Support MERGE on distributed tables with restrictions

This implements the phase - II of MERGE sql support

Support routable query where all the tables in the merge-sql are distributed, co-located, and both the source and
target relations are joined on the distribution column with a constant qual. This should be a Citus single-task
query. Below is an example.

SELECT create_distributed_table('t1', 'id');
SELECT create_distributed_table('s1', 'id', colocate_with => ‘t1’);

MERGE INTO t1
USING s1 ON t1.id = s1.id AND t1.id = 100
WHEN MATCHED THEN
UPDATE SET val = s1.val + 10
WHEN MATCHED THEN
DELETE
WHEN NOT MATCHED THEN
INSERT (id, val, src) VALUES (s1.id, s1.val, s1.src)

Basically, MERGE checks to see if

There are a minimum of two distributed tables (source and a target).
All the distributed tables are indeed colocated.
MERGE relations are joined on the distribution column
MERGE .. USING .. ON target.dist_key = source.dist_key
The query should touch only a single shard i.e. JOIN AND with a constant qual
MERGE .. USING .. ON target.dist_key = source.dist_key AND target.dist_key = <>
If any of the conditions are not met, it raises an exception.

(cherry picked from commit 44c387b978)

This implements MERGE phase3

Support pushdown query where all the tables in the merge-sql are Citus-distributed, co-located, and both
the source and target relations are joined on the distribution column. This will generate multiple tasks
which execute independently after pushdown.

SELECT create_distributed_table('t1', 'id');
SELECT create_distributed_table('s1', 'id', colocate_with => ‘t1’);

MERGE INTO t1
USING s1
ON t1.id = s1.id
        WHEN MATCHED THEN
                UPDATE SET val = s1.val + 10
        WHEN MATCHED THEN
                DELETE
        WHEN NOT MATCHED THEN
                INSERT (id, val, src) VALUES (s1.id, s1.val, s1.src)

*The only exception for both the phases II and III is, UPDATEs and INSERTs must be done on the same shard-group
as the joined key; for example, below scenarios are NOT supported as the key-value to be inserted/updated is not
guaranteed to be on the same node as the id distribution-column.

MERGE INTO target t
USING source s ON (t.customer_id = s.customer_id)
WHEN NOT MATCHED THEN - -
     INSERT(customer_id, …) VALUES (<non-local-constant-key-value>, ……);

OR this scenario where we update the distribution column itself

MERGE INTO target t
USING source s On (t.customer_id = s.customer_id)
WHEN MATCHED THEN
     UPDATE SET customer_id = 100;

(cherry picked from commit fa7b8949a8)
pull/6772/head
Teja Mupparti 2022-12-21 19:33:13 -08:00 committed by Teja Mupparti
parent b8b85072d6
commit 1e42cd3da0
13 changed files with 2450 additions and 248 deletions

View File

@ -75,9 +75,6 @@ static uint64 NextPlanId = 1;
/* keep track of planner call stack levels */
int PlannerLevel = 0;
static void ErrorIfQueryHasUnsupportedMergeCommand(Query *queryTree,
List *rangeTableList);
static bool ContainsMergeCommandWalker(Node *node);
static bool ListContainsDistributedTableRTE(List *rangeTableList,
bool *maybeHasForeignDistributedTable);
static bool IsUpdateOrDelete(Query *query);
@ -132,7 +129,7 @@ static PlannedStmt * PlanDistributedStmt(DistributedPlanningContext *planContext
static RTEListProperties * GetRTEListProperties(List *rangeTableList);
static List * TranslatedVars(PlannerInfo *root, int relationIndex);
static void WarnIfListHasForeignDistributedTable(List *rangeTableList);
static void ErrorIfMergeHasUnsupportedTables(Query *parse, List *rangeTableList);
/* Distributed planner hook */
PlannedStmt *
@ -200,12 +197,6 @@ distributed_planner(Query *parse,
if (!fastPathRouterQuery)
{
/*
* Fast path queries cannot have merge command, and we
* prevent the remaining here.
*/
ErrorIfQueryHasUnsupportedMergeCommand(parse, rangeTableList);
/*
* When there are partitioned tables (not applicable to fast path),
* pretend that they are regular tables to avoid unnecessary work
@ -304,44 +295,11 @@ distributed_planner(Query *parse,
}
/*
* ErrorIfQueryHasUnsupportedMergeCommand walks over the query tree and bails out
* if there is no Merge command (e.g., CMD_MERGE) in the query tree. For merge,
* looks for all supported combinations, throws an exception if any violations
* are seen.
*/
static void
ErrorIfQueryHasUnsupportedMergeCommand(Query *queryTree, List *rangeTableList)
{
/*
* Postgres currently doesn't support Merge queries inside subqueries and
* ctes, but lets be defensive and do query tree walk anyway.
*
* We do not call this path for fast-path queries to avoid this additional
* overhead.
*/
if (!ContainsMergeCommandWalker((Node *) queryTree))
{
/* No MERGE found */
return;
}
/*
* In Citus we have limited support for MERGE, it's allowed
* only if all the tables(target, source or any CTE) tables
* are are local i.e. a combination of Citus local and Non-Citus
* tables (regular Postgres tables).
*/
ErrorIfMergeHasUnsupportedTables(queryTree, rangeTableList);
}
/*
* ContainsMergeCommandWalker walks over the node and finds if there are any
* Merge command (e.g., CMD_MERGE) in the node.
*/
static bool
bool
ContainsMergeCommandWalker(Node *node)
{
#if PG_VERSION_NUM < PG_VERSION_15
@ -676,7 +634,8 @@ bool
IsUpdateOrDelete(Query *query)
{
return query->commandType == CMD_UPDATE ||
query->commandType == CMD_DELETE;
query->commandType == CMD_DELETE ||
query->commandType == CMD_MERGE;
}
@ -2611,148 +2570,3 @@ WarnIfListHasForeignDistributedTable(List *rangeTableList)
}
}
}
/*
* IsMergeAllowedOnRelation takes a relation entry and checks if MERGE command is
* permitted on special relations, such as materialized view, returns true only if
* it's a "source" relation.
*/
bool
IsMergeAllowedOnRelation(Query *parse, RangeTblEntry *rte)
{
if (!IsMergeQuery(parse))
{
return false;
}
RangeTblEntry *targetRte = rt_fetch(parse->resultRelation, parse->rtable);
/* Is it a target relation? */
if (targetRte->relid == rte->relid)
{
return false;
}
return true;
}
/*
* ErrorIfMergeHasUnsupportedTables checks if all the tables(target, source or any CTE
* present) in the MERGE command are local i.e. a combination of Citus local and Non-Citus
* tables (regular Postgres tables), raises an exception for all other combinations.
*/
static void
ErrorIfMergeHasUnsupportedTables(Query *parse, List *rangeTableList)
{
ListCell *tableCell = NULL;
foreach(tableCell, rangeTableList)
{
RangeTblEntry *rangeTableEntry = (RangeTblEntry *) lfirst(tableCell);
Oid relationId = rangeTableEntry->relid;
switch (rangeTableEntry->rtekind)
{
case RTE_RELATION:
{
/* Check the relation type */
break;
}
case RTE_SUBQUERY:
case RTE_FUNCTION:
case RTE_TABLEFUNC:
case RTE_VALUES:
case RTE_JOIN:
case RTE_CTE:
{
/* Skip them as base table(s) will be checked */
continue;
}
/*
* RTE_NAMEDTUPLESTORE is typically used in ephmeral named relations,
* such as, trigger data; until we find a genuine use case, raise an
* exception.
* RTE_RESULT is a node added by the planner and we shouldn't
* encounter it in the parse tree.
*/
case RTE_NAMEDTUPLESTORE:
case RTE_RESULT:
{
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("MERGE command is not supported with "
"Tuplestores and results")));
break;
}
default:
{
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("MERGE command: Unrecognized range table entry.")));
}
}
/* RTE Relation can be of various types, check them now */
/* skip the regular views as they are replaced with subqueries */
if (rangeTableEntry->relkind == RELKIND_VIEW)
{
continue;
}
if (rangeTableEntry->relkind == RELKIND_MATVIEW ||
rangeTableEntry->relkind == RELKIND_FOREIGN_TABLE)
{
/* Materialized view or Foreign table as target is not allowed */
if (IsMergeAllowedOnRelation(parse, rangeTableEntry))
{
/* Non target relation is ok */
continue;
}
else
{
ereport(ERROR, (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("MERGE command is not allowed "
"on materialized view")));
}
}
if (rangeTableEntry->relkind != RELKIND_RELATION &&
rangeTableEntry->relkind != RELKIND_PARTITIONED_TABLE)
{
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("Unexpected relation type(relkind:%c) in MERGE command",
rangeTableEntry->relkind)));
}
Assert(rangeTableEntry->relid != 0);
/* Distributed tables and Reference tables are not supported yet */
if (IsCitusTableType(relationId, REFERENCE_TABLE) ||
IsCitusTableType(relationId, DISTRIBUTED_TABLE))
{
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("MERGE command is not supported on "
"distributed/reference tables yet")));
}
/* Regular Postgres tables and Citus local tables are allowed */
if (!IsCitusTable(relationId) ||
IsCitusTableType(relationId, CITUS_LOCAL_TABLE))
{
continue;
}
/* Any other Citus table type missing ? */
}
/* All the tables are local, supported */
}

View File

@ -54,8 +54,6 @@
bool EnableFastPathRouterPlanner = true;
static bool ColumnAppearsMultipleTimes(Node *quals, Var *distributionKey);
static bool ConjunctionContainsColumnFilter(Node *node, Var *column,
Node **distributionKeyValue);
static bool DistKeyInSimpleOpExpression(Expr *clause, Var *distColumn,
Node **distributionKeyValue);
@ -294,7 +292,7 @@ ColumnAppearsMultipleTimes(Node *quals, Var *distributionKey)
*
* If the conjuction contains column filter which is const, distributionKeyValue is set.
*/
static bool
bool
ConjunctionContainsColumnFilter(Node *node, Var *column, Node **distributionKeyValue)
{
if (node == NULL)

View File

@ -2225,14 +2225,17 @@ QueryPushdownSqlTaskList(Query *query, uint64 jobId,
}
/*
* For left joins we don't care about the shards pruned for the right hand side.
* If the right hand side would prune to a smaller set we should still send it to
* all tables of the left hand side. However if the right hand side is bigger than
* the left hand side we don't have to send the query to any shard that is not
* matching anything on the left hand side.
* For left joins we don't care about the shards pruned for
* the right hand side. If the right hand side would prune
* to a smaller set we should still send it to all tables
* of the left hand side. However if the right hand side is
* bigger than the left hand side we don't have to send the
* query to any shard that is not matching anything on the
* left hand side.
*
* Instead we will simply skip any RelationRestriction if it is an OUTER join and
* the table is part of the non-outer side of the join.
* Instead we will simply skip any RelationRestriction if it
* is an OUTER join and the table is part of the non-outer
* side of the join.
*/
if (IsInnerTableOfOuterJoin(relationRestriction))
{

View File

@ -185,7 +185,6 @@ static DeferredErrorMessage * TargetlistAndFunctionsSupported(Oid resultRelation
List *targetList,
CmdType commandType,
List *returningList);
/*
* CreateRouterPlan attempts to create a router executor plan for the given
* SELECT statement. ->planningError is set if planning fails.
@ -905,6 +904,85 @@ NodeIsFieldStore(Node *node)
}
/*
* MergeQuerySupported does check for a MERGE command in the query, if it finds
* one, it will verify the below criteria
* - Supported tables and combinations in ErrorIfMergeHasUnsupportedTables
* - Distributed tables requirements in ErrorIfDistTablesNotColocated
* - Checks target-lists and functions-in-quals in TargetlistAndFunctionsSupported
*/
static DeferredErrorMessage *
MergeQuerySupported(Query *originalQuery,
PlannerRestrictionContext *plannerRestrictionContext)
{
/* For non-MERGE commands it's a no-op */
if (!QueryHasMergeCommand(originalQuery))
{
return NULL;
}
List *rangeTableList = ExtractRangeTableEntryList(originalQuery);
RangeTblEntry *resultRte = ExtractResultRelationRTE(originalQuery);
/*
* Fast path queries cannot have merge command, and we prevent the remaining here.
* In Citus we have limited support for MERGE, it's allowed only if all
* the tables(target, source or any CTE) tables are are local i.e. a
* combination of Citus local and Non-Citus tables (regular Postgres tables)
* or distributed tables with some restrictions, please see header of routine
* ErrorIfDistTablesNotColocated for details.
*/
DeferredErrorMessage *deferredError =
ErrorIfMergeHasUnsupportedTables(originalQuery,
rangeTableList,
plannerRestrictionContext);
if (deferredError)
{
return deferredError;
}
Oid resultRelationId = resultRte->relid;
deferredError =
TargetlistAndFunctionsSupported(resultRelationId,
originalQuery->jointree,
originalQuery->jointree->quals,
originalQuery->targetList,
originalQuery->commandType,
originalQuery->returningList);
if (deferredError)
{
return deferredError;
}
#if PG_VERSION_NUM >= PG_VERSION_15
/*
* MERGE is a special case where we have multiple modify statements
* within itself. Check each INSERT/UPDATE/DELETE individually.
*/
MergeAction *action = NULL;
foreach_ptr(action, originalQuery->mergeActionList)
{
Assert(originalQuery->returningList == NULL);
deferredError =
TargetlistAndFunctionsSupported(resultRelationId,
originalQuery->jointree,
action->qual,
action->targetList,
action->commandType,
originalQuery->returningList);
if (deferredError)
{
return deferredError;
}
}
#endif
return NULL;
}
/*
* ModifyQuerySupported returns NULL if the query only contains supported
* features, otherwise it returns an error description.
@ -920,8 +998,17 @@ ModifyQuerySupported(Query *queryTree, Query *originalQuery, bool multiShardQuer
PlannerRestrictionContext *plannerRestrictionContext)
{
Oid distributedTableId = InvalidOid;
DeferredErrorMessage *error = ModifyPartialQuerySupported(queryTree, multiShardQuery,
&distributedTableId);
DeferredErrorMessage *error = MergeQuerySupported(originalQuery,
plannerRestrictionContext);
if (error)
{
/*
* For MERGE, we do not do recursive plannning, simply bail out.
*/
RaiseDeferredError(error, ERROR);
}
error = ModifyPartialQuerySupported(queryTree, multiShardQuery, &distributedTableId);
if (error)
{
return error;
@ -3969,3 +4056,263 @@ CompareInsertValuesByShardId(const void *leftElement, const void *rightElement)
}
}
}
/*
* IsMergeAllowedOnRelation takes a relation entry and checks if MERGE command is
* permitted on special relations, such as materialized view, returns true only if
* it's a "source" relation.
*/
bool
IsMergeAllowedOnRelation(Query *parse, RangeTblEntry *rte)
{
if (!IsMergeQuery(parse))
{
return false;
}
RangeTblEntry *targetRte = rt_fetch(parse->resultRelation, parse->rtable);
/* Is it a target relation? */
if (targetRte->relid == rte->relid)
{
return false;
}
return true;
}
/*
* ErrorIfDistTablesNotColocated Checks to see if
*
* - There are a minimum of two distributed tables (source and a target).
* - All the distributed tables are indeed colocated.
* - MERGE relations are joined on the distribution column
* MERGE .. USING .. ON target.dist_key = source.dist_key
*
* If any of the conditions are not met, it raises an exception.
*/
static DeferredErrorMessage *
ErrorIfDistTablesNotColocated(Query *parse, List *distTablesList,
PlannerRestrictionContext *plannerRestrictionContext)
{
/* All MERGE tables must be distributed */
if (list_length(distTablesList) < 2)
{
return DeferredError(ERRCODE_FEATURE_NOT_SUPPORTED,
"For MERGE command, both the source and target "
"must be distributed", NULL, NULL);
}
/* All distributed tables must be colocated */
if (!AllRelationsInListColocated(distTablesList, RANGETABLE_ENTRY))
{
return DeferredError(ERRCODE_FEATURE_NOT_SUPPORTED,
"For MERGE command, all the distributed tables "
"must be colocated", NULL, NULL);
}
/* Are source and target tables joined on distribution column? */
if (!RestrictionEquivalenceForPartitionKeys(plannerRestrictionContext))
{
return DeferredError(ERRCODE_FEATURE_NOT_SUPPORTED,
"MERGE command is only supported when distributed "
"tables are joined on their distribution column",
NULL, NULL);
}
return NULL;
}
/*
* ErrorIfMergeHasUnsupportedTables checks if all the tables(target, source or any CTE
* present) in the MERGE command are local i.e. a combination of Citus local and Non-Citus
* tables (regular Postgres tables), or distributed tables with some restrictions, please
* see header of routine ErrorIfDistTablesNotColocated for details, raises an exception
* for all other combinations.
*/
static DeferredErrorMessage *
ErrorIfMergeHasUnsupportedTables(Query *parse, List *rangeTableList,
PlannerRestrictionContext *restrictionContext)
{
List *distTablesList = NIL;
bool foundLocalTables = false;
RangeTblEntry *rangeTableEntry = NULL;
foreach_ptr(rangeTableEntry, rangeTableList)
{
Oid relationId = rangeTableEntry->relid;
switch (rangeTableEntry->rtekind)
{
case RTE_RELATION:
{
/* Check the relation type */
break;
}
case RTE_SUBQUERY:
case RTE_FUNCTION:
case RTE_TABLEFUNC:
case RTE_VALUES:
case RTE_JOIN:
case RTE_CTE:
{
/* Skip them as base table(s) will be checked */
continue;
}
/*
* RTE_NAMEDTUPLESTORE is typically used in ephmeral named relations,
* such as, trigger data; until we find a genuine use case, raise an
* exception.
* RTE_RESULT is a node added by the planner and we shouldn't
* encounter it in the parse tree.
*/
case RTE_NAMEDTUPLESTORE:
case RTE_RESULT:
{
return DeferredError(ERRCODE_FEATURE_NOT_SUPPORTED,
"MERGE command is not supported with "
"Tuplestores and results",
NULL, NULL);
}
default:
{
return DeferredError(ERRCODE_FEATURE_NOT_SUPPORTED,
"MERGE command: Unrecognized range table entry.",
NULL, NULL);
}
}
/* RTE Relation can be of various types, check them now */
/* skip the regular views as they are replaced with subqueries */
if (rangeTableEntry->relkind == RELKIND_VIEW)
{
continue;
}
if (rangeTableEntry->relkind == RELKIND_MATVIEW ||
rangeTableEntry->relkind == RELKIND_FOREIGN_TABLE)
{
/* Materialized view or Foreign table as target is not allowed */
if (IsMergeAllowedOnRelation(parse, rangeTableEntry))
{
/* Non target relation is ok */
continue;
}
else
{
/* Usually we don't reach this exception as the Postgres parser catches it */
StringInfo errorMessage = makeStringInfo();
appendStringInfo(errorMessage,
"MERGE command is not allowed on "
"relation type(relkind:%c)", rangeTableEntry->relkind);
return DeferredError(ERRCODE_FEATURE_NOT_SUPPORTED, errorMessage->data,
NULL, NULL);
}
}
if (rangeTableEntry->relkind != RELKIND_RELATION &&
rangeTableEntry->relkind != RELKIND_PARTITIONED_TABLE)
{
StringInfo errorMessage = makeStringInfo();
appendStringInfo(errorMessage, "Unexpected table type(relkind:%c) "
"in MERGE command", rangeTableEntry->relkind);
return DeferredError(ERRCODE_FEATURE_NOT_SUPPORTED, errorMessage->data,
NULL, NULL);
}
Assert(rangeTableEntry->relid != 0);
/* Reference tables are not supported yet */
if (IsCitusTableType(relationId, REFERENCE_TABLE))
{
return DeferredError(ERRCODE_FEATURE_NOT_SUPPORTED,
"MERGE command is not supported on reference "
"tables yet", NULL, NULL);
}
/* Append/Range tables are not supported */
if (IsCitusTableType(relationId, APPEND_DISTRIBUTED) ||
IsCitusTableType(relationId, RANGE_DISTRIBUTED))
{
return DeferredError(ERRCODE_FEATURE_NOT_SUPPORTED,
"For MERGE command, all the distributed tables "
"must be colocated, for append/range distribution, "
"colocation is not supported", NULL,
"Consider using hash distribution instead");
}
/*
* For now, save all distributed tables, later (below) we will
* check for supported combination(s).
*/
if (IsCitusTableType(relationId, DISTRIBUTED_TABLE))
{
distTablesList = lappend(distTablesList, rangeTableEntry);
continue;
}
/* Regular Postgres tables and Citus local tables are allowed */
if (!IsCitusTable(relationId) ||
IsCitusTableType(relationId, CITUS_LOCAL_TABLE))
{
foundLocalTables = true;
continue;
}
/* Any other Citus table type missing ? */
}
/* Ensure all tables are indeed local */
if (foundLocalTables && list_length(distTablesList) == 0)
{
/* All the tables are local, supported */
return NULL;
}
else if (foundLocalTables && list_length(distTablesList) > 0)
{
return DeferredError(ERRCODE_FEATURE_NOT_SUPPORTED,
"MERGE command is not supported with "
"combination of distributed/local tables yet",
NULL, NULL);
}
/* Ensure all distributed tables are indeed co-located */
return ErrorIfDistTablesNotColocated(parse, distTablesList, restrictionContext);
}
/*
* QueryHasMergeCommand walks over the query tree and returns false if there
* is no Merge command (e.g., CMD_MERGE), true otherwise.
*/
static bool
QueryHasMergeCommand(Query *queryTree)
{
/* function is void for pre-15 versions of Postgres */
#if PG_VERSION_NUM < PG_VERSION_15
return false;
#else
/*
* Postgres currently doesn't support Merge queries inside subqueries and
* ctes, but lets be defensive and do query tree walk anyway.
*
* We do not call this path for fast-path queries to avoid this additional
* overhead.
*/
if (!ContainsMergeCommandWalker((Node *) queryTree))
{
/* No MERGE found */
return false;
}
return true;
#endif
}

View File

@ -151,8 +151,6 @@ static void ListConcatUniqueAttributeClassMemberLists(AttributeEquivalenceClass
secondClass);
static Var * PartitionKeyForRTEIdentityInQuery(Query *query, int targetRTEIndex,
Index *partitionKeyIndex);
static bool AllRelationsInRestrictionContextColocated(RelationRestrictionContext *
restrictionContext);
static bool IsNotSafeRestrictionToRecursivelyPlan(Node *node);
static JoinRestrictionContext * FilterJoinRestrictionContext(
JoinRestrictionContext *joinRestrictionContext, Relids
@ -383,7 +381,8 @@ SafeToPushdownUnionSubquery(Query *originalQuery,
return false;
}
if (!AllRelationsInRestrictionContextColocated(restrictionContext))
if (!AllRelationsInListColocated(restrictionContext->relationRestrictionList,
RESTRICTION_CONTEXT))
{
/* distribution columns are equal, but tables are not co-located */
return false;
@ -1919,19 +1918,33 @@ FindQueryContainingRTEIdentityInternal(Node *node,
/*
* AllRelationsInRestrictionContextColocated determines whether all of the relations in the
* given relation restrictions list are co-located.
* AllRelationsInListColocated determines whether all of the relations in the
* given list are co-located.
* Note: The list can be of dofferent types, which is specified by ListEntryType
*/
static bool
AllRelationsInRestrictionContextColocated(RelationRestrictionContext *restrictionContext)
bool
AllRelationsInListColocated(List *relationList, ListEntryType entryType)
{
void *varPtr = NULL;
RangeTblEntry *rangeTableEntry = NULL;
RelationRestriction *relationRestriction = NULL;
int initialColocationId = INVALID_COLOCATION_ID;
/* check whether all relations exists in the main restriction list */
foreach_ptr(relationRestriction, restrictionContext->relationRestrictionList)
foreach_ptr(varPtr, relationList)
{
Oid relationId = relationRestriction->relationId;
Oid relationId = InvalidOid;
if (entryType == RANGETABLE_ENTRY)
{
rangeTableEntry = (RangeTblEntry *) varPtr;
relationId = rangeTableEntry->relid;
}
else if (entryType == RESTRICTION_CONTEXT)
{
relationRestriction = (RelationRestriction *) varPtr;
relationId = relationRestriction->relationId;
}
if (IsCitusTable(relationId) && !HasDistributionKey(relationId))
{

View File

@ -256,5 +256,9 @@ extern struct DistributedPlan * CreateDistributedPlan(uint64 planId,
plannerRestrictionContext);
extern bool IsMergeAllowedOnRelation(Query *parse, RangeTblEntry *rte);
extern bool ConjunctionContainsColumnFilter(Node *node,
Var *column,
Node **distributionKeyValue);
extern bool ContainsMergeCommandWalker(Node *node);
#endif /* DISTRIBUTED_PLANNER_H */

View File

@ -17,6 +17,15 @@
#define SINGLE_RTE_INDEX 1
/*
* Represents the pointer type that's being passed in the list.
*/
typedef enum ListEntryType
{
RANGETABLE_ENTRY, /* RangeTblEntry */
RESTRICTION_CONTEXT /* RelationRestriction */
} ListEntryType;
extern bool AllDistributionKeysInQueryAreEqual(Query *originalQuery,
PlannerRestrictionContext *
plannerRestrictionContext);
@ -54,4 +63,6 @@ extern RelationRestrictionContext * FilterRelationRestrictionContext(
RelationRestrictionContext *relationRestrictionContext,
Relids
queryRteIdentities);
extern bool AllRelationsInListColocated(List *relationList, ListEntryType entryType);
#endif /* RELATION_RESTRICTION_EQUIVALENCE_H */

View File

@ -28,6 +28,10 @@ s/\(ref_id\)=\([0-9]+\)/(ref_id)=(X)/g
# shard table names for multi_subtransactions
s/"t2_[0-9]+"/"t2_xxxxxxx"/g
# shard table names for MERGE tests
s/merge_schema\.([_a-z0-9]+)_40[0-9]+ /merge_schema.\1_xxxxxxx /g
s/pgmerge_schema\.([_a-z0-9]+)_40[0-9]+ /pgmerge_schema.\1_xxxxxxx /g
# shard table names for multi_subquery
s/ keyval(1|2|ref)_[0-9]+ / keyval\1_xxxxxxx /g

File diff suppressed because it is too large Load Diff

View File

@ -315,7 +315,7 @@ SELECT create_reference_table('tbl2');
MERGE INTO tbl1 USING tbl2 ON (true)
WHEN MATCHED THEN DELETE;
ERROR: MERGE command is not supported on distributed/reference tables yet
ERROR: MERGE command is not supported on reference tables yet
-- now, both are reference, still not supported
SELECT create_reference_table('tbl1');
create_reference_table
@ -325,7 +325,7 @@ SELECT create_reference_table('tbl1');
MERGE INTO tbl1 USING tbl2 ON (true)
WHEN MATCHED THEN DELETE;
ERROR: MERGE command is not supported on distributed/reference tables yet
ERROR: MERGE command is not supported on reference tables yet
-- now, both distributed, not works
SELECT undistribute_table('tbl1');
NOTICE: creating a new table for pg15.tbl1
@ -419,14 +419,14 @@ SELECT create_distributed_table('tbl2', 'x');
MERGE INTO tbl1 USING tbl2 ON (true)
WHEN MATCHED THEN DELETE;
ERROR: MERGE command is not supported on distributed/reference tables yet
ERROR: MERGE command is only supported when distributed tables are joined on their distribution column
-- also, not inside subqueries & ctes
WITH targq AS (
SELECT * FROM tbl2
)
MERGE INTO tbl1 USING targq ON (true)
WHEN MATCHED THEN DELETE;
ERROR: MERGE command is not supported on distributed/reference tables yet
ERROR: MERGE command is only supported when distributed tables are joined on their distribution column
-- crashes on beta3, fixed on 15 stable
--WITH foo AS (
-- MERGE INTO tbl1 USING tbl2 ON (true)
@ -441,7 +441,7 @@ USING tbl2
ON (true)
WHEN MATCHED THEN
UPDATE SET x = (SELECT count(*) FROM tbl2);
ERROR: MERGE command is not supported on distributed/reference tables yet
ERROR: MERGE command is only supported when distributed tables are joined on their distribution column
-- test numeric types with negative scale
CREATE TABLE numeric_negative_scale(numeric_column numeric(3,-1), orig_value int);
INSERT into numeric_negative_scale SELECT x,x FROM generate_series(111, 115) x;

View File

@ -910,7 +910,15 @@ MERGE INTO wq_target t
USING wq_source s ON t.tid = s.sid
WHEN MATCHED AND (merge_when_and_write()) THEN
UPDATE SET balance = t.balance + s.balance;
ERROR: functions used in UPDATE queries on distributed tables must not be VOLATILE
ERROR: functions used in the WHERE/ON/WHEN clause of modification queries on distributed tables must not be VOLATILE
ROLLBACK;
-- Test preventing ON condition from writing to the database
BEGIN;
MERGE INTO wq_target t
USING wq_source s ON t.tid = s.sid AND (merge_when_and_write())
WHEN MATCHED THEN
UPDATE SET balance = t.balance + s.balance;
ERROR: functions used in the WHERE/ON/WHEN clause of modification queries on distributed tables must not be VOLATILE
ROLLBACK;
drop function merge_when_and_write();
DROP TABLE wq_target, wq_source;
@ -1891,7 +1899,7 @@ MERGE INTO pa_target t
UPDATE SET balance = balance + delta, val = val || ' updated by merge'
WHEN NOT MATCHED THEN
INSERT VALUES (slogts::timestamp, sid, delta, 'inserted by merge');
DEBUG: <Deparsed MERGE query: MERGE INTO pgmerge_schema.pa_target t USING (SELECT '2017-01-15'::text AS slogts, pa_source.sid, pa_source.delta FROM pgmerge_schema.pa_source_4001021 pa_source WHERE (pa_source.sid OPERATOR(pg_catalog.<) 10)) s ON (t.tid OPERATOR(pg_catalog.=) s.sid) WHEN MATCHED THEN UPDATE SET balance = (t.balance OPERATOR(pg_catalog.+) s.delta), val = (t.val OPERATOR(pg_catalog.||) ' updated by merge'::text) WHEN NOT MATCHED THEN INSERT (logts, tid, balance, val) VALUES ((s.slogts)::timestamp without time zone, s.sid, s.delta, 'inserted by merge'::text)>
DEBUG: <Deparsed MERGE query: MERGE INTO pgmerge_schema.pa_target t USING (SELECT '2017-01-15'::text AS slogts, pa_source.sid, pa_source.delta FROM pgmerge_schema.pa_source_xxxxxxx pa_source WHERE (pa_source.sid OPERATOR(pg_catalog.<) 10)) s ON (t.tid OPERATOR(pg_catalog.=) s.sid) WHEN MATCHED THEN UPDATE SET balance = (t.balance OPERATOR(pg_catalog.+) s.delta), val = (t.val OPERATOR(pg_catalog.||) ' updated by merge'::text) WHEN NOT MATCHED THEN INSERT (logts, tid, balance, val) VALUES ((s.slogts)::timestamp without time zone, s.sid, s.delta, 'inserted by merge'::text)>
SELECT * FROM pa_target ORDER BY tid;
logts | tid | balance | val
---------------------------------------------------------------------
@ -2083,7 +2091,7 @@ WHEN MATCHED THEN UPDATE
WHEN NOT MATCHED THEN INSERT
(city_id, logdate, peaktemp, unitsales)
VALUES (city_id, logdate, peaktemp, unitsales);
DEBUG: <Deparsed MERGE query: MERGE INTO pgmerge_schema.measurement m USING pgmerge_schema.new_measurement_4001026 nm ON ((m.city_id OPERATOR(pg_catalog.=) nm.city_id) AND (m.logdate OPERATOR(pg_catalog.=) nm.logdate)) WHEN MATCHED AND (nm.peaktemp IS NULL) THEN DELETE WHEN MATCHED THEN UPDATE SET peaktemp = GREATEST(m.peaktemp, nm.peaktemp), unitsales = (m.unitsales OPERATOR(pg_catalog.+) COALESCE(nm.unitsales, 0)) WHEN NOT MATCHED THEN INSERT (city_id, logdate, peaktemp, unitsales) VALUES (nm.city_id, nm.logdate, nm.peaktemp, nm.unitsales)>
DEBUG: <Deparsed MERGE query: MERGE INTO pgmerge_schema.measurement m USING pgmerge_schema.new_measurement_xxxxxxx nm ON ((m.city_id OPERATOR(pg_catalog.=) nm.city_id) AND (m.logdate OPERATOR(pg_catalog.=) nm.logdate)) WHEN MATCHED AND (nm.peaktemp IS NULL) THEN DELETE WHEN MATCHED THEN UPDATE SET peaktemp = GREATEST(m.peaktemp, nm.peaktemp), unitsales = (m.unitsales OPERATOR(pg_catalog.+) COALESCE(nm.unitsales, 0)) WHEN NOT MATCHED THEN INSERT (city_id, logdate, peaktemp, unitsales) VALUES (nm.city_id, nm.logdate, nm.peaktemp, nm.unitsales)>
RESET client_min_messages;
SELECT tableoid::regclass, * FROM measurement ORDER BY city_id, logdate;
tableoid | city_id | logdate | peaktemp | unitsales

View File

@ -19,6 +19,7 @@ SET search_path TO merge_schema;
SET citus.shard_count TO 4;
SET citus.next_shard_id TO 4000000;
SET citus.explain_all_tasks to true;
SET citus.shard_replication_factor TO 1;
SELECT 1 FROM master_add_node('localhost', :master_port, groupid => 0);
CREATE TABLE source
@ -143,9 +144,33 @@ SELECT undistribute_table('source');
SELECT create_distributed_table('target', 'customer_id');
SELECT create_distributed_table('source', 'customer_id');
-- Updates one of the row with customer_id = 30002
SELECT * from target t WHERE t.customer_id = 30002;
-- Turn on notice to print tasks sent to nodes
SET citus.log_remote_commands to true;
MERGE INTO target t
USING source s
ON (t.customer_id = s.customer_id)
ON (t.customer_id = s.customer_id) AND t.customer_id = 30002
WHEN MATCHED AND t.order_center = 'XX' THEN
DELETE
WHEN MATCHED THEN
UPDATE SET -- Existing customer, update the order count and last_order_id
order_count = t.order_count + 1,
last_order_id = s.order_id
WHEN NOT MATCHED THEN
DO NOTHING;
SET citus.log_remote_commands to false;
SELECT * from target t WHERE t.customer_id = 30002;
-- Deletes one of the row with customer_id = 30004
SELECT * from target t WHERE t.customer_id = 30004;
MERGE INTO target t
USING source s
ON (t.customer_id = s.customer_id) AND t.customer_id = 30004
WHEN MATCHED AND t.order_center = 'XX' THEN
DELETE
@ -158,6 +183,7 @@ MERGE INTO target t
WHEN NOT MATCHED THEN -- New entry, record it.
INSERT (customer_id, last_order_id, order_center, order_count, last_order)
VALUES (customer_id, s.order_id, s.order_center, 123, s.order_time);
SELECT * from target t WHERE t.customer_id = 30004;
--
-- Test MERGE with CTE as source
@ -243,11 +269,13 @@ SELECT create_distributed_table('t1', 'id');
SELECT create_distributed_table('s1', 'id');
SELECT * FROM t1 order by id;
SET citus.log_remote_commands to true;
WITH s1_res AS (
SELECT * FROM s1
)
MERGE INTO t1
USING s1_res ON (s1_res.id = t1.id)
USING s1_res ON (s1_res.id = t1.id) AND t1.id = 6
WHEN MATCHED AND s1_res.val = 0 THEN
DELETE
@ -255,6 +283,9 @@ MERGE INTO t1
UPDATE SET val = t1.val + 1
WHEN NOT MATCHED THEN
INSERT (id, val) VALUES (s1_res.id, s1_res.val);
SET citus.log_remote_commands to false;
-- Other than id 6 everything else is a NO match, and should appear in target
SELECT * FROM t1 order by 1, 2;
--
-- Test with multiple join conditions
@ -325,15 +356,21 @@ SELECT undistribute_table('s2');
SELECT create_distributed_table('t2', 'id');
SELECT create_distributed_table('s2', 'id');
SELECT * FROM t2 ORDER BY 1;
SET citus.log_remote_commands to true;
MERGE INTO t2
USING s2
ON t2.id = s2.id AND t2.src = s2.src
ON t2.id = s2.id AND t2.src = s2.src AND t2.id = 4
WHEN MATCHED AND t2.val = 1 THEN
UPDATE SET val = s2.val + 10
WHEN MATCHED THEN
DELETE
WHEN NOT MATCHED THEN
INSERT (id, val, src) VALUES (s2.id, s2.val, s2.src);
DO NOTHING;
SET citus.log_remote_commands to false;
-- Row with id = 4 is a match for delete clause, row should be deleted
-- Row with id = 3 is a NO match, row from source will be inserted
SELECT * FROM t2 ORDER BY 1;
--
-- With sub-query as the MERGE source
@ -824,10 +861,577 @@ RESET client_min_messages;
SELECT * FROM ft_target;
--
-- complex joins on the source side
--
-- source(join of two relations) relation is an unaliased join
CREATE TABLE target_cj(tid int, src text, val int);
CREATE TABLE source_cj1(sid1 int, src1 text, val1 int);
CREATE TABLE source_cj2(sid2 int, src2 text, val2 int);
INSERT INTO target_cj VALUES (1, 'target', 0);
INSERT INTO target_cj VALUES (2, 'target', 0);
INSERT INTO target_cj VALUES (2, 'target', 0);
INSERT INTO target_cj VALUES (3, 'target', 0);
INSERT INTO source_cj1 VALUES (2, 'source-1', 10);
INSERT INTO source_cj2 VALUES (2, 'source-2', 20);
BEGIN;
MERGE INTO target_cj t
USING source_cj1 s1 INNER JOIN source_cj2 s2 ON sid1 = sid2
ON t.tid = sid1 AND t.tid = 2
WHEN MATCHED THEN
UPDATE SET src = src2
WHEN NOT MATCHED THEN
DO NOTHING;
-- Gold result to compare against
SELECT * FROM target_cj ORDER BY 1;
ROLLBACK;
BEGIN;
-- try accessing columns from either side of the source join
MERGE INTO target_cj t
USING source_cj1 s2
INNER JOIN source_cj2 s1 ON sid1 = sid2 AND val1 = 10
ON t.tid = sid1 AND t.tid = 2
WHEN MATCHED THEN
UPDATE SET tid = sid2, src = src1, val = val2
WHEN NOT MATCHED THEN
DO NOTHING;
-- Gold result to compare against
SELECT * FROM target_cj ORDER BY 1;
ROLLBACK;
-- Test the same scenarios with distributed tables
SELECT create_distributed_table('target_cj', 'tid');
SELECT create_distributed_table('source_cj1', 'sid1');
SELECT create_distributed_table('source_cj2', 'sid2');
BEGIN;
SET citus.log_remote_commands to true;
MERGE INTO target_cj t
USING source_cj1 s1 INNER JOIN source_cj2 s2 ON sid1 = sid2
ON t.tid = sid1 AND t.tid = 2
WHEN MATCHED THEN
UPDATE SET src = src2
WHEN NOT MATCHED THEN
DO NOTHING;
SET citus.log_remote_commands to false;
SELECT * FROM target_cj ORDER BY 1;
ROLLBACK;
BEGIN;
-- try accessing columns from either side of the source join
MERGE INTO target_cj t
USING source_cj1 s2
INNER JOIN source_cj2 s1 ON sid1 = sid2 AND val1 = 10
ON t.tid = sid1 AND t.tid = 2
WHEN MATCHED THEN
UPDATE SET src = src1, val = val2
WHEN NOT MATCHED THEN
DO NOTHING;
SELECT * FROM target_cj ORDER BY 1;
ROLLBACK;
-- sub-query as a source
BEGIN;
MERGE INTO target_cj t
USING (SELECT * FROM source_cj1 WHERE sid1 = 2) sub
ON t.tid = sub.sid1 AND t.tid = 2
WHEN MATCHED THEN
UPDATE SET src = sub.src1, val = val1
WHEN NOT MATCHED THEN
DO NOTHING;
SELECT * FROM target_cj ORDER BY 1;
ROLLBACK;
-- Test self-join
BEGIN;
SELECT * FROM target_cj ORDER BY 1;
set citus.log_remote_commands to true;
MERGE INTO target_cj t1
USING (SELECT * FROM target_cj) sub
ON t1.tid = sub.tid AND t1.tid = 3
WHEN MATCHED THEN
UPDATE SET src = sub.src, val = sub.val + 100
WHEN NOT MATCHED THEN
DO NOTHING;
set citus.log_remote_commands to false;
SELECT * FROM target_cj ORDER BY 1;
ROLLBACK;
-- Test PREPARE
PREPARE foo(int) AS
MERGE INTO target_cj target
USING (SELECT * FROM source_cj1) sub
ON target.tid = sub.sid1 AND target.tid = $1
WHEN MATCHED THEN
UPDATE SET val = sub.val1
WHEN NOT MATCHED THEN
DO NOTHING;
SELECT * FROM target_cj ORDER BY 1;
BEGIN;
EXECUTE foo(2);
EXECUTE foo(2);
EXECUTE foo(2);
EXECUTE foo(2);
EXECUTE foo(2);
SELECT * FROM target_cj ORDER BY 1;
ROLLBACK;
BEGIN;
SET citus.log_remote_commands to true;
SET client_min_messages TO DEBUG1;
EXECUTE foo(2);
RESET client_min_messages;
EXECUTE foo(2);
SET citus.log_remote_commands to false;
SELECT * FROM target_cj ORDER BY 1;
ROLLBACK;
-- Test distributed tables, must be co-located and joined on distribution column.
--
-- We create two sets of source and target tables, one set is Postgres and the other
-- is Citus distributed. Run the _exact_ MERGE SQL on both the sets and compare the
-- final results of target tables of Postgres and Citus, the result should match.
-- This is repeated for various MERGE SQL combinations
--
CREATE TABLE pg_target(id int, val varchar);
CREATE TABLE pg_source(id int, val varchar);
CREATE TABLE citus_target(id int, val varchar);
CREATE TABLE citus_source(id int, val varchar);
-- Half of the source rows do not match
INSERT INTO pg_target SELECT i, 'target' FROM generate_series(250, 500) i;
INSERT INTO pg_source SELECT i, 'source' FROM generate_series(1, 500) i;
INSERT INTO citus_target SELECT i, 'target' FROM generate_series(250, 500) i;
INSERT INTO citus_source SELECT i, 'source' FROM generate_series(1, 500) i;
SELECT create_distributed_table('citus_target', 'id');
SELECT create_distributed_table('citus_source', 'id');
--
-- This routine compares the target tables of Postgres and Citus and
-- returns true if they match, false if the results do not match.
--
CREATE OR REPLACE FUNCTION compare_tables() RETURNS BOOLEAN AS $$
DECLARE ret BOOL;
BEGIN
SELECT count(1) = 0 INTO ret
FROM pg_target
FULL OUTER JOIN citus_target
USING (id, val)
WHERE pg_target.id IS NULL
OR citus_target.id IS NULL;
RETURN ret;
END
$$ LANGUAGE PLPGSQL;
-- Make sure we start with exact data in Postgres and Citus
SELECT compare_tables();
-- Run the MERGE on both Postgres and Citus, and compare the final target tables
BEGIN;
SET citus.log_remote_commands to true;
MERGE INTO pg_target t
USING pg_source s
ON t.id = s.id
WHEN MATCHED AND t.id > 400 THEN
UPDATE SET val = t.val || 'Updated by Merge'
WHEN MATCHED THEN
DELETE
WHEN NOT MATCHED THEN
INSERT VALUES(s.id, s.val);
MERGE INTO citus_target t
USING citus_source s
ON t.id = s.id
WHEN MATCHED AND t.id > 400 THEN
UPDATE SET val = t.val || 'Updated by Merge'
WHEN MATCHED THEN
DELETE
WHEN NOT MATCHED THEN
INSERT VALUES(s.id, s.val);
SET citus.log_remote_commands to false;
SELECT compare_tables();
ROLLBACK;
--
-- ON clause filter on source
--
BEGIN;
SET citus.log_remote_commands to true;
MERGE INTO pg_target t
USING pg_source s
ON t.id = s.id AND s.id < 100
WHEN MATCHED AND t.id > 400 THEN
UPDATE SET val = t.val || 'Updated by Merge'
WHEN MATCHED THEN
DELETE
WHEN NOT MATCHED THEN
INSERT VALUES(s.id, s.val);
MERGE INTO citus_target t
USING citus_source s
ON t.id = s.id AND s.id < 100
WHEN MATCHED AND t.id > 400 THEN
UPDATE SET val = t.val || 'Updated by Merge'
WHEN MATCHED THEN
DELETE
WHEN NOT MATCHED THEN
INSERT VALUES(s.id, s.val);
SET citus.log_remote_commands to false;
SELECT compare_tables();
ROLLBACK;
--
-- ON clause filter on target
--
BEGIN;
SET citus.log_remote_commands to true;
MERGE INTO pg_target t
USING pg_source s
ON t.id = s.id AND t.id < 100
WHEN MATCHED AND t.id > 400 THEN
UPDATE SET val = t.val || 'Updated by Merge'
WHEN MATCHED THEN
DELETE
WHEN NOT MATCHED THEN
INSERT VALUES(s.id, s.val);
MERGE INTO citus_target t
USING citus_source s
ON t.id = s.id AND t.id < 100
WHEN MATCHED AND t.id > 400 THEN
UPDATE SET val = t.val || 'Updated by Merge'
WHEN MATCHED THEN
DELETE
WHEN NOT MATCHED THEN
INSERT VALUES(s.id, s.val);
SET citus.log_remote_commands to false;
SELECT compare_tables();
ROLLBACK;
--
-- NOT MATCHED clause filter on source
--
BEGIN;
SET citus.log_remote_commands to true;
MERGE INTO pg_target t
USING pg_source s
ON t.id = s.id
WHEN MATCHED THEN
DO NOTHING
WHEN NOT MATCHED AND s.id < 100 THEN
INSERT VALUES(s.id, s.val);
MERGE INTO citus_target t
USING citus_source s
ON t.id = s.id
WHEN MATCHED THEN
DO NOTHING
WHEN NOT MATCHED AND s.id < 100 THEN
INSERT VALUES(s.id, s.val);
SET citus.log_remote_commands to false;
SELECT compare_tables();
ROLLBACK;
--
-- Test constant filter in ON clause to check if shards are pruned
-- with restriction information
--
--
-- Though constant filter is present, this won't prune shards as
-- NOT MATCHED clause is present
--
BEGIN;
SET citus.log_remote_commands to true;
MERGE INTO pg_target t
USING pg_source s
ON t.id = s.id AND s.id = 250
WHEN MATCHED THEN
UPDATE SET val = t.val || 'Updated by Merge'
WHEN NOT MATCHED THEN
INSERT VALUES(s.id, s.val);
MERGE INTO citus_target t
USING citus_source s
ON t.id = s.id AND s.id = 250
WHEN MATCHED THEN
UPDATE SET val = t.val || 'Updated by Merge'
WHEN NOT MATCHED THEN
INSERT VALUES(s.id, s.val);
SET citus.log_remote_commands to false;
SELECT compare_tables();
ROLLBACK;
-- This will prune shards with restriction information as NOT MATCHED is void
BEGIN;
SET citus.log_remote_commands to true;
MERGE INTO pg_target t
USING pg_source s
ON t.id = s.id AND s.id = 250
WHEN MATCHED THEN
UPDATE SET val = t.val || 'Updated by Merge'
WHEN NOT MATCHED THEN
DO NOTHING;
MERGE INTO citus_target t
USING citus_source s
ON t.id = s.id AND s.id = 250
WHEN MATCHED THEN
UPDATE SET val = t.val || 'Updated by Merge'
WHEN NOT MATCHED THEN
DO NOTHING;
SET citus.log_remote_commands to false;
SELECT compare_tables();
ROLLBACK;
-- Test CTE with distributed tables
CREATE VIEW pg_source_view AS SELECT * FROM pg_source WHERE id < 400;
CREATE VIEW citus_source_view AS SELECT * FROM citus_source WHERE id < 400;
BEGIN;
SEt citus.log_remote_commands to true;
WITH cte AS (
SELECT * FROM pg_source_view
)
MERGE INTO pg_target t
USING cte
ON cte.id = t.id
WHEN MATCHED AND t.id > 350 THEN
UPDATE SET val = t.val || 'Updated by CTE'
WHEN NOT MATCHED THEN
INSERT VALUES (cte.id, cte.val)
WHEN MATCHED AND t.id < 350 THEN
DELETE;
WITH cte AS (
SELECT * FROM citus_source_view
)
MERGE INTO citus_target t
USING cte
ON cte.id = t.id
WHEN MATCHED AND t.id > 350 THEN
UPDATE SET val = t.val || 'Updated by CTE'
WHEN NOT MATCHED THEN
INSERT VALUES (cte.id, cte.val)
WHEN MATCHED AND t.id < 350 THEN
DELETE;
SET citus.log_remote_commands to false;
SELECT compare_tables();
ROLLBACK;
-- Test sub-query with distributed tables
BEGIN;
SEt citus.log_remote_commands to true;
MERGE INTO pg_target t
USING (SELECT * FROM pg_source) subq
ON subq.id = t.id
WHEN MATCHED AND t.id > 350 THEN
UPDATE SET val = t.val || 'Updated by subquery'
WHEN NOT MATCHED THEN
INSERT VALUES (subq.id, subq.val)
WHEN MATCHED AND t.id < 350 THEN
DELETE;
MERGE INTO citus_target t
USING (SELECT * FROM citus_source) subq
ON subq.id = t.id
WHEN MATCHED AND t.id > 350 THEN
UPDATE SET val = t.val || 'Updated by subquery'
WHEN NOT MATCHED THEN
INSERT VALUES (subq.id, subq.val)
WHEN MATCHED AND t.id < 350 THEN
DELETE;
SET citus.log_remote_commands to false;
SELECT compare_tables();
ROLLBACK;
-- Test PREPARE
PREPARE pg_prep(int) AS
MERGE INTO pg_target
USING (SELECT * FROM pg_source) sub
ON pg_target.id = sub.id AND pg_target.id = $1
WHEN MATCHED THEN
UPDATE SET val = 'Updated by prepare using ' || sub.val
WHEN NOT MATCHED THEN
DO NOTHING;
PREPARE citus_prep(int) AS
MERGE INTO citus_target
USING (SELECT * FROM citus_source) sub
ON citus_target.id = sub.id AND citus_target.id = $1
WHEN MATCHED THEN
UPDATE SET val = 'Updated by prepare using ' || sub.val
WHEN NOT MATCHED THEN
DO NOTHING;
BEGIN;
SET citus.log_remote_commands to true;
SELECT * FROM pg_target WHERE id = 500; -- before merge
EXECUTE pg_prep(500);
SELECT * FROM pg_target WHERE id = 500; -- non-cached
EXECUTE pg_prep(500);
EXECUTE pg_prep(500);
EXECUTE pg_prep(500);
EXECUTE pg_prep(500);
EXECUTE pg_prep(500);
SELECT * FROM pg_target WHERE id = 500; -- cached
SELECT * FROM citus_target WHERE id = 500; -- before merge
EXECUTE citus_prep(500);
SELECT * FROM citus_target WHERE id = 500; -- non-cached
EXECUTE citus_prep(500);
EXECUTE citus_prep(500);
EXECUTE citus_prep(500);
EXECUTE citus_prep(500);
EXECUTE citus_prep(500);
SELECT * FROM citus_target WHERE id = 500; -- cached
SET citus.log_remote_commands to false;
SELECT compare_tables();
ROLLBACK;
-- Test partitions + distributed tables
CREATE TABLE pg_pa_target (tid integer, balance float, val text)
PARTITION BY LIST (tid);
CREATE TABLE citus_pa_target (tid integer, balance float, val text)
PARTITION BY LIST (tid);
CREATE TABLE part1 PARTITION OF pg_pa_target FOR VALUES IN (1,4)
WITH (autovacuum_enabled=off);
CREATE TABLE part2 PARTITION OF pg_pa_target FOR VALUES IN (2,5,6)
WITH (autovacuum_enabled=off);
CREATE TABLE part3 PARTITION OF pg_pa_target FOR VALUES IN (3,8,9)
WITH (autovacuum_enabled=off);
CREATE TABLE part4 PARTITION OF pg_pa_target DEFAULT
WITH (autovacuum_enabled=off);
CREATE TABLE part5 PARTITION OF citus_pa_target FOR VALUES IN (1,4)
WITH (autovacuum_enabled=off);
CREATE TABLE part6 PARTITION OF citus_pa_target FOR VALUES IN (2,5,6)
WITH (autovacuum_enabled=off);
CREATE TABLE part7 PARTITION OF citus_pa_target FOR VALUES IN (3,8,9)
WITH (autovacuum_enabled=off);
CREATE TABLE part8 PARTITION OF citus_pa_target DEFAULT
WITH (autovacuum_enabled=off);
CREATE TABLE pg_pa_source (sid integer, delta float);
CREATE TABLE citus_pa_source (sid integer, delta float);
-- insert many rows to the source table
INSERT INTO pg_pa_source SELECT id, id * 10 FROM generate_series(1,14) AS id;
INSERT INTO citus_pa_source SELECT id, id * 10 FROM generate_series(1,14) AS id;
-- insert a few rows in the target table (odd numbered tid)
INSERT INTO pg_pa_target SELECT id, id * 100, 'initial' FROM generate_series(1,14,2) AS id;
INSERT INTO citus_pa_target SELECT id, id * 100, 'initial' FROM generate_series(1,14,2) AS id;
SELECT create_distributed_table('citus_pa_target', 'tid');
SELECT create_distributed_table('citus_pa_source', 'sid');
CREATE OR REPLACE FUNCTION pa_compare_tables() RETURNS BOOLEAN AS $$
DECLARE ret BOOL;
BEGIN
SELECT count(1) = 0 INTO ret
FROM pg_pa_target
FULL OUTER JOIN citus_pa_target
USING (tid, balance, val)
WHERE pg_pa_target.tid IS NULL
OR citus_pa_target.tid IS NULL;
RETURN ret;
END
$$ LANGUAGE PLPGSQL;
-- try simple MERGE
BEGIN;
MERGE INTO pg_pa_target t
USING pg_pa_source s
ON t.tid = s.sid
WHEN MATCHED THEN
UPDATE SET balance = balance + delta, val = val || ' updated by merge'
WHEN NOT MATCHED THEN
INSERT VALUES (sid, delta, 'inserted by merge');
MERGE INTO citus_pa_target t
USING citus_pa_source s
ON t.tid = s.sid
WHEN MATCHED THEN
UPDATE SET balance = balance + delta, val = val || ' updated by merge'
WHEN NOT MATCHED THEN
INSERT VALUES (sid, delta, 'inserted by merge');
SELECT pa_compare_tables();
ROLLBACK;
-- same with a constant qual
BEGIN;
MERGE INTO pg_pa_target t
USING pg_pa_source s
ON t.tid = s.sid AND tid = 1
WHEN MATCHED THEN
UPDATE SET balance = balance + delta, val = val || ' updated by merge'
WHEN NOT MATCHED THEN
INSERT VALUES (sid, delta, 'inserted by merge');
MERGE INTO citus_pa_target t
USING citus_pa_source s
ON t.tid = s.sid AND tid = 1
WHEN MATCHED THEN
UPDATE SET balance = balance + delta, val = val || ' updated by merge'
WHEN NOT MATCHED THEN
INSERT VALUES (sid, delta, 'inserted by merge');
SELECT pa_compare_tables();
ROLLBACK;
--
-- Error and Unsupported scenarios
--
-- try updating the distribution key column
BEGIN;
MERGE INTO target_cj t
USING source_cj1 s
ON t.tid = s.sid1 AND t.tid = 2
WHEN MATCHED THEN
UPDATE SET tid = tid + 9, src = src || ' updated by merge'
WHEN NOT MATCHED THEN
INSERT VALUES (sid1, 'inserted by merge', val1);
ROLLBACK;
-- Foreign table as target
MERGE INTO foreign_table
USING ft_target ON (foreign_table.id = ft_target.id)
@ -854,6 +1458,38 @@ MERGE INTO t1
WHEN NOT MATCHED THEN
INSERT (id, val) VALUES (s1.id, s1.val);
-- Now both s1 and t1 are distributed tables
SELECT undistribute_table('t1');
SELECT create_distributed_table('t1', 'id');
-- We have a potential pitfall where a function can be invoked in
-- the MERGE conditions which can insert/update to a random shard
CREATE OR REPLACE function merge_when_and_write() RETURNS BOOLEAN
LANGUAGE PLPGSQL AS
$$
BEGIN
INSERT INTO t1 VALUES (100, 100);
RETURN TRUE;
END;
$$;
-- Test preventing "ON" join condition from writing to the database
BEGIN;
MERGE INTO t1
USING s1 ON t1.id = s1.id AND t1.id = 2 AND (merge_when_and_write())
WHEN MATCHED THEN
UPDATE SET val = t1.val + s1.val;
ROLLBACK;
-- Test preventing WHEN clause(s) from writing to the database
BEGIN;
MERGE INTO t1
USING s1 ON t1.id = s1.id AND t1.id = 2
WHEN MATCHED AND (merge_when_and_write()) THEN
UPDATE SET val = t1.val + s1.val;
ROLLBACK;
-- Joining on partition columns with sub-query
MERGE INTO t1
USING (SELECT * FROM s1) sub ON (sub.val = t1.id) -- sub.val is not a distribution column
@ -997,6 +1633,104 @@ WHEN MATCHED THEN
WHEN NOT MATCHED THEN
INSERT VALUES(mv_source.id, mv_source.val);
-- Distributed tables *must* be colocated
CREATE TABLE dist_target(id int, val varchar);
SELECT create_distributed_table('dist_target', 'id');
CREATE TABLE dist_source(id int, val varchar);
SELECT create_distributed_table('dist_source', 'id', colocate_with => 'none');
MERGE INTO dist_target
USING dist_source
ON dist_target.id = dist_source.id
WHEN MATCHED THEN
UPDATE SET val = dist_source.val
WHEN NOT MATCHED THEN
INSERT VALUES(dist_source.id, dist_source.val);
-- Distributed tables *must* be joined on distribution column
CREATE TABLE dist_colocated(id int, val int);
SELECT create_distributed_table('dist_colocated', 'id', colocate_with => 'dist_target');
MERGE INTO dist_target
USING dist_colocated
ON dist_target.id = dist_colocated.val -- val is not the distribution column
WHEN MATCHED THEN
UPDATE SET val = dist_colocated.val
WHEN NOT MATCHED THEN
INSERT VALUES(dist_colocated.id, dist_colocated.val);
-- Both the source and target must be distributed
MERGE INTO dist_target
USING (SELECT 100 id) AS source
ON dist_target.id = source.id AND dist_target.val = 'const'
WHEN MATCHED THEN
UPDATE SET val = 'source'
WHEN NOT MATCHED THEN
INSERT VALUES(source.id, 'source');
-- Non-hash distributed tables (append/range).
CREATE VIEW show_tables AS
SELECT logicalrelid, partmethod
FROM pg_dist_partition
WHERE (logicalrelid = 'dist_target'::regclass) OR (logicalrelid = 'dist_source'::regclass)
ORDER BY 1;
SELECT undistribute_table('dist_source');
SELECT create_distributed_table('dist_source', 'id', 'append');
SELECT * FROM show_tables;
MERGE INTO dist_target
USING dist_source
ON dist_target.id = dist_source.id
WHEN MATCHED THEN
UPDATE SET val = dist_source.val
WHEN NOT MATCHED THEN
INSERT VALUES(dist_source.id, dist_source.val);
SELECT undistribute_table('dist_source');
SELECT create_distributed_table('dist_source', 'id', 'range');
SELECT * FROM show_tables;
MERGE INTO dist_target
USING dist_source
ON dist_target.id = dist_source.id
WHEN MATCHED THEN
UPDATE SET val = dist_source.val
WHEN NOT MATCHED THEN
INSERT VALUES(dist_source.id, dist_source.val);
-- Both are append tables
SELECT undistribute_table('dist_target');
SELECT undistribute_table('dist_source');
SELECT create_distributed_table('dist_target', 'id', 'append');
SELECT create_distributed_table('dist_source', 'id', 'append');
SELECT * FROM show_tables;
MERGE INTO dist_target
USING dist_source
ON dist_target.id = dist_source.id
WHEN MATCHED THEN
UPDATE SET val = dist_source.val
WHEN NOT MATCHED THEN
INSERT VALUES(dist_source.id, dist_source.val);
-- Both are range tables
SELECT undistribute_table('dist_target');
SELECT undistribute_table('dist_source');
SELECT create_distributed_table('dist_target', 'id', 'range');
SELECT create_distributed_table('dist_source', 'id', 'range');
SELECT * FROM show_tables;
MERGE INTO dist_target
USING dist_source
ON dist_target.id = dist_source.id
WHEN MATCHED THEN
UPDATE SET val = dist_source.val
WHEN NOT MATCHED THEN
INSERT VALUES(dist_source.id, dist_source.val);
DROP SERVER foreign_server CASCADE;
DROP FUNCTION merge_when_and_write();
DROP SCHEMA merge_schema CASCADE;
SELECT 1 FROM master_remove_node('localhost', :master_port);

View File

@ -608,6 +608,14 @@ USING wq_source s ON t.tid = s.sid
WHEN MATCHED AND (merge_when_and_write()) THEN
UPDATE SET balance = t.balance + s.balance;
ROLLBACK;
-- Test preventing ON condition from writing to the database
BEGIN;
MERGE INTO wq_target t
USING wq_source s ON t.tid = s.sid AND (merge_when_and_write())
WHEN MATCHED THEN
UPDATE SET balance = t.balance + s.balance;
ROLLBACK;
drop function merge_when_and_write();
DROP TABLE wq_target, wq_source;