Acquire relation locks on partitions while operation on parent table

pull/1544/head
Burak Yucesoy 2017-08-11 15:13:16 +03:00
parent 52b9e35d50
commit a321e750c0
5 changed files with 157 additions and 3 deletions

View File

@ -14,8 +14,10 @@
#include "distributed/insert_select_planner.h"
#include "distributed/multi_copy.h"
#include "distributed/multi_executor.h"
#include "distributed/multi_partitioning_utils.h"
#include "distributed/multi_physical_planner.h"
#include "distributed/multi_planner.h"
#include "distributed/resource_lock.h"
#include "distributed/transaction_management.h"
#include "executor/executor.h"
#include "nodes/execnodes.h"
@ -60,6 +62,16 @@ CoordinatorInsertSelectExecScan(CustomScanState *node)
ereport(DEBUG1, (errmsg("Collecting INSERT ... SELECT results on coordinator")));
/*
* If we are dealing with partitioned table, we also need to lock its
* partitions. Here we only lock targetRelation, we acquire necessary
* locks on selected tables during execution of those select queries.
*/
if (PartitionedTable(targetRelationId))
{
LockPartitionRelations(targetRelationId, RowExclusiveLock);
}
ExecuteSelectIntoRelation(targetRelationId, insertTargetList, selectQuery,
executorState);

View File

@ -25,6 +25,7 @@
#include "distributed/multi_resowner.h"
#include "distributed/multi_server_executor.h"
#include "distributed/multi_utility.h"
#include "distributed/resource_lock.h"
#include "distributed/worker_protocol.h"
#include "executor/execdebug.h"
#include "commands/copy.h"
@ -248,6 +249,9 @@ RealTimeExecScan(CustomScanState *node)
MultiPlan *multiPlan = scanState->multiPlan;
Job *workerJob = multiPlan->workerJob;
/* we are taking locks on partitions of partitioned tables */
LockPartitionsInRelationList(multiPlan->relationIdList, AccessShareLock);
PrepareMasterJobDirectory(workerJob);
MultiRealTimeExecute(workerJob);
@ -445,6 +449,9 @@ TaskTrackerExecScan(CustomScanState *node)
MultiPlan *multiPlan = scanState->multiPlan;
Job *workerJob = multiPlan->workerJob;
/* we are taking locks on partitions of partitioned tables */
LockPartitionsInRelationList(multiPlan->relationIdList, AccessShareLock);
PrepareMasterJobDirectory(workerJob);
MultiTaskTrackerExecute(workerJob);

View File

@ -34,6 +34,7 @@
#include "distributed/master_metadata_utility.h"
#include "distributed/metadata_cache.h"
#include "distributed/multi_executor.h"
#include "distributed/multi_partitioning_utils.h"
#include "distributed/multi_physical_planner.h"
#include "distributed/multi_planner.h"
#include "distributed/multi_router_executor.h"
@ -98,6 +99,7 @@ static bool StoreQueryResult(CitusScanState *scanState, MultiConnection *connect
bool failOnError, int64 *rows);
static bool ConsumeQueryResult(MultiConnection *connection, bool failOnError,
int64 *rows);
static LOCKMODE LockModeForModifyTask(Task *task);
/*
@ -434,6 +436,12 @@ CitusModifyBeginScan(CustomScanState *node, EState *estate, int eflags)
/* prevent concurrent placement changes */
AcquireMetadataLocks(taskList);
/*
* We are taking locks on partitions of partitioned tables. These locks are
* necessary for locking tables that appear in the SELECT part of the query.
*/
LockPartitionsInRelationList(multiPlan->relationIdList, AccessShareLock);
/* modify tasks are always assigned using first-replica policy */
workerJob->taskList = FirstReplicaAssignTaskList(taskList);
}
@ -519,6 +527,9 @@ RouterSelectExecScan(CustomScanState *node)
Job *workerJob = multiPlan->workerJob;
List *taskList = workerJob->taskList;
/* we are taking locks on partitions of partitioned tables */
LockPartitionsInRelationList(multiPlan->relationIdList, AccessShareLock);
if (list_length(taskList) > 0)
{
Task *task = (Task *) linitial(taskList);
@ -682,6 +693,9 @@ ExecuteSingleModifyTask(CitusScanState *scanState, Task *task, bool expectResult
char *queryString = task->queryString;
bool taskRequiresTwoPhaseCommit = (task->replicationModel == REPLICATION_MODEL_2PC);
ShardInterval *shardInterval = LoadShardInterval(task->anchorShardId);
Oid relationId = shardInterval->relationId;
/*
* Modifications for reference tables are always done using 2PC. First
* ensure that distributed transaction is started. Then force the
@ -712,6 +726,17 @@ ExecuteSingleModifyTask(CitusScanState *scanState, Task *task, bool expectResult
*/
connectionList = GetModifyConnections(task, taskRequiresTwoPhaseCommit);
/*
* If we are dealing with a partitioned table, we also need to lock its
* partitions.
*/
if (PartitionedTable(relationId))
{
LOCKMODE lockMode = LockModeForModifyTask(task);
LockPartitionRelations(relationId, lockMode);
}
/* prevent replicas of the same shard from diverging */
AcquireExecutorShardLock(task, operation);
@ -947,6 +972,7 @@ ExecuteModifyTasks(List *taskList, bool expectResults, ParamListInfo paramListIn
int64 totalAffectedTupleCount = 0;
ListCell *taskCell = NULL;
Task *firstTask = NULL;
ShardInterval *firstShardInterval = NULL;
int connectionFlags = 0;
List *affectedTupleCountList = NIL;
HTAB *shardConnectionHash = NULL;
@ -958,13 +984,25 @@ ExecuteModifyTasks(List *taskList, bool expectResults, ParamListInfo paramListIn
return 0;
}
/*
* In multi shard modification, we expect that all tasks operates on the
* same relation, so it is enough to acquire a lock on the first task's
* anchor relation's partitions.
*/
firstTask = (Task *) linitial(taskList);
firstShardInterval = LoadShardInterval(firstTask->anchorShardId);
if (PartitionedTable(firstShardInterval->relationId))
{
LOCKMODE lockMode = LockModeForModifyTask(firstTask);
LockPartitionRelations(firstShardInterval->relationId, lockMode);
}
/* ensure that there are no concurrent modifications on the same shards */
AcquireExecutorMultiShardLocks(taskList);
BeginOrContinueCoordinatedTransaction();
firstTask = (Task *) linitial(taskList);
if (MultiShardCommitProtocol == COMMIT_PROTOCOL_2PC ||
firstTask->replicationModel == REPLICATION_MODEL_2PC)
{
@ -1464,3 +1502,29 @@ ConsumeQueryResult(MultiConnection *connection, bool failOnError, int64 *rows)
return gotResponse && !commandFailed;
}
/*
* LockModeForRouterModifyTask returns appropriate LOCKMODE for given router
* modify task.
*/
static LOCKMODE
LockModeForModifyTask(Task *task)
{
LOCKMODE lockMode = NoLock;
if (task->taskType == DDL_TASK)
{
lockMode = AccessExclusiveLock;
}
else if (task->taskType == MODIFY_TASK)
{
lockMode = RowExclusiveLock;
}
else
{
/* we do not allow any other task type in these code path */
Assert(false);
}
return lockMode;
}

View File

@ -17,11 +17,13 @@
#include "c.h"
#include "miscadmin.h"
#include "distributed/colocation_utils.h"
#include "distributed/listutils.h"
#include "distributed/master_metadata_utility.h"
#include "distributed/metadata_cache.h"
#include "distributed/multi_router_executor.h"
#include "distributed/multi_partitioning_utils.h"
#include "distributed/multi_planner.h"
#include "distributed/multi_router_executor.h"
#include "distributed/relay_utility.h"
#include "distributed/resource_lock.h"
#include "distributed/shardinterval_utils.h"
@ -316,3 +318,69 @@ LockRelationShardResources(List *relationShardList, LOCKMODE lockMode)
}
}
}
/*
* LockParentShardResourceIfPartition checks whether given shard belongs to a
* partition. If it is, LockParentShardResourceIfPartition acquires a shard
* resource lock on the colocated shard of the parent table.
*/
void
LockParentShardResourceIfPartition(uint64 shardId, LOCKMODE lockMode)
{
ShardInterval *shardInterval = LoadShardInterval(shardId);
Oid relationId = shardInterval->relationId;
if (PartitionTable(relationId))
{
int shardIndex = ShardIndex(shardInterval);
Oid parentRelationId = PartitionParentOid(relationId);
uint64 parentShardId = ColocatedShardIdInRelation(parentRelationId, shardIndex);
LockShardResource(parentShardId, lockMode);
}
}
/*
* LockPartitionsInRelationList iterates over given list and acquires locks on
* partitions of each partitioned table. It does nothing for non-partitioned tables.
*/
void
LockPartitionsInRelationList(List *relationIdList, LOCKMODE lockmode)
{
ListCell *relationIdCell = NULL;
foreach(relationIdCell, relationIdList)
{
Oid relationId = lfirst_oid(relationIdCell);
if (PartitionedTable(relationId))
{
LockPartitionRelations(relationId, lockmode);
}
}
}
/*
* LockPartitionRelations acquires relation lock on all partitions of given
* partitioned relation. This function expects that given relation is a
* partitioned relation.
*/
void
LockPartitionRelations(Oid relationId, LOCKMODE lockMode)
{
/*
* PartitionList function generates partition list in the same order
* as PostgreSQL. Therefore we do not need to sort it before acquiring
* locks.
*/
List *partitionList = PartitionList(relationId);
ListCell *partitionCell = NULL;
foreach(partitionCell, partitionList)
{
Oid partitionRelationId = lfirst_oid(partitionCell);
LockRelationOid(partitionRelationId, lockMode);
}
}

View File

@ -80,5 +80,8 @@ extern void LockShardListMetadata(List *shardIntervalList, LOCKMODE lockMode);
extern void LockShardListResources(List *shardIntervalList, LOCKMODE lockMode);
extern void LockRelationShardResources(List *relationShardList, LOCKMODE lockMode);
/* Lock partitions of partitioned table */
extern void LockPartitionsInRelationList(List *relationIdList, LOCKMODE lockmode);
extern void LockPartitionRelations(Oid relationId, LOCKMODE lockMode);
#endif /* RESOURCE_LOCK_H */