #endif
-	}
 
 		if (es->format == EXPLAIN_FORMAT_TEXT)
 		{
 			es->indent -= 3;
 		}
+	}
 
 	ExplainCloseGroup("Subplans", "Subplans", false, es);
 }

src/backend/distributed/planner/recursive_planning.c

@@ -140,6 +140,17 @@ RecursivelyPlanSubqueriesAndCTEs(Query *query,
 	DeferredErrorMessage *error = NULL;
 	RecursivePlanningContext context;
 
+	context.level = 0;
+	context.planId = planId;
+	context.subPlanList = NIL;
+	context.plannerRestrictionContext = plannerRestrictionContext;
+
+	error = RecursivelyPlanCTEs(query, &context);
+	if (error != NULL)
+	{
+		return error;
+	}
+
 	if (SubqueryPushdown)
 	{
 		/*
@@ -155,17 +166,6 @@ RecursivelyPlanSubqueriesAndCTEs(Query *query,
 		return NULL;
 	}
 
-	context.level = 0;
-	context.planId = planId;
-	context.subPlanList = NIL;
-	context.plannerRestrictionContext = plannerRestrictionContext;
-
-	error = RecursivelyPlanCTEs(query, &context);
-	if (error != NULL)
-	{
-		return error;
-	}
-
 	/* XXX: plan subqueries */
 
 	*subPlanList = context.subPlanList;

src/test/regress/expected/multi_explain.out

@@ -713,6 +713,7 @@ Limit
                                                               Sort Key: events.event_time DESC
                                                               ->  Seq Scan on events_1400029 events
                                                                     Filter: (composite_id = users.composite_id)
+RESET citus.subquery_pushdown;
 -- Test all tasks output
 SET citus.explain_all_tasks TO on;
 EXPLAIN (COSTS FALSE)
@@ -1115,3 +1116,75 @@ Custom Scan (Citus INSERT ... SELECT via coordinator)
         ->  Append
               ->  Function Scan on generate_series s
               ->  Function Scan on generate_series s_1
+-- explain with recursive planning
+EXPLAIN (COSTS OFF, VERBOSE true)
+WITH keys AS (
+  SELECT DISTINCT l_orderkey FROM lineitem_hash_part
+),
+series AS (
+  SELECT s FROM generate_series(1,10) s
+)
+SELECT l_orderkey FROM series JOIN keys ON (s = l_orderkey)
+ORDER BY s;
+Custom Scan (Citus Router)
+  Output: remote_scan.l_orderkey
+  ->  Distributed Subplan 54_1
+        ->  HashAggregate
+              Output: remote_scan.l_orderkey
+              Group Key: remote_scan.l_orderkey
+              ->  Custom Scan (Citus Real-Time)
+                    Output: remote_scan.l_orderkey
+                    Task Count: 4
+                    Tasks Shown: One of 4
+                    ->  Task
+                          Node: host=localhost port=57637 dbname=regression
+                          ->  HashAggregate
+                                Output: l_orderkey
+                                Group Key: lineitem_hash_part.l_orderkey
+                                ->  Seq Scan on public.lineitem_hash_part_360038 lineitem_hash_part
+                                      Output: l_orderkey, l_partkey, l_suppkey, l_linenumber, l_quantity, l_extendedprice, l_discount, l_tax, l_returnflag, l_linestatus, l_shipdate, l_commitdate, l_receiptdate, l_shipinstruct, l_shipmode, l_comment
+  ->  Distributed Subplan 54_2
+        ->  Function Scan on pg_catalog.generate_series s
+              Output: s
+              Function Call: generate_series(1, 10)
+  Task Count: 1
+  Tasks Shown: All
+  ->  Task
+        Node: host=localhost port=57638 dbname=regression
+        ->  Merge Join
+              Output: intermediate_result_1.l_orderkey, intermediate_result.s
+              Merge Cond: (intermediate_result.s = intermediate_result_1.l_orderkey)
+              ->  Sort
+                    Output: intermediate_result.s
+                    Sort Key: intermediate_result.s
+                    ->  Function Scan on pg_catalog.read_intermediate_result intermediate_result
+                          Output: intermediate_result.s
+                          Function Call: read_intermediate_result('54_2'::text, 'binary'::citus_copy_format)
+              ->  Sort
+                    Output: intermediate_result_1.l_orderkey
+                    Sort Key: intermediate_result_1.l_orderkey
+                    ->  Function Scan on pg_catalog.read_intermediate_result intermediate_result_1
+                          Output: intermediate_result_1.l_orderkey
+                          Function Call: read_intermediate_result('54_1'::text, 'binary'::citus_copy_format)
+SELECT true AS valid FROM explain_json($$
+  WITH result AS (
+    SELECT l_quantity, count(*) count_quantity FROM lineitem
+	GROUP BY l_quantity ORDER BY count_quantity, l_quantity
+  ),
+  series AS (
+    SELECT s FROM generate_series(1,10) s
+  )
+  SELECT * FROM result JOIN series ON (s = count_quantity) JOIN orders_hash_part ON (s = o_orderkey)
+$$);
+t
+SELECT true AS valid FROM explain_xml($$
+  WITH result AS (
+    SELECT l_quantity, count(*) count_quantity FROM lineitem
+	GROUP BY l_quantity ORDER BY count_quantity, l_quantity
+  ),
+  series AS (
+    SELECT s FROM generate_series(1,10) s
+  )
+  SELECT * FROM result JOIN series ON (s = l_quantity) JOIN orders_hash_part ON (s = o_orderkey)
+$$);
+t

src/test/regress/expected/multi_explain_0.out

@@ -713,6 +713,7 @@ Limit
                                                               Sort Key: events.event_time DESC
                                                               ->  Seq Scan on events_1400029 events
                                                                     Filter: (composite_id = users.composite_id)
+RESET citus.subquery_pushdown;
 -- Test all tasks output
 SET citus.explain_all_tasks TO on;
 EXPLAIN (COSTS FALSE)
@@ -1115,3 +1116,75 @@ Custom Scan (Citus INSERT ... SELECT via coordinator)
         ->  Append
               ->  Function Scan on generate_series s
               ->  Function Scan on generate_series s_1
+-- explain with recursive planning
+EXPLAIN (COSTS OFF, VERBOSE true)
+WITH keys AS (
+  SELECT DISTINCT l_orderkey FROM lineitem_hash_part
+),
+series AS (
+  SELECT s FROM generate_series(1,10) s
+)
+SELECT l_orderkey FROM series JOIN keys ON (s = l_orderkey)
+ORDER BY s;
+Custom Scan (Citus Router)
+  Output: remote_scan.l_orderkey
+  ->  Distributed Subplan 54_1
+        ->  HashAggregate
+              Output: remote_scan.l_orderkey
+              Group Key: remote_scan.l_orderkey
+              ->  Custom Scan (Citus Real-Time)
+                    Output: remote_scan.l_orderkey
+                    Task Count: 4
+                    Tasks Shown: One of 4
+                    ->  Task
+                          Node: host=localhost port=57637 dbname=regression
+                          ->  HashAggregate
+                                Output: l_orderkey
+                                Group Key: lineitem_hash_part.l_orderkey
+                                ->  Seq Scan on public.lineitem_hash_part_360038 lineitem_hash_part
+                                      Output: l_orderkey, l_partkey, l_suppkey, l_linenumber, l_quantity, l_extendedprice, l_discount, l_tax, l_returnflag, l_linestatus, l_shipdate, l_commitdate, l_receiptdate, l_shipinstruct, l_shipmode, l_comment
+  ->  Distributed Subplan 54_2
+        ->  Function Scan on pg_catalog.generate_series s
+              Output: s
+              Function Call: generate_series(1, 10)
+  Task Count: 1
+  Tasks Shown: All
+  ->  Task
+        Node: host=localhost port=57638 dbname=regression
+        ->  Merge Join
+              Output: intermediate_result_1.l_orderkey, intermediate_result.s
+              Merge Cond: (intermediate_result.s = intermediate_result_1.l_orderkey)
+              ->  Sort
+                    Output: intermediate_result.s
+                    Sort Key: intermediate_result.s
+                    ->  Function Scan on pg_catalog.read_intermediate_result intermediate_result
+                          Output: intermediate_result.s
+                          Function Call: read_intermediate_result('54_2'::text, 'binary'::citus_copy_format)
+              ->  Sort
+                    Output: intermediate_result_1.l_orderkey
+                    Sort Key: intermediate_result_1.l_orderkey
+                    ->  Function Scan on pg_catalog.read_intermediate_result intermediate_result_1
+                          Output: intermediate_result_1.l_orderkey
+                          Function Call: read_intermediate_result('54_1'::text, 'binary'::citus_copy_format)
+SELECT true AS valid FROM explain_json($$
+  WITH result AS (
+    SELECT l_quantity, count(*) count_quantity FROM lineitem
+	GROUP BY l_quantity ORDER BY count_quantity, l_quantity
+  ),
+  series AS (
+    SELECT s FROM generate_series(1,10) s
+  )
+  SELECT * FROM result JOIN series ON (s = count_quantity) JOIN orders_hash_part ON (s = o_orderkey)
+$$);
+t
+SELECT true AS valid FROM explain_xml($$
+  WITH result AS (
+    SELECT l_quantity, count(*) count_quantity FROM lineitem
+	GROUP BY l_quantity ORDER BY count_quantity, l_quantity
+  ),
+  series AS (
+    SELECT s FROM generate_series(1,10) s
+  )
+  SELECT * FROM result JOIN series ON (s = l_quantity) JOIN orders_hash_part ON (s = o_orderkey)
+$$);
+t

src/test/regress/sql/multi_explain.sql

@@ -356,6 +356,8 @@ ORDER BY
 LIMIT
 	10;
 
+RESET citus.subquery_pushdown;
+
 -- Test all tasks output
 SET citus.explain_all_tasks TO on;
 
@@ -519,3 +521,36 @@ EXPLAIN (COSTS OFF)
 INSERT INTO lineitem_hash_part
 ( SELECT s FROM generate_series(1,5) s) UNION
 ( SELECT s FROM generate_series(5,10) s);
+
+-- explain with recursive planning
+EXPLAIN (COSTS OFF, VERBOSE true)
+WITH keys AS (
+  SELECT DISTINCT l_orderkey FROM lineitem_hash_part
+),
+series AS (
+  SELECT s FROM generate_series(1,10) s
+)
+SELECT l_orderkey FROM series JOIN keys ON (s = l_orderkey)
+ORDER BY s;
+
+SELECT true AS valid FROM explain_json($$
+  WITH result AS (
+    SELECT l_quantity, count(*) count_quantity FROM lineitem
+	GROUP BY l_quantity ORDER BY count_quantity, l_quantity
+  ),
+  series AS (
+    SELECT s FROM generate_series(1,10) s
+  )
+  SELECT * FROM result JOIN series ON (s = count_quantity) JOIN orders_hash_part ON (s = o_orderkey)
+$$);
+
+SELECT true AS valid FROM explain_xml($$
+  WITH result AS (
+    SELECT l_quantity, count(*) count_quantity FROM lineitem
+	GROUP BY l_quantity ORDER BY count_quantity, l_quantity
+  ),
+  series AS (
+    SELECT s FROM generate_series(1,10) s
+  )
+  SELECT * FROM result JOIN series ON (s = l_quantity) JOIN orders_hash_part ON (s = o_orderkey)
+$$);