Remove default statistics control in join planning

Commit e0409357 moved to using default estimates rather
than interrogating the QEs for relations which lack statistics.
As an effect of this, the cdb_default_stats_used member was
hardcoded to false and the warnings for missing statistics did
never fire.

Rather than resurrecting the warnings, this removes the code
that attempts to figure out if the warnings at all apply since
it seems quite expensive to run that in the hot path of every
join query being planned.

Discussion: https://github.com/greenplum-db/gpdb/pull/5216Reviewed-by: NVenkatesh Raghavan <vraghavan@pivotal.io>

src/backend/nodes/outfast.c

@@ -672,7 +672,6 @@ _outIndexOptInfo(StringInfo str, IndexOptInfo *node)
 	WRITE_BOOL_FIELD(hypothetical);
 
 	WRITE_BOOL_FIELD(amoptionalkey);
-	WRITE_BOOL_FIELD(cdb_default_stats_used);
 }
 
 /*****************************************************************************

src/backend/nodes/outfuncs.c

@@ -2338,7 +2338,6 @@ _outIndexOptInfo(StringInfo str, IndexOptInfo *node)
 	WRITE_BOOL_FIELD(hypothetical);
 
 	WRITE_BOOL_FIELD(amoptionalkey);
-	WRITE_BOOL_FIELD(cdb_default_stats_used);
 }
 #endif /* COMPILING_BINARY_FUNCS */
 

src/backend/optimizer/path/allpaths.c

@@ -112,27 +112,6 @@ make_one_rel(PlannerInfo *root, List *joinlist)
 	 */
 	set_base_rel_pathlists(root);
 
-	/*
-	 * CDB: If join, warn of any tables that need ANALYZE.
-	 */
-	if (has_multiple_baserels(root))
-	{
-		Index		rti;
-		RelOptInfo *brel;
-		RangeTblEntry *brte;
-
-		for (rti = 1; rti < root->simple_rel_array_size; rti++)
-		{
-			brel = root->simple_rel_array[rti];
-			if (brel &&
-				brel->cdb_default_stats_used)
-			{
-				brte = rt_fetch(rti, root->parse->rtable);
-				cdb_default_stats_warning_for_table(brte->relid);
-			}
-		}
-	}
-
 	/*
 	 * Generate access paths for the entire join tree.
 	 */

src/backend/optimizer/util/plancat.c

@@ -65,11 +65,7 @@ cdb_estimate_rel_size(RelOptInfo   *relOptInfo,
                       Relation      rel,
                       int32        *attr_widths,
 				      BlockNumber  *pages,
-                      double       *tuples,
-                      bool         *default_stats_used);
-
-static void
-cdb_default_stats_warning_for_index(Oid reloid, Oid indexoid);
+                      double       *tuples);
 
 static void get_external_relation_info(Relation relation, RelOptInfo *rel);
 
@@ -155,9 +151,7 @@ get_relation_info(PlannerInfo *root, Oid relationObjectId, bool inhparent,
 			relation,
 			rel->attr_widths - rel->min_attr,
 			&rel->pages,
-			&rel->tuples,
-			&rel->cdb_default_stats_used
-			);
+			&rel->tuples);
 	}
 
 	/*
@@ -176,10 +170,6 @@ get_relation_info(PlannerInfo *root, Oid relationObjectId, bool inhparent,
 		ListCell   *l;
 		LOCKMODE	lmode;
 
-        /* Warn if indexed table needs ANALYZE. */
-        if (rel->cdb_default_stats_used)
-            cdb_default_stats_warning_for_table(relation->rd_id);
-
 		indexoidlist = RelationGetIndexList(relation);
 
 		/*
@@ -379,17 +369,12 @@ get_relation_info(PlannerInfo *root, Oid relationObjectId, bool inhparent,
                                   indexRelation,
                                   NULL,
                                   &info->pages,
-                                  &info->tuples,
-                                  &info->cdb_default_stats_used);
+                                  &info->tuples);
 
 			if (!info->indpred ||
 				info->tuples > rel->tuples)
 				info->tuples = rel->tuples;
 
-            if (info->cdb_default_stats_used &&
-                !rel->cdb_default_stats_used)
-                cdb_default_stats_warning_for_index(relation->rd_id, indexoid);
-
 			index_close(indexRelation, needs_longlock ? NoLock : lmode);
 
 			indexinfos = lcons(info, indexinfos);
@@ -442,16 +427,13 @@ cdb_estimate_rel_size(RelOptInfo   *relOptInfo,
                       Relation      rel,
                       int32        *attr_widths,
 				      BlockNumber  *pages,
-                      double       *tuples,
-                      bool         *default_stats_used)
+                      double       *tuples)
 {
 	BlockNumber relpages;
 	double		reltuples;
 	double		density;
     BlockNumber curpages = 0;
 
-    *default_stats_used = false;
-
     /* Rel not distributed?  RelationGetNumberOfBlocks can get actual #pages. */
     if (!relOptInfo->cdbpolicy ||
         relOptInfo->cdbpolicy->ptype == POLICYTYPE_ENTRY)
@@ -1222,94 +1204,3 @@ has_unique_index(RelOptInfo *rel, AttrNumber attno)
 	}
 	return false;
 }
-
-
-/*
- * cdb_default_stats_warning_needed
- */
-static bool
-cdb_default_stats_warning_needed(Oid reloid)
-{
-    Relation    relation;
-    bool        warn = true;
-
-    /* Find relcache entry. */
-    relation = relation_open(reloid, NoLock);
-
-    /* Keep quiet if temporary or system table. */
-    if (relation->rd_rel->relpersistence == RELPERSISTENCE_TEMP ||
-        IsSystemClass(relation->rd_rel))
-        warn = false;
-
-    /* Warn at most once during lifetime of relcache entry. */
-    else if (relation->rd_cdbDefaultStatsWarningIssued)
-        warn = false;
-
-    /* Caller will issue warning.  Set flag so warning won't be repeated. */
-    else
-        relation->rd_cdbDefaultStatsWarningIssued = true;
-
-    /* Close rel.  Don't disturb the lock. */
-    relation_close(relation, NoLock);
-
-    return warn;
-}                               /* cdb_default_stats_warning_needed */
-
-
-/*
- * cdb_default_stats_warning_for_index
- */
-void
-cdb_default_stats_warning_for_index(Oid reloid, Oid indexoid)
-{
-    char           *relname;
-    char           *indexname;
-
-    /* Warn at most once during lifetime of relcache entry.  Skip if temp. */
-    if (!cdb_default_stats_warning_needed(indexoid))
-        return;
-
-    /* Get name from catalog, not from relcache, in case it has been renamed. */
-    relname = get_rel_name(reloid);
-    indexname = get_rel_name(indexoid);
-
-    ereport(NOTICE,
-            (errmsg("Query planner will use default statistics for index \"%s\" "
-                    "on table \"%s\"",
-                    indexname ? indexname : "??",
-                    relname ? relname : "??"),
-             errhint("To cache a sample of the table's actual statistics for "
-                     "optimization, use the ANALYZE or VACUUM ANALYZE command.")
-             ));
-
-    if (relname)
-        pfree(relname);
-    if (indexname)
-        pfree(indexname);
-}                               /* cdb_default_stats_warning_for_index */
-
-/*
- * cdb_default_stats_warning_for_table
- */
-void
-cdb_default_stats_warning_for_table(Oid reloid)
-{
-    char   *relname;
-
-    /* Warn at most once during lifetime of relcache entry.  Skip if temp. */
-    if (!cdb_default_stats_warning_needed(reloid))
-        return;
-
-    /* Get name from catalog, not from relcache, in case name has changed. */
-    relname = get_rel_name(reloid);
-
-    ereport(NOTICE,
-            (errmsg("Query planner will use default statistics for table \"%s\"",
-                    relname ? relname : "??"),
-             errhint("To cache a sample of the table's actual statistics for "
-                     "optimization, use the ANALYZE or VACUUM ANALYZE command.")
-             ));
-
-    if (relname)
-        pfree(relname);
-}                               /* cdb_default_stats_warning_for_table */

src/include/nodes/relation.h

@@ -550,7 +550,6 @@ typedef struct RelOptInfo
 	double		tuples;
     struct GpPolicy   *cdbpolicy;      /* distribution of stored tuples */
 	char		relstorage;		/* from pg_class.relstorage */
-    bool        cdb_default_stats_used; /* true if ANALYZE needed */
 	struct Plan *subplan;		/* if subquery */
 	List	   *subrtable;		/* if subquery */
 	List	   *subrowmark;		/* if subquery */
@@ -637,7 +636,6 @@ typedef struct IndexOptInfo
 	bool		amsearchnulls;	/* can AM search for NULL/NOT NULL entries? */
 	bool		amhasgettuple;	/* does AM have amgettuple interface? */
 	bool		amhasgetbitmap; /* does AM have amgetbitmap interface? */
-    bool        cdb_default_stats_used; /* true if ANALYZE needed */
     int         num_leading_eq; /* CDB: always 0, except amcostestimate proc may
                                  * set it briefly; it is transferred forthwith
                                  * to the IndexPath (q.v.), then reset. Kludge.