Streamlined Lock Distribution

* Packaged the logic in a single method which can be invoked by different operations which need to acquire distributed locks.
* Error checking, cleaned up some duplicated code, removed some redundant validation, fixed bug in reading nowait in `lock_relation_if_exists`
pull/5899/head
gledis69 2022-04-19 13:12:20 +03:00
parent a7fa46ea63
commit 3debcd4f38
10 changed files with 188 additions and 149 deletions

View File

@ -43,7 +43,6 @@
static void ErrorIfUnsupportedTruncateStmt(TruncateStmt *truncateStatement);
static void ExecuteTruncateStmtSequentialIfNecessary(TruncateStmt *command);
static void EnsurePartitionTableNotReplicatedForTruncate(TruncateStmt *truncateStatement);
static void LockTruncatedRelationMetadataInWorkers(TruncateStmt *truncateStatement);
static List * TruncateTaskList(Oid relationId);
@ -243,7 +242,8 @@ PreprocessTruncateStatement(TruncateStmt *truncateStatement)
ErrorIfUnsupportedTruncateStmt(truncateStatement);
EnsurePartitionTableNotReplicatedForTruncate(truncateStatement);
ExecuteTruncateStmtSequentialIfNecessary(truncateStatement);
LockTruncatedRelationMetadataInWorkers(truncateStatement);
AcquireDistributedLockOnRelations(truncateStatement->relations, AccessExclusiveLock,
DIST_LOCK_REFERENCING_TABLES);
}
@ -340,65 +340,3 @@ ExecuteTruncateStmtSequentialIfNecessary(TruncateStmt *command)
}
}
}
/*
* LockTruncatedRelationMetadataInWorkers determines if distributed
* lock is necessary for truncated relations, and acquire locks.
*
* LockTruncatedRelationMetadataInWorkers handles distributed locking
* of truncated tables before standard utility takes over.
*
* Actual distributed truncation occurs inside truncate trigger.
*
* This is only for distributed serialization of truncate commands.
* The function assumes that there is no foreign key relation between
* non-distributed and distributed relations.
*/
static void
LockTruncatedRelationMetadataInWorkers(TruncateStmt *truncateStatement)
{
List *distributedRelationList = NIL;
/* nothing to do if there is no metadata at worker nodes */
if (!ClusterHasKnownMetadataWorkers())
{
return;
}
RangeVar *rangeVar = NULL;
foreach_ptr(rangeVar, truncateStatement->relations)
{
Oid relationId = RangeVarGetRelid(rangeVar, NoLock, false);
Oid referencingRelationId = InvalidOid;
if (!IsCitusTable(relationId))
{
continue;
}
if (list_member_oid(distributedRelationList, relationId))
{
continue;
}
distributedRelationList = lappend_oid(distributedRelationList, relationId);
CitusTableCacheEntry *cacheEntry = GetCitusTableCacheEntry(relationId);
Assert(cacheEntry != NULL);
List *referencingTableList = cacheEntry->referencingRelationsViaForeignKey;
foreach_oid(referencingRelationId, referencingTableList)
{
distributedRelationList = list_append_unique_oid(distributedRelationList,
referencingRelationId);
}
}
if (distributedRelationList != NIL)
{
bool nowait = false;
AcquireDistributedLockOnRelations(distributedRelationList, AccessExclusiveLock,
nowait);
}
}

View File

@ -471,31 +471,13 @@ ProcessUtilityInternal(PlannedStmt *pstmt,
if (IsA(parsetree, LockStmt))
{
LockStmt *stmt = (LockStmt *) parsetree;
List *distributedRelationList = NIL;
RangeVar *rangeVar = NULL;
foreach_ptr(rangeVar, stmt->relations)
{
Oid relationId = RangeVarGetRelid(rangeVar, NoLock, false);
ereport(NOTICE, errmsg("Processing LOCK command."));
if (!IsCitusTable(relationId))
{
continue;
}
if (list_member_oid(distributedRelationList, relationId))
{
continue;
}
distributedRelationList = lappend_oid(distributedRelationList, relationId);
}
if (distributedRelationList != NIL)
{
AcquireDistributedLockOnRelations(distributedRelationList, stmt->mode,
stmt->nowait);
}
ErrorIfUnsupportedLockStmt(stmt);
uint32 nowaitFlag = stmt->nowait ? DIST_LOCK_NOWAIT : 0;
AcquireDistributedLockOnRelations(stmt->relations, stmt->mode,
DIST_LOCK_VIEWS_RECUR | nowaitFlag);
}
/*

View File

@ -105,3 +105,5 @@ GRANT SELECT ON pg_catalog.pg_dist_object TO public;
#include "udfs/citus_pid_for_gpid/11.0-1.sql"
#include "udfs/citus_coordinator_nodeid/11.0-1.sql"
#include "udfs/lock_relation_if_exists/11.0-1.sql"

View File

@ -6,14 +6,3 @@ DROP FUNCTION pg_catalog.worker_hash_partition_table(bigint, integer, text, text
DROP FUNCTION pg_catalog.worker_merge_files_into_table(bigint, integer, text[], text[]);
DROP FUNCTION pg_catalog.worker_range_partition_table(bigint, integer, text, text, oid, anyarray);
DROP FUNCTION pg_catalog.worker_repartition_cleanup(bigint);
SET search_path = 'pg_catalog';
CREATE OR REPLACE FUNCTION lock_relation_if_exists(table_name text, lock_mode text, nowait boolean)
RETURNS BOOL
LANGUAGE C STRICT as 'MODULE_PATHNAME',
$$lock_relation_if_exists$$;
COMMENT ON FUNCTION lock_relation_if_exists(table_name text, lock_mode text, nowait boolean)
IS 'locks relation in the lock_mode if the relation exists';
RESET search_path;

View File

@ -0,0 +1,6 @@
CREATE OR REPLACE FUNCTION pg_catalog.lock_relation_if_exists(table_name text, lock_mode text, nowait boolean)
RETURNS BOOL
LANGUAGE C STRICT as 'MODULE_PATHNAME',
$$lock_relation_if_exists$$;
COMMENT ON FUNCTION pg_catalog.lock_relation_if_exists(table_name text, lock_mode text, nowait boolean)
IS 'locks relation in the lock_mode if the relation exists';

View File

@ -0,0 +1,6 @@
CREATE OR REPLACE FUNCTION pg_catalog.lock_relation_if_exists(table_name text, lock_mode text, nowait boolean)
RETURNS BOOL
LANGUAGE C STRICT as 'MODULE_PATHNAME',
$$lock_relation_if_exists$$;
COMMENT ON FUNCTION pg_catalog.lock_relation_if_exists(table_name text, lock_mode text, nowait boolean)
IS 'locks relation in the lock_mode if the relation exists';

View File

@ -36,6 +36,7 @@
#include "distributed/resource_lock.h"
#include "distributed/shardinterval_utils.h"
#include "distributed/worker_protocol.h"
#include "distributed/worker_shard_visibility.h"
#include "distributed/utils/array_type.h"
#include "distributed/version_compat.h"
#include "storage/lmgr.h"
@ -1006,7 +1007,7 @@ lock_relation_if_exists(PG_FUNCTION_ARGS)
char *lockModeCString = text_to_cstring(lockModeText);
bool nowait = false;
if (!PG_ARGISNULL(2))
if (PG_NARGS() == 3)
{
nowait = PG_GETARG_BOOL(2);
}
@ -1088,32 +1089,31 @@ CitusLockTableAclCheck(Oid relationId, LOCKMODE lockmode, Oid userId)
/*
* AcquireDistributedLockOnRelations acquire a distributed lock on worker nodes
* for given list of relations ids. Relation id list and worker node list
* sorted so that the lock is acquired in the same order regardless of which
* node it was run on. Notice that no lock is acquired on coordinator node.
* AcquireDistributedLockOnRelations_Internal acquire a distributed lock on worker nodes
* for given list of relations ids. Worker node list is sorted so that the lock
* is acquired in the same order regardless of which node it was run on.
* Notice that no lock is acquired on coordinator node if the coordinator is not
* added to the metadata.
*
* Notice that the locking functions is sent to all workers regardless of if
* it has metadata or not. This is because a worker node only knows itself
* and previous workers that has metadata sync turned on. The node does not
* know about other nodes that have metadata sync turned on afterwards.
* Notice that no validation is done on the relationIds, that is the responsibility of
* AcquireDistributedLockOnRelations.
*
* A nowait flag is used to require the locks to be available immediately
* and if that is not the case, an error will be thrown
*/
void
AcquireDistributedLockOnRelations(List *relationIdList, LOCKMODE lockMode, bool nowait)
static void
AcquireDistributedLockOnRelations_Internal(List *relationIdList, LOCKMODE lockMode, bool
nowait)
{
Oid relationId = InvalidOid;
List *workerNodeList = ActivePrimaryNodeList(NoLock);
Oid userId = GetUserId();
const char *lockModeCString = LockModeToLockModeCString(lockMode);
/*
* We want to acquire locks in the same order across the nodes.
* Although relation ids may change, their ordering will not.
*/
relationIdList = SortList(relationIdList, CompareOids);
workerNodeList = SortList(workerNodeList, CompareWorkerNodes);
UseCoordinatedTransaction();
@ -1122,44 +1122,153 @@ AcquireDistributedLockOnRelations(List *relationIdList, LOCKMODE lockMode, bool
foreach_oid(relationId, relationIdList)
{
/*
* We only acquire distributed lock on relation if
* the relation is sync'ed between mx nodes.
*
* Even if users disable metadata sync, we cannot
* allow them not to acquire the remote locks.
* Hence, we have !IsCoordinator() check.
*/
if (ShouldSyncTableMetadata(relationId) || !IsCoordinator())
char *qualifiedRelationName = generate_qualified_relation_name(relationId);
StringInfo lockRelationCommand = makeStringInfo();
char *lockCommand = nowait ? LOCK_RELATION_IF_EXISTS_NOWAIT :
LOCK_RELATION_IF_EXISTS;
appendStringInfo(lockRelationCommand, lockCommand,
quote_literal_cstr(qualifiedRelationName),
lockModeCString);
/* preemptive permission check so a worker connection is not */
/* established if the user is not allowed to acquire the lock */
CitusLockTableAclCheck(relationId, lockMode, userId);
WorkerNode *workerNode = NULL;
foreach_ptr(workerNode, workerNodeList)
{
char *qualifiedRelationName = generate_qualified_relation_name(relationId);
StringInfo lockRelationCommand = makeStringInfo();
const char *nodeName = workerNode->workerName;
int nodePort = workerNode->workerPort;
char *lockCommand = nowait ? LOCK_RELATION_IF_EXISTS_NOWAIT :
LOCK_RELATION_IF_EXISTS;
appendStringInfo(lockRelationCommand, lockCommand,
quote_literal_cstr(qualifiedRelationName),
lockModeCString);
WorkerNode *workerNode = NULL;
foreach_ptr(workerNode, workerNodeList)
/* if local node is one of the targets, acquire the lock locally */
if (workerNode->groupId == localGroupId)
{
const char *nodeName = workerNode->workerName;
int nodePort = workerNode->workerPort;
LockRelationIfExists(
cstring_to_text(qualifiedRelationName),
lockModeCString,
nowait);
/* if local node is one of the targets, acquire the lock locally */
if (workerNode->groupId == localGroupId)
{
LockRelationIfExists(
cstring_to_text(qualifiedRelationName),
lockModeCString,
nowait
);
continue;
}
SendCommandToWorker(nodeName, nodePort, lockRelationCommand->data);
continue;
}
SendCommandToWorker(nodeName, nodePort, lockRelationCommand->data);
}
}
}
/*
* AcquireDistributedLockOnRelations filters relations before passing them to
* AcquireDistributedLockOnRelations_Internal to acquire the locks.
*
* Only tables, views, and foreign tables can be locked with this function. Other relations
* will cause an error.
*
* Gracefully locks relations so no errors are thrown for things like invalid id-s
* or missing relations on worker nodes.
*
* Considers different types of relations based on a 'configs' parameter:
* - DIST_LOCK_DEFAULT: locks citus tables
* - DIST_LOCK_VIEWS_RECUR: locks citus tables that locked views are dependent on recursively
* - DIST_LOCK_REFERENCING_TABLES: locks tables that refer to locked citus tables with a foreign key
* - DIST_LOCK_NOWAIT: throws an error if the lock is not immediately available
*/
void
AcquireDistributedLockOnRelations(List *relationList, LOCKMODE lockMode, uint32 configs)
{
/* nothing to do if there is no metadata at worker nodes */
if (!ClusterHasKnownMetadataWorkers())
{
return;
}
List *distributedRelationList = NIL;
RangeVar *rangeVar = NULL;
List *relations = list_copy(relationList);
foreach_ptr(rangeVar, relations)
{
Oid relationId = RangeVarGetRelid(rangeVar, NoLock, false);
if ((configs & DIST_LOCK_VIEWS_RECUR) > 0 && get_rel_relkind(relationId) ==
RELKIND_VIEW)
{
ObjectAddress viewAddress = { 0 };
Oid schemaOid = RangeVarGetCreationNamespace(rangeVar);
ObjectAddressSet(viewAddress, schemaOid, relationId);
List *distDependencies = GetDistributableDependenciesForObject(&viewAddress);
ObjectAddress *address = NULL;
foreach_ptr(address, distDependencies)
{
distributedRelationList = list_append_unique_oid(distributedRelationList,
address->objectId);
}
continue;
}
if (!IsCitusTable(relationId))
{
continue;
}
if (list_member_oid(distributedRelationList, relationId))
{
continue;
}
distributedRelationList = lappend_oid(distributedRelationList, relationId);
if ((configs & DIST_LOCK_REFERENCING_TABLES) > 0)
{
Oid referencingRelationId = InvalidOid;
CitusTableCacheEntry *cacheEntry = GetCitusTableCacheEntry(relationId);
Assert(cacheEntry != NULL);
List *referencingTableList = cacheEntry->referencingRelationsViaForeignKey;
foreach_oid(referencingRelationId, referencingTableList)
{
distributedRelationList = list_append_unique_oid(distributedRelationList,
referencingRelationId);
}
}
}
if (distributedRelationList != NIL)
{
bool nowait = (configs & DIST_LOCK_NOWAIT) > 0;
AcquireDistributedLockOnRelations_Internal(distributedRelationList, lockMode,
nowait);
}
}
/*
* ErrorIfUnsupportedLockStmt errors out if:
* - The lock statement is not in a transaction block
* - The relation id-s being locked do not exist
* - Locking shard, but citus.enable_manual_changes_to_shards is false
*/
void
ErrorIfUnsupportedLockStmt(LockStmt *stmt)
{
RequireTransactionBlock(true, "LOCK TABLE");
RangeVar *rangeVar = NULL;
foreach_ptr(rangeVar, stmt->relations)
{
bool missingOk = false;
Oid relationId = RangeVarGetRelid(rangeVar, NoLock, missingOk);
/* Note that allowing the user to lock shards could lead to */
/* distributed deadlocks due to shards not being locked when */
/* a distributed table is locked. */
/* However, because citus.enable_manual_changes_to_shards */
/* is a guc which is not visible by default, whoever is using this */
/* guc will hopefully know what they're doing and avoid such scenarios. */
ErrorIfIllegallyChangingKnownShard(relationId);
}
}

View File

@ -13,6 +13,7 @@
#define CITUS_LISTUTILS_H
#include "postgres.h"
#include "storage/lock.h"
#include "c.h"
#include "nodes/pg_list.h"

View File

@ -16,6 +16,7 @@
#include "distributed/worker_transaction.h"
#include "nodes/pg_list.h"
#include "storage/lock.h"
#include "tcop/utility.h"
/*
@ -110,9 +111,14 @@ typedef enum CitusOperations
ADV_LOCKTAG_CLASS_CITUS_PLACEMENT_CLEANUP)
#define LOCK_RELATION_IF_EXISTS "SELECT lock_relation_if_exists(%s, '%s');"
#define DIST_LOCK_DEFAULT 0
#define DIST_LOCK_VIEWS_RECUR 1 << 0
#define DIST_LOCK_REFERENCING_TABLES 1 << 1
#define DIST_LOCK_NOWAIT 1 << 2
#define LOCK_RELATION_IF_EXISTS "SELECT pg_catalog.lock_relation_if_exists(%s, '%s');"
#define LOCK_RELATION_IF_EXISTS_NOWAIT \
"SELECT lock_relation_if_exists(%s, '%s', nowait => true);"
"SELECT pg_catalog.lock_relation_if_exists(%s, '%s', nowait => true);"
/* Lock shard/relation metadata for safe modifications */
extern void LockShardDistributionMetadata(int64 shardId, LOCKMODE lockMode);
@ -155,6 +161,7 @@ extern void LockParentShardResourceIfPartition(List *shardIntervalList,
/* Lock mode translation between text and enum */
extern LOCKMODE LockModeCStringToLockMode(const char *lockModeName);
extern const char * LockModeToLockModeCString(LOCKMODE lockMode);
extern void AcquireDistributedLockOnRelations(List *relationIdList, LOCKMODE lockMode,
bool nowait);
extern void AcquireDistributedLockOnRelations(List *relationList, LOCKMODE lockMode,
uint32 configs);
extern void ErrorIfUnsupportedLockStmt(LockStmt *stmt);
#endif /* RESOURCE_LOCK_H */

View File

@ -89,7 +89,6 @@ INSERT INTO super_user_owned_regular_user_granted VALUES (1, 1), (2, 1) ON CONFL
ERROR: permission denied for table super_user_owned_regular_user_granted
TRUNCATE super_user_owned_regular_user_granted;
ERROR: permission denied for table super_user_owned_regular_user_granted
CONTEXT: while executing command on localhost:xxxxx
DELETE FROM super_user_owned_regular_user_granted;
ERROR: permission denied for table super_user_owned_regular_user_granted
UPDATE super_user_owned_regular_user_granted SET a = 1;