external
/
citus
mirror of https://github.com/citusdata/citus.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Merge pull request #1455 from citusdata/join_xact_modification_levels 

Rework connection API to remove transaction restrictions
pull/1462/head
Marco Slot 2017-07-12 12:49:54 +02:00 committed by GitHub
parent 0d569b722e 9f7e4769e2
commit 04e2d64764
22 changed files with 1057 additions and 470 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

744
src/backend/distributed/connection/placement_connection.c
View File

@ -14,7 +14,9 @@
#include "access/hash.h"
#include "distributed/connection_management.h"
#include "distributed/hash_helpers.h"
#include "distributed/master_protocol.h"
#include "distributed/metadata_cache.h"
#include "distributed/multi_planner.h"
#include "distributed/placement_connection.h"
#include "utils/hsearch.h"
#include "utils/memutils.h"
@ -47,17 +49,14 @@ typedef struct ConnectionReference
bool hadDML;
bool hadDDL;
/*
* Membership in ConnectionPlacementHashEntry->connectionReferences or
* ColocatedPlacementsHashEntry->connectionReferences.
*/
dlist_node placementNode;
/* membership in MultiConnection->referencedPlacements */
dlist_node connectionNode;
} ConnectionReference;
struct ColocatedPlacementsHashEntry;
/*
* Hash table mapping placements to a list of connections.
*
@ -85,11 +84,14 @@ typedef struct ConnectionPlacementHashEntry
/* did any remote transactions fail? */
bool failed;
/* list of connections to remote nodes */
dlist_head connectionReferences;
/* primary connection used to access the placement */
ConnectionReference *primaryConnection;
/* if non-NULL, connection executing DML/DDL*/
ConnectionReference *modifyingConnection;
/* are any other connections reading from the placements? */
bool hasSecondaryConnections;
/* entry for the set of co-located placements */
struct ColocatedPlacementsHashEntry *colocatedEntry;
/* membership in ConnectionShardHashEntry->placementConnections */
dlist_node shardNode;
@ -137,11 +139,11 @@ typedef struct ColocatedPlacementsHashEntry
{
ColocatedPlacementsHashKey key;
/* list of connections to remote nodes */
dlist_head connectionReferences;
/* primary connection used to access the co-located placements */
ConnectionReference *primaryConnection;
/* if non-NULL, connection executing DML/DDL*/
ConnectionReference *modifyingConnection;
/* are any other connections reading from the placements? */
bool hasSecondaryConnections;
} ColocatedPlacementsHashEntry;
static HTAB *ColocatedPlacementsHash;
@ -172,19 +174,10 @@ typedef struct ConnectionShardHashEntry
static HTAB *ConnectionShardHash;
static MultiConnection * StartColocatedPlacementConnection(uint32 flags,
ShardPlacement *placement,
const char *userName);
static ConnectionReference * CheckExistingPlacementConnections(uint32 flags,
ConnectionPlacementHashEntry
*placementEntry,
const char *userName);
static ConnectionReference * CheckExistingColocatedConnections(uint32 flags,
ColocatedPlacementsHashEntry
*connectionEntry,
const char *userName);
static ConnectionPlacementHashEntry * FindOrCreatePlacementEntry(
ShardPlacement *placement);
static bool CanUseExistingConnection(uint32 flags, const char *userName,
ConnectionReference *connectionReference);
ConnectionReference *placementConnection);
static void AssociatePlacementWithShard(ConnectionPlacementHashEntry *placementEntry,
ShardPlacement *placement);
static bool CheckShardPlacements(ConnectionShardHashEntry *shardEntry);
@ -229,76 +222,355 @@ GetPlacementConnection(uint32 flags, ShardPlacement *placement, const char *user
MultiConnection *
StartPlacementConnection(uint32 flags, ShardPlacement *placement, const char *userName)
{
ConnectionPlacementHashKey key;
ConnectionPlacementHashEntry *placementEntry = NULL;
ConnectionReference *returnConnectionReference = NULL;
ShardPlacementAccess *placementAccess =
(ShardPlacementAccess *) palloc0(sizeof(ShardPlacementAccess));
placementAccess->placement = placement;
if (flags & FOR_DDL)
{
placementAccess->accessType = PLACEMENT_ACCESS_DDL;
}
else if (flags & FOR_DML)
{
placementAccess->accessType = PLACEMENT_ACCESS_DML;
}
else
{
placementAccess->accessType = PLACEMENT_ACCESS_SELECT;
}
return StartPlacementListConnection(flags, list_make1(placementAccess), userName);
}
/*
* GetPlacementListConnection establishes a connection for a set of placement
* accesses.
*
* See StartPlacementListConnection for details.
*/
MultiConnection *
GetPlacementListConnection(uint32 flags, List *placementAccessList, const char *userName)
{
MultiConnection *connection = StartPlacementListConnection(flags, placementAccessList,
userName);
FinishConnectionEstablishment(connection);
return connection;
}
/*
* StartPlacementListConnection returns a connection to a remote node suitable for
* a placement accesses (SELECT, DML, DDL) or throws an error if no suitable
* connection can be established if would cause a self-deadlock or consistency
* violation.
*/
MultiConnection *
StartPlacementListConnection(uint32 flags, List *placementAccessList,
const char *userName)
{
char *freeUserName = NULL;
bool found = false;
bool foundModifyingConnection = false;
ListCell *placementAccessCell = NULL;
List *placementEntryList = NIL;
ListCell *placementEntryCell = NULL;
MultiConnection *chosenConnection = NULL;
if (userName == NULL)
{
userName = freeUserName = CurrentUserName();
}
key.placementId = placement->placementId;
/* lookup relevant hash entry */
placementEntry = hash_search(ConnectionPlacementHash, &key, HASH_ENTER, &found);
if (!found)
{
dlist_init(&placementEntry->connectionReferences);
placementEntry->failed = false;
placementEntry->modifyingConnection = NULL;
}
/*
* Check whether any of the connections already associated with the
* placement can be reused, or violates FOR_DML/FOR_DDL constraints.
*/
returnConnectionReference =
CheckExistingPlacementConnections(flags, placementEntry, userName);
/*
* Either no caching desired, or no connection already associated with the
* placement present. Check whether there's a usable connection associated
* for the set of colocated placements, or establish a new one.
* Go through all placement accesses to find a suitable connection.
*
* Allocations are performed in transaction context, so we don't have to
* care about freeing in case of an early disconnect.
* If none of the placements have been accessed in this transaction, connection
* remains NULL.
*
* If one or more of the placements have been modified in this transaction, then
* use the connection that performed the write. If placements have been written
* over multiple connections or the connection is not available, error out.
*
* If placements have only been read in this transaction, then use the last
* suitable connection found for a placement in the placementAccessList.
*/
if (returnConnectionReference == NULL)
foreach(placementAccessCell, placementAccessList)
{
MultiConnection *connection =
StartColocatedPlacementConnection(flags, placement, userName);
ShardPlacementAccess *placementAccess =
(ShardPlacementAccess *) lfirst(placementAccessCell);
ShardPlacement *placement = placementAccess->placement;
ShardPlacementAccessType accessType = placementAccess->accessType;
returnConnectionReference = (ConnectionReference *)
MemoryContextAlloc(TopTransactionContext,
sizeof(ConnectionReference));
returnConnectionReference->connection = connection;
returnConnectionReference->hadDDL = false;
returnConnectionReference->hadDML = false;
returnConnectionReference->userName =
MemoryContextStrdup(TopTransactionContext, userName);
dlist_push_tail(&placementEntry->connectionReferences,
&returnConnectionReference->placementNode);
ConnectionPlacementHashEntry *placementEntry = NULL;
ColocatedPlacementsHashEntry *colocatedEntry = NULL;
ConnectionReference *placementConnection = NULL;
/* record association with shard, for invalidation */
AssociatePlacementWithShard(placementEntry, placement);
if (placement->shardId == INVALID_SHARD_ID)
{
/*
* When a SELECT prunes down to 0 shard, we use a dummy placement.
* In that case, we can fall back to the default connection.
*
* FIXME: this can be removed if we evaluate empty SELECTs locally.
*/
continue;
}
/* record association with connection, to handle connection closure */
dlist_push_tail(&connection->referencedPlacements,
&returnConnectionReference->connectionNode);
placementEntry = FindOrCreatePlacementEntry(placement);
colocatedEntry = placementEntry->colocatedEntry;
placementConnection = placementEntry->primaryConnection;
/* note: the Asserts below are primarily for clarifying the conditions */
if (placementConnection->connection == NULL)
{
/* no connection has been chosen for the placement */
}
else if (accessType == PLACEMENT_ACCESS_DDL &&
placementEntry->hasSecondaryConnections)
{
/*
* If a placement has been read over multiple connections (typically as
* a result of a reference table join) then a DDL command on the placement
* would create a self-deadlock.
*/
Assert(placementConnection != NULL);
ereport(ERROR,
(errcode(ERRCODE_ACTIVE_SQL_TRANSACTION),
errmsg(
"cannot perform DDL on placement %ld, which has been read over "
"multiple connections",
placement->placementId)));
}
else if (accessType == PLACEMENT_ACCESS_DDL && colocatedEntry != NULL &&
colocatedEntry->hasSecondaryConnections)
{
/*
* If a placement has been read over multiple (uncommitted) connections
* then a DDL command on a co-located placement may create a self-deadlock
* if there exist some relationship between the co-located placements
* (e.g. foreign key, partitioning).
*/
Assert(placementConnection != NULL);
ereport(ERROR,
(errcode(ERRCODE_ACTIVE_SQL_TRANSACTION),
errmsg("cannot perform DDL on placement %ld since a co-located "
"placement has been read over multiple connections",
placement->placementId)));
}
else if (foundModifyingConnection)
{
/*
* We already found a connection that performed writes on of the placements
* and must use it.
*/
if ((placementConnection->hadDDL || placementConnection->hadDML) &&
placementConnection->connection != chosenConnection)
{
/*
* The current placement may have been modified over a different
* connection. Neither connection is guaranteed to see all uncomitted
* writes and therefore we cannot proceed.
*/
ereport(ERROR,
(errcode(ERRCODE_ACTIVE_SQL_TRANSACTION),
errmsg("cannot perform query with placements that were "
"modified over multiple connections")));
}
}
else if (CanUseExistingConnection(flags, userName, placementConnection))
{
/*
* There is an existing connection for the placement and we can use it.
*/
Assert(placementConnection != NULL);
chosenConnection = placementConnection->connection;
if (placementConnection->hadDDL || placementConnection->hadDML)
{
/* this connection performed writes, we must use it */
foundModifyingConnection = true;
}
}
else if (placementConnection->hadDDL)
{
/*
* There is an existing connection, but we cannot use it and it executed
* DDL. Any subsequent operation needs to be able to see the results of
* the DDL command and thus cannot proceed if it cannot use the connection.
*/
Assert(placementConnection != NULL);
Assert(!CanUseExistingConnection(flags, userName, placementConnection));
ereport(ERROR,
(errcode(ERRCODE_ACTIVE_SQL_TRANSACTION),
errmsg("cannot establish a new connection for placement %ld, since "
"DDL has been executed on a connection that is in use",
placement->placementId)));
}
else if (placementConnection->hadDML)
{
/*
* There is an existing connection, but we cannot use it and it executed
* DML. Any subsequent operation needs to be able to see the results of
* the DML command and thus cannot proceed if it cannot use the connection.
*
* Note that this is not meaningfully different from the previous case. We
* just produce a different error message based on whether DDL was or only
* DML was executed.
*/
Assert(placementConnection != NULL);
Assert(!CanUseExistingConnection(flags, userName, placementConnection));
Assert(!placementConnection->hadDDL);
ereport(ERROR,
(errcode(ERRCODE_ACTIVE_SQL_TRANSACTION),
errmsg("cannot establish a new connection for placement %ld, since "
"DML has been executed on a connection that is in use",
placement->placementId)));
}
else if (accessType == PLACEMENT_ACCESS_DDL)
{
/*
* There is an existing connection, but we cannot use it and we want to
* execute DDL. The operation on the existing connection might conflict
* with the DDL statement.
*/
Assert(placementConnection != NULL);
Assert(!CanUseExistingConnection(flags, userName, placementConnection));
Assert(!placementConnection->hadDDL);
Assert(!placementConnection->hadDML);
ereport(ERROR,
(errcode(ERRCODE_ACTIVE_SQL_TRANSACTION),
errmsg("cannot perform a parallel DDL command because multiple "
"placements have been accessed over the same connection")));
}
else
{
/*
* The placement has a connection assigned to it, but it cannot be used,
* most likely because it has been claimed exclusively. Fortunately, it
* has only been used for reads and we're not performing a DDL command.
* We can therefore use a different connection for this placement.
*/
Assert(placementConnection != NULL);
Assert(!CanUseExistingConnection(flags, userName, placementConnection));
Assert(!placementConnection->hadDDL);
Assert(!placementConnection->hadDML);
Assert(accessType != PLACEMENT_ACCESS_DDL);
}
placementEntryList = lappend(placementEntryList, placementEntry);
}
if (flags & FOR_DDL)
if (chosenConnection == NULL)
{
placementEntry->modifyingConnection = returnConnectionReference;
returnConnectionReference->hadDDL = true;
/* use the first placement from the list to extract nodename and nodeport */
ShardPlacementAccess *placementAccess =
(ShardPlacementAccess *) linitial(placementAccessList);
ShardPlacement *placement = placementAccess->placement;
/*
* No suitable connection in the placement->connection mapping, get one from
* the node->connection pool.
*/
chosenConnection = StartNodeConnection(flags, placement->nodeName,
placement->nodePort);
}
if (flags & FOR_DML)
/*
* Now that a connection has been chosen, initialise or update the connection
* references for all placements.
*/
forboth(placementAccessCell, placementAccessList,
placementEntryCell, placementEntryList)
{
placementEntry->modifyingConnection = returnConnectionReference;
returnConnectionReference->hadDML = true;
ShardPlacementAccess *placementAccess =
(ShardPlacementAccess *) lfirst(placementAccessCell);
ShardPlacementAccessType accessType = placementAccess->accessType;
ConnectionPlacementHashEntry *placementEntry =
(ConnectionPlacementHashEntry *) lfirst(placementEntryCell);
ConnectionReference *placementConnection = placementEntry->primaryConnection;
if (placementConnection->connection == chosenConnection)
{
/* using the connection that was already assigned to the placement */
}
else if (placementConnection->connection == NULL)
{
/* placement does not have a connection assigned yet */
placementConnection->connection = chosenConnection;
placementConnection->hadDDL = false;
placementConnection->hadDML = false;
placementConnection->userName = MemoryContextStrdup(TopTransactionContext,
userName);
/* record association with connection, to handle connection closure */
dlist_push_tail(&chosenConnection->referencedPlacements,
&placementConnection->connectionNode);
}
else
{
/* using a different connection than the one assigned to the placement */
if (accessType != PLACEMENT_ACCESS_SELECT)
{
/*
* We previously read from the placement, but now we're writing to
* it (if we had written to the placement, we would have either chosen
* the same connection, or errored out). Update the connection reference
* to point to the connection used for writing. We don't need to remember
* the existing connection since we won't be able to reuse it for
* accessing the placement. However, we do register that it exists in
* hasSecondaryConnections.
*/
placementConnection->connection = chosenConnection;
placementConnection->userName = MemoryContextStrdup(TopTransactionContext,
userName);
Assert(!placementConnection->hadDDL);
Assert(!placementConnection->hadDML);
}
/*
* There are now multiple connections that read from the placement
* and DDL commands are forbidden.
*/
placementEntry->hasSecondaryConnections = true;
if (placementEntry->colocatedEntry != NULL)
{
/* we also remember this for co-located placements */
placementEntry->colocatedEntry->hasSecondaryConnections = true;
}
}
/*
* Remember that we used the current connection for writes.
*/
if (accessType == PLACEMENT_ACCESS_DDL)
{
placementConnection->hadDDL = true;
}
if (accessType == PLACEMENT_ACCESS_DML)
{
placementConnection->hadDML = true;
}
}
if (freeUserName)
@ -306,195 +578,82 @@ StartPlacementConnection(uint32 flags, ShardPlacement *placement, const char *us
pfree(freeUserName);
}
return returnConnectionReference->connection;
return chosenConnection;
}
/*
* CheckExistingPlacementConnections check whether any of the existing
* connections is usable. If so, return it, otherwise return NULL.
*
* A connection is usable if it is not in use, the user matches, DDL/DML usage
* match and cached connection are allowed. If no existing connection
* matches, but a new connection would conflict (e.g. a connection already
* exists but isn't usable, and the desired connection needs to execute
* DML/DML) an error is thrown.
* FindOrCreatePlacementEntry finds a placement entry in either the
* placement->connection hash or the co-located placements->connection hash,
* or adds a new entry if the placement has not yet been accessed in the
* current transaction.
*/
static ConnectionReference *
CheckExistingPlacementConnections(uint32 flags,
ConnectionPlacementHashEntry *placementEntry,
const char *userName)
static ConnectionPlacementHashEntry *
FindOrCreatePlacementEntry(ShardPlacement *placement)
{
dlist_iter it;
ConnectionPlacementHashKey key;
ConnectionPlacementHashEntry *placementEntry = NULL;
bool found = false;
/*
* If there's a connection that has executed DML/DDL, always return it if
* possible. That's because we only can execute DML/DDL over that
* connection. Checking this first also allows us to detect conflicts due
* to opening a second connection that wants to execute DML/DDL.
*/
if (placementEntry->modifyingConnection)
key.placementId = placement->placementId;
placementEntry = hash_search(ConnectionPlacementHash, &key, HASH_ENTER, &found);
if (!found)
{
ConnectionReference *connectionReference = placementEntry->modifyingConnection;
/* no connection has been chosen for this placement */
placementEntry->failed = false;
placementEntry->primaryConnection = NULL;
placementEntry->hasSecondaryConnections = false;
placementEntry->colocatedEntry = NULL;
if (CanUseExistingConnection(flags, userName, connectionReference))
if (placement->partitionMethod == DISTRIBUTE_BY_HASH ||
placement->partitionMethod == DISTRIBUTE_BY_NONE)
{
return connectionReference;
}
ColocatedPlacementsHashKey key;
ColocatedPlacementsHashEntry *colocatedEntry = NULL;
if (connectionReference->hadDDL)
{
/* would deadlock otherwise */
ereport(ERROR,
(errcode(ERRCODE_ACTIVE_SQL_TRANSACTION),
errmsg("cannot establish new placement connection when DDL has been "
"executed on existing placement connection")));
}
else if (connectionReference->hadDML)
{
/* we'd not see the other connection's contents */
ereport(ERROR,
(errcode(ERRCODE_ACTIVE_SQL_TRANSACTION),
errmsg("cannot establish new placement connection when DML has been "
"executed on existing placement connection")));
strcpy(key.nodeName, placement->nodeName);
key.nodePort = placement->nodePort;
key.colocationGroupId = placement->colocationGroupId;
key.representativeValue = placement->representativeValue;
/* look for a connection assigned to co-located placements */
colocatedEntry = hash_search(ColocatedPlacementsHash, &key, HASH_ENTER,
&found);
if (!found)
{
void *conRef = MemoryContextAllocZero(TopTransactionContext,
sizeof(ConnectionReference));
/*
* Create a connection reference that can be used for the entire
* set of co-located placements.
*/
colocatedEntry->primaryConnection = (ConnectionReference *) conRef;
colocatedEntry->hasSecondaryConnections = false;
}
/*
* Assign the connection reference for the set of co-located placements
* to the current placement.
*/
placementEntry->primaryConnection = colocatedEntry->primaryConnection;
placementEntry->colocatedEntry = colocatedEntry;
}
else
{
/* modifying xact should have executed DML/DDL */
Assert(false);
void *conRef = MemoryContextAllocZero(TopTransactionContext,
sizeof(ConnectionReference));
placementEntry->primaryConnection = (ConnectionReference *) conRef;
}
}
/*
* Search existing connections for a reusable connection.
*/
dlist_foreach(it, &placementEntry->connectionReferences)
{
ConnectionReference *connectionReference =
dlist_container(ConnectionReference, placementNode, it.cur);
/* record association with shard, for invalidation */
AssociatePlacementWithShard(placementEntry, placement);
if (CanUseExistingConnection(flags, userName, connectionReference))
{
return connectionReference;
}
/*
* If not using the connection, verify that FOR_DML/DDL flags are
* compatible.
*/
if (flags & FOR_DDL)
{
/* would deadlock otherwise */
ereport(ERROR,
(errcode(ERRCODE_ACTIVE_SQL_TRANSACTION),
errmsg("cannot establish new placement connection for DDL when "
"other connections exist")));
}
else if (flags & FOR_DML)
{
/*
* We could allow this case (as presumably the SELECT is done, but
* it'd be a bit prone to programming errors, so we disallow for
* now.
*/
ereport(ERROR,
(errcode(ERRCODE_ACTIVE_SQL_TRANSACTION),
errmsg("cannot establish new placement connection for DML when "
"other connections exist")));
}
}
/* establish a new connection */
return NULL;
}
static ConnectionReference *
CheckExistingColocatedConnections(uint32 flags,
ColocatedPlacementsHashEntry *connectionsEntry,
const char *userName)
{
dlist_iter it;
/*
* If there's a connection that has executed DML/DDL, always return it if
* possible. That's because we only can execute DML/DDL over that
* connection. Checking this first also allows us to detect conflicts due
* to opening a second connection that wants to execute DML/DDL.
*/
if (connectionsEntry->modifyingConnection)
{
ConnectionReference *connectionReference = connectionsEntry->modifyingConnection;
if (CanUseExistingConnection(flags, userName, connectionReference))
{
return connectionReference;
}
if (connectionReference->hadDDL)
{
/* would deadlock otherwise */
ereport(ERROR,
(errcode(ERRCODE_ACTIVE_SQL_TRANSACTION),
errmsg("cannot establish new placement connection when DDL has been "
"executed on existing placement connection for a colocated placement")));
}
else if (connectionReference->hadDML)
{
/* we'd not see the other connection's contents */
ereport(ERROR,
(errcode(ERRCODE_ACTIVE_SQL_TRANSACTION),
errmsg("cannot establish new placement connection when DML has been "
"executed on existing placement connection for a colocated placement")));
}
else
{
/* modifying xact should have executed DML/DDL */
Assert(false);
}
}
/*
* Search existing connections for a reusable connection.
*/
dlist_foreach(it, &connectionsEntry->connectionReferences)
{
ConnectionReference *connectionReference =
dlist_container(ConnectionReference, placementNode, it.cur);
if (CanUseExistingConnection(flags, userName, connectionReference))
{
return connectionReference;
}
/*
* If not using the connection, verify that FOR_DML/DDL flags are
* compatible.
*/
if (flags & FOR_DDL)
{
/* would deadlock otherwise */
ereport(ERROR,
(errcode(ERRCODE_ACTIVE_SQL_TRANSACTION),
errmsg("cannot establish new placement connection for DDL when "
"other connections exist for colocated placements")));
}
else if (flags & FOR_DML)
{
/*
* We could allow this case (as presumably the SELECT is done, but
* it'd be a bit prone to programming errors, so we disallow for
* now.
*/
ereport(ERROR,
(errcode(ERRCODE_ACTIVE_SQL_TRANSACTION),
errmsg("cannot establish new placement connection for DML when "
"other connections exist for colocated placements")));
}
}
/* establish a new connection */
return NULL;
return placementEntry;
}
@ -535,86 +694,6 @@ CanUseExistingConnection(uint32 flags, const char *userName,
}
/*
* StartColocatedPlacementConnection returns a connection that's usable for
* the passed in flags/placement.
*
* It does so by first checking whether any connection is already associated
* with the relevant entry in the colocated placement hash, or by asking for a
* formerly unassociated (possibly new) connection. Errors out if there's
* conflicting connections.
*/
static MultiConnection *
StartColocatedPlacementConnection(uint32 flags, ShardPlacement *placement,
const char *userName)
{
ConnectionReference *connectionReference = NULL;
ColocatedPlacementsHashKey key;
ColocatedPlacementsHashEntry *entry;
bool found;
/*
* Currently only hash partitioned tables can be colocated. For everything
* else we just return a new connection.
*/
if (placement->partitionMethod != DISTRIBUTE_BY_HASH)
{
return StartNodeConnection(flags, placement->nodeName, placement->nodePort);
}
strcpy(key.nodeName, placement->nodeName);
key.nodePort = placement->nodePort;
key.colocationGroupId = placement->colocationGroupId;
key.representativeValue = placement->representativeValue;
entry = hash_search(ColocatedPlacementsHash, &key, HASH_ENTER, &found);
if (!found)
{
memset(((char *) entry) + sizeof(ColocatedPlacementsHashKey),
0,
sizeof(ColocatedPlacementsHashEntry) - sizeof(ColocatedPlacementsHashKey));
dlist_init(&entry->connectionReferences);
}
connectionReference =
CheckExistingColocatedConnections(flags, entry, userName);
if (!connectionReference)
{
MultiConnection *connection = NULL;
connection = StartNodeConnection(flags, placement->nodeName, placement->nodePort);
connectionReference = (ConnectionReference *)
MemoryContextAlloc(TopTransactionContext,
sizeof(ConnectionReference));
connectionReference->connection = connection;
connectionReference->hadDDL = false;
connectionReference->hadDML = false;
connectionReference->userName =
MemoryContextStrdup(TopTransactionContext, userName);
dlist_push_tail(&entry->connectionReferences,
&connectionReference->placementNode);
/* record association with connection, to handle connection closure */
dlist_push_tail(&connection->referencedPlacements,
&connectionReference->connectionNode);
}
if (flags & FOR_DDL)
{
entry->modifyingConnection = connectionReference;
connectionReference->hadDDL = true;
}
if (flags & FOR_DML)
{
entry->modifyingConnection = connectionReference;
connectionReference->hadDML = true;
}
return connectionReference->connection;
}
/*
* AssociatePlacementWithShard records shard->placement relation in
* ConnectionShardHash.
@ -680,7 +759,7 @@ CloseShardPlacementAssociation(struct MultiConnection *connection)
/*
* Note that we don't reset ConnectionPlacementHashEntry's
* modifyingConnection here, that'd more complicated than it seems
* primaryConnection here, that'd more complicated than it seems
* worth. That means we'll error out spuriously if a DML/DDL
* executing connection is closed earlier in a transaction.
*/
@ -823,16 +902,17 @@ CheckShardPlacements(ConnectionShardHashEntry *shardEntry)
{
ConnectionPlacementHashEntry *placementEntry =
dlist_container(ConnectionPlacementHashEntry, shardNode, placementIter.cur);
ConnectionReference *modifyingConnection = placementEntry->modifyingConnection;
ConnectionReference *primaryConnection = placementEntry->primaryConnection;
MultiConnection *connection = NULL;
/* we only consider shards that are modified */
if (modifyingConnection == NULL)
if (primaryConnection == NULL ||
!(primaryConnection->hadDDL || primaryConnection->hadDML))
{
continue;
}
connection = modifyingConnection->connection;
connection = primaryConnection->connection;
if (!connection || connection->remoteTransaction.transactionFailed)
{

166
src/backend/distributed/executor/multi_router_executor.c
View File

@ -75,16 +75,17 @@ bool EnableDeadlockPrevention = true;
/* functions needed during run phase */
static void ReacquireMetadataLocks(List *taskList);
static void AssignInsertTaskShardId(Query *jobQuery, List *taskList);
static ShardPlacementAccess * CreatePlacementAccess(ShardPlacement *placement,
ShardPlacementAccessType accessType);
static void ExecuteSingleModifyTask(CitusScanState *scanState, Task *task,
bool expectResults);
static void ExecuteSingleSelectTask(CitusScanState *scanState, Task *task);
static List * GetModifyConnections(List *taskPlacementList, bool markCritical,
static List * GetModifyConnections(Task *task, bool markCritical,
bool startedInTransaction);
static void ExecuteMultipleTasks(CitusScanState *scanState, List *taskList,
bool isModificationQuery, bool expectResults);
static int64 ExecuteModifyTasks(List *taskList, bool expectResults,
ParamListInfo paramListInfo, CitusScanState *scanState);
static List * TaskShardIntervalList(List *taskList);
static void AcquireExecutorShardLock(Task *task, CmdType commandType);
static void AcquireExecutorMultiShardLocks(List *taskList);
static bool RequiresConsistentSnapshot(Task *task);
@ -601,13 +602,7 @@ ExecuteSingleSelectTask(CitusScanState *scanState, Task *task)
List *taskPlacementList = task->taskPlacementList;
ListCell *taskPlacementCell = NULL;
char *queryString = task->queryString;
if (XactModificationLevel == XACT_MODIFICATION_MULTI_SHARD)
{
ereport(ERROR, (errcode(ERRCODE_ACTIVE_SQL_TRANSACTION),
errmsg("single-shard query may not appear in transaction blocks "
"which contain multi-shard data modifications")));
}
List *relationShardList = task->relationShardList;
/*
* Try to run the query to completion on one placement. If the query fails
@ -620,8 +615,32 @@ ExecuteSingleSelectTask(CitusScanState *scanState, Task *task)
bool dontFailOnError = false;
int64 currentAffectedTupleCount = 0;
int connectionFlags = SESSION_LIFESPAN;
MultiConnection *connection =
GetPlacementConnection(connectionFlags, taskPlacement, NULL);
List *placementAccessList = NIL;
MultiConnection *connection = NULL;
if (list_length(relationShardList) > 0)
{
placementAccessList = BuildPlacementSelectList(taskPlacement->nodeName,
taskPlacement->nodePort,
relationShardList);
}
else
{
/*
* When the SELECT prunes down to 0 shards, just use the dummy placement.
*
* FIXME: it would be preferable to evaluate the SELECT locally since no
* data from the workers is required.
*/
ShardPlacementAccess *placementAccess =
CreatePlacementAccess(taskPlacement, PLACEMENT_ACCESS_SELECT);
placementAccessList = list_make1(placementAccess);
}
connection = GetPlacementListConnection(connectionFlags, placementAccessList,
NULL);
queryOK = SendQueryInSingleRowMode(connection, queryString, paramListInfo);
if (!queryOK)
@ -641,6 +660,56 @@ ExecuteSingleSelectTask(CitusScanState *scanState, Task *task)
}
/*
* BuildPlacementSelectList builds a list of SELECT placement accesses
* which can be used to call StartPlacementListConnection or
* GetPlacementListConnection.
*/
List *
BuildPlacementSelectList(char *nodeName, int nodePort, List *relationShardList)
{
ListCell *relationShardCell = NULL;
List *placementAccessList = NIL;
foreach(relationShardCell, relationShardList)
{
RelationShard *relationShard = (RelationShard *) lfirst(relationShardCell);
ShardPlacement *placement = NULL;
ShardPlacementAccess *placementAccess = NULL;
placement = FindShardPlacementOnNode(nodeName, nodePort, relationShard->shardId);
if (placement == NULL)
{
ereport(ERROR, (errmsg("no active placement of shard %ld found on node "
"%s:%d",
relationShard->shardId, nodeName, nodePort)));
}
placementAccess = CreatePlacementAccess(placement, PLACEMENT_ACCESS_SELECT);
placementAccessList = lappend(placementAccessList, placementAccess);
}
return placementAccessList;
}
/*
* CreatePlacementAccess returns a new ShardPlacementAccess for the given placement
* and access type.
*/
static ShardPlacementAccess *
CreatePlacementAccess(ShardPlacement *placement, ShardPlacementAccessType accessType)
{
ShardPlacementAccess *placementAccess = NULL;
placementAccess = (ShardPlacementAccess *) palloc0(sizeof(ShardPlacementAccess));
placementAccess->placement = placement;
placementAccess->accessType = accessType;
return placementAccess;
}
/*
* ExecuteSingleModifyTask executes the task on the remote node, retrieves the
* results and stores them, if RETURNING is used, in a tuple store.
@ -669,14 +738,6 @@ ExecuteSingleModifyTask(CitusScanState *scanState, Task *task, bool expectResult
bool startedInTransaction =
InCoordinatedTransaction() && XactModificationLevel == XACT_MODIFICATION_DATA;
if (XactModificationLevel == XACT_MODIFICATION_MULTI_SHARD)
{
ereport(ERROR, (errcode(ERRCODE_ACTIVE_SQL_TRANSACTION),
errmsg("single-shard DML commands must not appear in "
"transaction blocks which contain multi-shard data "
"modifications")));
}
/*
* Modifications for reference tables are always done using 2PC. First
* ensure that distributed transaction is started. Then force the
@ -705,7 +766,7 @@ ExecuteSingleModifyTask(CitusScanState *scanState, Task *task, bool expectResult
* establish the connection, mark as critical (when modifying reference
* table) and start a transaction (when in a transaction).
*/
connectionList = GetModifyConnections(taskPlacementList,
connectionList = GetModifyConnections(task,
taskRequiresTwoPhaseCommit,
startedInTransaction);
@ -808,10 +869,12 @@ ExecuteSingleModifyTask(CitusScanState *scanState, Task *task, bool expectResult
* transaction in progress.
*/
static List *
GetModifyConnections(List *taskPlacementList, bool markCritical, bool noNewTransactions)
GetModifyConnections(Task *task, bool markCritical, bool noNewTransactions)
{
List *taskPlacementList = task->taskPlacementList;
ListCell *taskPlacementCell = NULL;
List *multiConnectionList = NIL;
List *relationShardList = task->relationShardList;
/* first initiate connection establishment for all necessary connections */
foreach(taskPlacementCell, taskPlacementList)
@ -819,14 +882,22 @@ GetModifyConnections(List *taskPlacementList, bool markCritical, bool noNewTrans
ShardPlacement *taskPlacement = (ShardPlacement *) lfirst(taskPlacementCell);
int connectionFlags = SESSION_LIFESPAN | FOR_DML;
MultiConnection *multiConnection = NULL;
List *placementAccessList = NIL;
ShardPlacementAccess *placementModification = NULL;
/*
* FIXME: It's not actually correct to use only one shard placement
* here for router queries involving multiple relations. We should
* check that this connection is the only modifying one associated
* with all the involved shards.
*/
multiConnection = StartPlacementConnection(connectionFlags, taskPlacement, NULL);
/* create placement accesses for placements that appear in a subselect */
placementAccessList = BuildPlacementSelectList(taskPlacement->nodeName,
taskPlacement->nodePort,
relationShardList);
/* create placement access for the placement that we're modifying */
placementModification = CreatePlacementAccess(taskPlacement,
PLACEMENT_ACCESS_DML);
placementAccessList = lappend(placementAccessList, placementModification);
/* get an appropriate connection for the DML statement */
multiConnection = GetPlacementListConnection(connectionFlags, placementAccessList,
NULL);
/*
* If already in a transaction, disallow expanding set of remote
@ -943,7 +1014,6 @@ ExecuteModifyTasks(List *taskList, bool expectResults, ParamListInfo paramListIn
ListCell *taskCell = NULL;
Task *firstTask = NULL;
int connectionFlags = 0;
List *shardIntervalList = NIL;
List *affectedTupleCountList = NIL;
HTAB *shardConnectionHash = NULL;
bool tasksPending = true;
@ -954,16 +1024,6 @@ ExecuteModifyTasks(List *taskList, bool expectResults, ParamListInfo paramListIn
return 0;
}
if (XactModificationLevel == XACT_MODIFICATION_DATA)
{
ereport(ERROR, (errcode(ERRCODE_ACTIVE_SQL_TRANSACTION),
errmsg("multi-shard data modifications must not appear in "
"transaction blocks which contain single-shard DML "
"commands")));
}
shardIntervalList = TaskShardIntervalList(taskList);
/* ensure that there are no concurrent modifications on the same shards */
AcquireExecutorMultiShardLocks(taskList);
@ -987,10 +1047,9 @@ ExecuteModifyTasks(List *taskList, bool expectResults, ParamListInfo paramListIn
}
/* open connection to all relevant placements, if not already open */
shardConnectionHash = OpenTransactionsToAllShardPlacements(shardIntervalList,
connectionFlags);
shardConnectionHash = OpenTransactionsForAllTasks(taskList, connectionFlags);
XactModificationLevel = XACT_MODIFICATION_MULTI_SHARD;
XactModificationLevel = XACT_MODIFICATION_DATA;
/* iterate over placements in rounds, to ensure in-order execution */
while (tasksPending)
@ -1126,29 +1185,6 @@ ExecuteModifyTasks(List *taskList, bool expectResults, ParamListInfo paramListIn
}
/*
* TaskShardIntervalList returns a list of shard intervals for a given list of
* tasks.
*/
static List *
TaskShardIntervalList(List *taskList)
{
ListCell *taskCell = NULL;
List *shardIntervalList = NIL;
foreach(taskCell, taskList)
{
Task *task = (Task *) lfirst(taskCell);
int64 shardId = task->anchorShardId;
ShardInterval *shardInterval = LoadShardInterval(shardId);
shardIntervalList = lappend(shardIntervalList, shardInterval);
}
return shardIntervalList;
}
/*
* SendQueryInSingleRowMode sends the given query on the connection in an
* asynchronous way. The function also sets the single-row mode on the

8
src/backend/distributed/executor/multi_utility.c
View File

@ -2494,14 +2494,6 @@ ExecuteDistributedDDLJob(DDLJob *ddlJob)
{
bool shouldSyncMetadata = ShouldSyncTableMetadata(ddlJob->targetRelationId);
if (XactModificationLevel == XACT_MODIFICATION_DATA)
{
ereport(ERROR, (errcode(ERRCODE_ACTIVE_SQL_TRANSACTION),
errmsg("distributed DDL commands must not appear within "
"transaction blocks containing single-shard data "
"modifications")));
}
EnsureCoordinator();
if (!ddlJob->concurrentIndexCmd)

2
src/backend/distributed/master/master_metadata_utility.c
View File

@ -146,7 +146,7 @@ DistributedTableSize(Oid relationId, char *sizeQuery)
ListCell *workerNodeCell = NULL;
uint64 totalRelationSize = 0;
if (XactModificationLevel == XACT_MODIFICATION_MULTI_SHARD)
if (XactModificationLevel == XACT_MODIFICATION_DATA)
{
ereport(ERROR, (errcode(ERRCODE_ACTIVE_SQL_TRANSACTION),
errmsg("citus size functions cannot be called in transaction"

2
src/backend/distributed/master/master_modify_multiple_shards.c
View File

@ -210,7 +210,7 @@ ModifyMultipleShardsTaskList(Query *query, List *shardIntervalList, Oid relation
task = CitusMakeNode(Task);
task->jobId = jobId;
task->taskId = taskId++;
task->taskType = SQL_TASK;
task->taskType = MODIFY_TASK;
task->queryString = shardQueryString->data;
task->dependedTaskList = NULL;
task->replicationModel = REPLICATION_MODEL_INVALID;

57
src/backend/distributed/transaction/multi_shard_transaction.c
View File

@ -17,6 +17,7 @@
#include "distributed/connection_management.h"
#include "distributed/master_metadata_utility.h"
#include "distributed/metadata_cache.h"
#include "distributed/multi_router_executor.h"
#include "distributed/multi_shard_transaction.h"
#include "distributed/placement_connection.h"
#include "distributed/shardinterval_utils.h"
@ -30,25 +31,27 @@
/*
* OpenTransactionsToAllShardPlacements opens connections to all placements
* using the provided shard identifier list and returns it as a shard ID ->
* ShardConnections hash. connectionFlags can be used to specify whether
* the command is FOR_DML or FOR_DDL.
* OpenTransactionsForAllTasks opens a connection for each task,
* taking into account which shards are read and modified by the task
* to select the appopriate connection, or error out if no appropriate
* connection can be found. The set of connections is returned as an
* anchor shard ID -> ShardConnections hash.
*/
HTAB *
OpenTransactionsToAllShardPlacements(List *shardIntervalList, int connectionFlags)
OpenTransactionsForAllTasks(List *taskList, int connectionFlags)
{
HTAB *shardConnectionHash = NULL;
ListCell *shardIntervalCell = NULL;
ListCell *taskCell = NULL;
List *newConnectionList = NIL;
shardConnectionHash = CreateShardConnectionHash(CurrentMemoryContext);
/* open connections to shards which don't have connections yet */
foreach(shardIntervalCell, shardIntervalList)
foreach(taskCell, taskList)
{
ShardInterval *shardInterval = (ShardInterval *) lfirst(shardIntervalCell);
uint64 shardId = shardInterval->shardId;
Task *task = (Task *) lfirst(taskCell);
ShardPlacementAccessType accessType = PLACEMENT_ACCESS_SELECT;
uint64 shardId = task->anchorShardId;
ShardConnections *shardConnections = NULL;
bool shardConnectionsFound = false;
List *shardPlacementList = NIL;
@ -69,9 +72,24 @@ OpenTransactionsToAllShardPlacements(List *shardIntervalList, int connectionFlag
UINT64_FORMAT, shardId)));
}
if (task->taskType == MODIFY_TASK)
{
accessType = PLACEMENT_ACCESS_DML;
}
else
{
/* can only open connections for DDL and DML commands */
Assert(task->taskType == DDL_TASK);
accessType = PLACEMENT_ACCESS_DDL;
}
foreach(placementCell, shardPlacementList)
{
ShardPlacement *shardPlacement = (ShardPlacement *) lfirst(placementCell);
ShardPlacementAccess placementModification;
List *placementAccessList = NIL;
List *placementSelectList = NIL;
MultiConnection *connection = NULL;
WorkerNode *workerNode = FindWorkerNode(shardPlacement->nodeName,
@ -83,9 +101,24 @@ OpenTransactionsToAllShardPlacements(List *shardIntervalList, int connectionFlag
shardPlacement->nodePort)));
}
connection = StartPlacementConnection(connectionFlags,
shardPlacement,
NULL);
/* add placement access for modification */
placementModification.placement = shardPlacement;
placementModification.accessType = accessType;
placementAccessList = lappend(placementAccessList, &placementModification);
/* add additional placement accesses for subselects (e.g. INSERT .. SELECT) */
placementSelectList = BuildPlacementSelectList(shardPlacement->nodeName,
shardPlacement->nodePort,
task->relationShardList);
placementAccessList = list_concat(placementAccessList, placementSelectList);
/*
* Find a connection that sees preceding writes and cannot self-deadlock,
* or error out if no such connection exists.
*/
connection = StartPlacementListConnection(connectionFlags,
placementAccessList, NULL);
ClaimConnectionExclusively(connection);

37
src/backend/distributed/utils/metadata_cache.c
View File

@ -356,6 +356,43 @@ LoadShardPlacement(uint64 shardId, uint64 placementId)
}
/*
* FindShardPlacementOnNode returns the shard placement for the given shard
* on the given node, or returns NULL of no placement for the shard exists
* on the node.
*/
ShardPlacement *
FindShardPlacementOnNode(char *nodeName, int nodePort, uint64 shardId)
{
ShardCacheEntry *shardEntry = NULL;
DistTableCacheEntry *tableEntry = NULL;
ShardPlacement *placementArray = NULL;
int numberOfPlacements = 0;
ShardPlacement *placementOnNode = NULL;
int placementIndex = 0;
shardEntry = LookupShardCacheEntry(shardId);
tableEntry = shardEntry->tableEntry;
placementArray = tableEntry->arrayOfPlacementArrays[shardEntry->shardIndex];
numberOfPlacements = tableEntry->arrayOfPlacementArrayLengths[shardEntry->shardIndex];
for (placementIndex = 0; placementIndex < numberOfPlacements; placementIndex++)
{
ShardPlacement *placement = &placementArray[placementIndex];
if (strncmp(nodeName, placement->nodeName, WORKER_LENGTH) == 0 &&
nodePort == placement->nodePort)
{
placementOnNode = CitusMakeNode(ShardPlacement);
CopyShardPlacement(placement, placementOnNode);
break;
}
}
return placementOnNode;
}
/*
* ShardPlacementList returns the list of placements for the given shard from
* the cache.

2
src/include/distributed/metadata_cache.h
View File

@ -70,6 +70,8 @@ extern bool IsDistributedTable(Oid relationId);
extern List * DistributedTableList(void);
extern ShardInterval * LoadShardInterval(uint64 shardId);
extern ShardPlacement * LoadShardPlacement(uint64 shardId, uint64 placementId);
extern ShardPlacement * FindShardPlacementOnNode(char *nodeName, int nodePort,
uint64 shardId);
extern DistTableCacheEntry * DistributedTableCacheEntry(Oid distributedRelationId);
extern int GetLocalGroupId(void);
extern List * DistTableOidList(void);

2
src/include/distributed/multi_physical_planner.h
View File

@ -186,7 +186,7 @@ typedef struct Task
char replicationModel; /* only applies to modify tasks */
bool insertSelectQuery;
List *relationShardList; /* only applies INSERT/SELECT tasks */
List *relationShardList;
} Task;

2
src/include/distributed/multi_router_executor.h
View File

@ -43,5 +43,7 @@ extern TupleTableSlot * RouterMultiModifyExecScan(CustomScanState *node);
extern int64 ExecuteModifyTasksWithoutResults(List *taskList);
extern void ExecuteTasksSequentiallyWithoutResults(List *taskList);
extern List * BuildPlacementSelectList(char *nodeName, int nodePort,
List *relationShardList);
#endif /* MULTI_ROUTER_EXECUTOR_H_ */

3
src/include/distributed/multi_shard_transaction.h
View File

@ -27,8 +27,7 @@ typedef struct ShardConnections
} ShardConnections;
extern HTAB * OpenTransactionsToAllShardPlacements(List *shardIdList,
int connectionFlags);
extern HTAB * OpenTransactionsForAllTasks(List *taskList, int connectionFlags);
extern HTAB * CreateShardConnectionHash(MemoryContext memoryContext);
extern ShardConnections * GetShardHashConnections(HTAB *connectionHash, int64 shardId,
bool *connectionsFound);

31
src/include/distributed/placement_connection.h
View File

@ -15,6 +15,30 @@
/* forward declare, to avoid dependency on ShardPlacement definition */
struct ShardPlacement;
/* represents the way in which a placement is accessed */
typedef enum ShardPlacementAccessType
{
/* read from placement */
PLACEMENT_ACCESS_SELECT,
/* modify rows in placement */
PLACEMENT_ACCESS_DML,
/* modify placement schema */
PLACEMENT_ACCESS_DDL
} ShardPlacementAccessType;
/* represents access to a placement */
typedef struct ShardPlacementAccess
{
/* placement that is accessed */
struct ShardPlacement *placement;
/* the way in which the placement is accessed */
ShardPlacementAccessType accessType;
} ShardPlacementAccess;
extern MultiConnection * GetPlacementConnection(uint32 flags,
struct ShardPlacement *placement,
const char *userName);
@ -22,6 +46,13 @@ extern MultiConnection * StartPlacementConnection(uint32 flags,
struct ShardPlacement *placement,
const char *userName);
extern MultiConnection * GetPlacementListConnection(uint32 flags,
List *placementAccessList,
const char *userName);
extern MultiConnection * StartPlacementListConnection(uint32 flags,
List *placementAccessList,
const char *userName);
extern void ResetPlacementConnectionManagement(void);
extern void MarkFailedShardPlacements(void);
extern void PostCommitMarkFailedShardPlacements(bool using2PC);

6
src/test/regress/expected/multi_alter_table_add_constraints.out
View File

@ -441,9 +441,9 @@ SELECT create_distributed_table('products', 'product_no');
BEGIN;
INSERT INTO products VALUES(1,'product_1', 5);
-- Should error out since conflicts with the above single-shard data modification command.
-- DDL may error out after an INSERT because it might pick the wrong connection
ALTER TABLE products ADD CONSTRAINT unn_pno UNIQUE(product_no);
ERROR: distributed DDL commands must not appear within transaction blocks containing single-shard data modifications
ERROR: cannot establish a new connection for placement 2327, since DML has been executed on a connection that is in use
ROLLBACK;
BEGIN;
-- Add constraints
@ -451,9 +451,7 @@ ALTER TABLE products ADD CONSTRAINT unn_pno UNIQUE(product_no);
ALTER TABLE products ADD CONSTRAINT check_price CHECK(price > discounted_price);
ALTER TABLE products ALTER COLUMN product_no SET NOT NULL;
ALTER TABLE products ADD CONSTRAINT p_key_product PRIMARY KEY(product_no);
-- Single shard DML command can't be located in the same transaction with above commands.
INSERT INTO products VALUES(1,'product_1', 10, 8);
ERROR: single-shard DML commands must not appear in transaction blocks which contain multi-shard data modifications
ROLLBACK;
-- There should be no constraint on master and worker(s)
SELECT "Constraint", "Definition" FROM table_checks WHERE relid='products'::regclass;

20
src/test/regress/expected/multi_insert_select.out
View File

@ -1636,15 +1636,24 @@ ROLLBACK;
BEGIN;
ALTER TABLE raw_events_second DROP COLUMN value_4;
INSERT INTO raw_events_first SELECT * FROM raw_events_second WHERE user_id = 100;
ERROR: single-shard DML commands must not appear in transaction blocks which contain multi-shard data modifications
ROLLBACK;
-- Insert after copy is currently disallowed because of the way the
-- transaction modification state is currently handled. Copy is also
-- rolled back.
-- Altering a reference table and then performing an INSERT ... SELECT which
-- joins with the reference table is not allowed, since the INSERT ... SELECT
-- would read from the reference table over others connections than the ones
-- that performed the DDL.
BEGIN;
ALTER TABLE reference_table ADD COLUMN z int;
INSERT INTO raw_events_first (user_id)
SELECT user_id FROM raw_events_second JOIN reference_table USING (user_id);
ERROR: cannot establish a new connection for placement 655, since DDL has been executed on a connection that is in use
ROLLBACK;
-- Insert after copy is disallowed when the INSERT INTO ... SELECT chooses
-- to use a connection for one shard, while the connection already modified
-- another shard.
BEGIN;
COPY raw_events_second (user_id, value_1) FROM STDIN DELIMITER ',';
INSERT INTO raw_events_first SELECT * FROM raw_events_second;
ERROR: multi-shard data modifications must not appear in transaction blocks which contain single-shard DML commands
ERROR: cannot establish a new connection for placement 636, since DML has been executed on a connection that is in use
ROLLBACK;
-- Insert after copy is currently allowed for single-shard operation.
-- Both insert and copy are rolled back successfully.
@ -1658,7 +1667,6 @@ SELECT user_id FROM raw_events_first WHERE user_id = 101;
(1 row)
ROLLBACK;
-- Copy after insert is currently disallowed.
BEGIN;
INSERT INTO raw_events_first SELECT * FROM raw_events_second;
COPY raw_events_first (user_id, value_1) FROM STDIN DELIMITER ',';

238
src/test/regress/expected/multi_modifying_xacts.out
View File

@ -172,18 +172,16 @@ SELECT count(*) FROM researchers WHERE lab_id = 6;
(1 row)
ABORT;
-- applies to DDL, too
-- we can mix DDL and INSERT
BEGIN;
INSERT INTO labs VALUES (6, 'Bell Labs');
ALTER TABLE labs ADD COLUMN motto text;
ERROR: distributed DDL commands must not appear within transaction blocks containing single-shard data modifications
COMMIT;
ABORT;
-- whether it occurs first or second
BEGIN;
ALTER TABLE labs ADD COLUMN motto text;
INSERT INTO labs VALUES (6, 'Bell Labs');
ERROR: single-shard DML commands must not appear in transaction blocks which contain multi-shard data modifications
COMMIT;
ABORT;
-- but the DDL should correctly roll back
SELECT "Column", "Type", "Modifiers" FROM table_desc WHERE relid='public.labs'::regclass;
Column | Type | Modifiers
@ -207,7 +205,7 @@ BEGIN;
INSERT INTO researchers VALUES (2, 1, 'Knuth Donald');
INSERT INTO researchers VALUES (10, 6, 'Lamport Leslie');
\copy researchers from stdin delimiter ','
ERROR: cannot establish new placement connection when DML has been executed on existing placement connection
ERROR: cannot establish a new connection for placement 2704, since DML has been executed on a connection that is in use
CONTEXT: COPY researchers, line 2: "10,6,Lesport Lampie"
ROLLBACK;
-- after a COPY you can modify multiple shards, since they'll use different connections
@ -383,16 +381,29 @@ ORDER BY nodeport;
localhost | 57638 | t | DROP FUNCTION
(2 rows)
-- ALTER TABLE and COPY are compatible if ALTER TABLE precedes COPY
-- ALTER and copy are compatible
BEGIN;
ALTER TABLE labs ADD COLUMN motto text;
\copy labs from stdin delimiter ','
ROLLBACK;
-- but not if COPY precedes ALTER TABLE
BEGIN;
\copy labs from stdin delimiter ','
ALTER TABLE labs ADD COLUMN motto text;
ERROR: distributed DDL commands must not appear within transaction blocks containing single-shard data modifications
ABORT;
-- cannot perform DDL once a connection is used for multiple shards
BEGIN;
SELECT lab_id FROM researchers WHERE lab_id = 1 AND id = 0;
lab_id
--------
(0 rows)
SELECT lab_id FROM researchers WHERE lab_id = 2 AND id = 0;
lab_id
--------
(0 rows)
ALTER TABLE researchers ADD COLUMN motto text;
ERROR: cannot perform a parallel DDL command because multiple placements have been accessed over the same connection
ROLLBACK;
-- multi-shard operations can co-exist with DDL in a transactional way
BEGIN;
@ -1283,3 +1294,212 @@ SELECT * FROM run_command_on_workers('DROP USER test_user');
(2 rows)
DROP USER test_user;
-- set up foreign keys to test transactions with co-located and reference tables
BEGIN;
SET LOCAL citus.shard_replication_factor TO 1;
SET LOCAL citus.shard_count TO 4;
CREATE TABLE usergroups (
gid int PRIMARY KEY,
name text
);
SELECT create_reference_table('usergroups');
create_reference_table
------------------------
(1 row)
CREATE TABLE itemgroups (
gid int PRIMARY KEY,
name text
);
SELECT create_reference_table('itemgroups');
create_reference_table
------------------------
(1 row)
CREATE TABLE users (
id int PRIMARY KEY,
name text,
user_group int
);
SELECT create_distributed_table('users', 'id');
create_distributed_table
--------------------------
(1 row)
CREATE TABLE items (
user_id int REFERENCES users (id) ON DELETE CASCADE,
item_name text,
item_group int
);
SELECT create_distributed_table('items', 'user_id');
create_distributed_table
--------------------------
(1 row)
-- Table to find values that live in different shards on the same node
SELECT id, shard_name('users', shardid), nodename, nodeport
FROM
pg_dist_shard_placement
JOIN
( SELECT id, get_shard_id_for_distribution_column('users', id) shardid FROM generate_series(1,10) id ) ids
USING (shardid)
ORDER BY
id;
id | shard_name | nodename | nodeport
----+----------------------+-----------+----------
1 | public.users_1200022 | localhost | 57637
2 | public.users_1200025 | localhost | 57638
3 | public.users_1200023 | localhost | 57638
4 | public.users_1200023 | localhost | 57638
5 | public.users_1200022 | localhost | 57637
6 | public.users_1200024 | localhost | 57637
7 | public.users_1200023 | localhost | 57638
8 | public.users_1200022 | localhost | 57637
9 | public.users_1200025 | localhost | 57638
10 | public.users_1200022 | localhost | 57637
(10 rows)
END;
-- the INSERTs into items should see the users
BEGIN;
\COPY users FROM STDIN WITH CSV
INSERT INTO items VALUES (1, 'item-1');
INSERT INTO items VALUES (6, 'item-6');
END;
SELECT user_id FROM items ORDER BY user_id;
user_id
---------
1
6
(2 rows)
-- should not be able to open multiple connections per node after INSERTing over one connection
BEGIN;
INSERT INTO users VALUES (2, 'burak');
INSERT INTO users VALUES (3, 'burak');
\COPY items FROM STDIN WITH CSV
ERROR: cannot establish a new connection for placement 2743, since DML has been executed on a connection that is in use
END;
-- cannot perform DDL after a co-located table has been read over 1 connection
BEGIN;
SELECT id FROM users WHERE id = 1;
id
----
1
(1 row)
SELECT id FROM users WHERE id = 6;
id
----
6
(1 row)
ALTER TABLE items ADD COLUMN last_update timestamptz;
NOTICE: using one-phase commit for distributed DDL commands
ERROR: cannot perform a parallel DDL command because multiple placements have been accessed over the same connection
END;
-- but the other way around is fine
BEGIN;
ALTER TABLE items ADD COLUMN last_update timestamptz;
SELECT id FROM users JOIN items ON (id = user_id) WHERE id = 1;
id
----
1
(1 row)
SELECT id FROM users JOIN items ON (id = user_id) WHERE id = 6;
id
----
6
(1 row)
END;
BEGIN;
-- establish multiple connections to a node
\COPY users FROM STDIN WITH CSV
-- now read from the reference table over each connection
SELECT user_id FROM items JOIN itemgroups ON (item_group = gid) WHERE user_id = 2;
user_id
---------
(0 rows)
SELECT user_id FROM items JOIN itemgroups ON (item_group = gid) WHERE user_id = 3;
user_id
---------
(0 rows)
-- perform a DDL command on the reference table
ALTER TABLE itemgroups ADD COLUMN last_update timestamptz;
ERROR: cannot perform DDL on placement 2737, which has been read over multiple connections
END;
BEGIN;
-- establish multiple connections to a node
\COPY users FROM STDIN WITH CSV
-- read from the reference table over each connection
SELECT user_id FROM items JOIN itemgroups ON (item_group = gid) WHERE user_id = 2;
user_id
---------
(0 rows)
SELECT user_id FROM items JOIN itemgroups ON (item_group = gid) WHERE user_id = 3;
user_id
---------
(0 rows)
-- perform a DDL command on a co-located reference table
ALTER TABLE usergroups ADD COLUMN last_update timestamptz;
ERROR: cannot perform DDL on placement 2735 since a co-located placement has been read over multiple connections
END;
BEGIN;
-- make a modification over connection 1
INSERT INTO usergroups VALUES (0,'istanbul');
-- copy over connections 1 and 2
\COPY users FROM STDIN WITH CSV
-- cannot read modifications made over different connections
SELECT id FROM users JOIN usergroups ON (gid = user_group) WHERE id = 3;
ERROR: cannot perform query with placements that were modified over multiple connections
END;
-- make sure we can see cascading deletes
BEGIN;
SELECT master_modify_multiple_shards('DELETE FROM users');
master_modify_multiple_shards
-------------------------------
2
(1 row)
SELECT user_id FROM items JOIN itemgroups ON (item_group = gid) WHERE user_id = 1;
user_id
---------
(0 rows)
SELECT user_id FROM items JOIN itemgroups ON (item_group = gid) WHERE user_id = 6;
user_id
---------
(0 rows)
END;
-- test visibility after COPY
INSERT INTO usergroups VALUES (2,'group');
BEGIN;
-- opens two separate connections to node
\COPY users FROM STDIN WITH CSV
-- Uses first connection, which wrote the row with id = 2
SELECT * FROM users JOIN usergroups ON (user_group = gid) WHERE id = 2;
id | name | user_group | gid | name
----+-------+------------+-----+-------
2 | onder | 2 | 2 | group
(1 row)
-- Should use second connection, which wrote the row with id = 4
SELECT * FROM users JOIN usergroups ON (user_group = gid) WHERE id = 4;
id | name | user_group | gid | name
----+-------+------------+-----+-------
4 | murat | 2 | 2 | group
(1 row)
END;
DROP TABLE items, users, itemgroups, usergroups;

6
src/test/regress/expected/multi_mx_modifying_xacts.out
View File

@ -179,12 +179,6 @@ SELECT count(*) FROM researchers_mx WHERE lab_id = 6;
(1 row)
ABORT;
-- applies to DDL
BEGIN;
INSERT INTO labs_mx VALUES (6, 'Bell labs_mx');
ALTER TABLE labs_mx ADD COLUMN motto text;
ERROR: distributed DDL commands must not appear within transaction blocks containing single-shard data modifications
COMMIT;
-- doesn't apply to COPY after modifications
BEGIN;
INSERT INTO labs_mx VALUES (6, 'Bell labs_mx');

11
src/test/regress/expected/multi_reference_table.out
View File

@ -1609,19 +1609,22 @@ SELECT * FROM reference_table_test;
10 | 2 | 2 | Fri Dec 02 00:00:00 2016
(1 row)
-- do not allow mixing transactions
-- DML+master_modify_multiple_shards is allowed
BEGIN;
INSERT INTO reference_table_test VALUES (2, 2.0, '2', '2016-12-02');
SELECT master_modify_multiple_shards('DELETE FROM colocated_table_test');
ERROR: multi-shard data modifications must not appear in transaction blocks which contain single-shard DML commands
master_modify_multiple_shards
-------------------------------
10
(1 row)
ROLLBACK;
-- Do not allow DDL and modification in the same transaction
-- DDL+DML is allowed
BEGIN;
ALTER TABLE reference_table_test ADD COLUMN value_dummy INT;
NOTICE: using one-phase commit for distributed DDL commands
HINT: You can enable two-phase commit for extra safety with: SET citus.multi_shard_commit_protocol TO '2pc'
INSERT INTO reference_table_test VALUES (2, 2.0, '2', '2016-12-02');
ERROR: single-shard DML commands must not appear in transaction blocks which contain multi-shard data modifications
ROLLBACK;
-- clean up tables
DROP TABLE reference_table_test, reference_table_test_second, reference_table_test_third,

3
src/test/regress/sql/multi_alter_table_add_constraints.sql
View File

@ -382,7 +382,7 @@ SELECT create_distributed_table('products', 'product_no');
BEGIN;
INSERT INTO products VALUES(1,'product_1', 5);
-- Should error out since conflicts with the above single-shard data modification command.
-- DDL may error out after an INSERT because it might pick the wrong connection
ALTER TABLE products ADD CONSTRAINT unn_pno UNIQUE(product_no);
ROLLBACK;
@ -393,7 +393,6 @@ ALTER TABLE products ADD CONSTRAINT check_price CHECK(price > discounted_price);
ALTER TABLE products ALTER COLUMN product_no SET NOT NULL;
ALTER TABLE products ADD CONSTRAINT p_key_product PRIMARY KEY(product_no);
-- Single shard DML command can't be located in the same transaction with above commands.
INSERT INTO products VALUES(1,'product_1', 10, 8);
ROLLBACK;

17
src/test/regress/sql/multi_insert_select.sql
View File

@ -1332,9 +1332,19 @@ ALTER TABLE raw_events_second DROP COLUMN value_4;
INSERT INTO raw_events_first SELECT * FROM raw_events_second WHERE user_id = 100;
ROLLBACK;
-- Insert after copy is currently disallowed because of the way the
-- transaction modification state is currently handled. Copy is also
-- rolled back.
-- Altering a reference table and then performing an INSERT ... SELECT which
-- joins with the reference table is not allowed, since the INSERT ... SELECT
-- would read from the reference table over others connections than the ones
-- that performed the DDL.
BEGIN;
ALTER TABLE reference_table ADD COLUMN z int;
INSERT INTO raw_events_first (user_id)
SELECT user_id FROM raw_events_second JOIN reference_table USING (user_id);
ROLLBACK;
-- Insert after copy is disallowed when the INSERT INTO ... SELECT chooses
-- to use a connection for one shard, while the connection already modified
-- another shard.
BEGIN;
COPY raw_events_second (user_id, value_1) FROM STDIN DELIMITER ',';
100,100
@ -1352,7 +1362,6 @@ INSERT INTO raw_events_first SELECT * FROM raw_events_second WHERE user_id = 101
SELECT user_id FROM raw_events_first WHERE user_id = 101;
ROLLBACK;
-- Copy after insert is currently disallowed.
BEGIN;
INSERT INTO raw_events_first SELECT * FROM raw_events_second;
COPY raw_events_first (user_id, value_1) FROM STDIN DELIMITER ',';

160
src/test/regress/sql/multi_modifying_xacts.sql
View File

@ -138,17 +138,17 @@ INSERT INTO labs VALUES (6, 'Bell Labs');
SELECT count(*) FROM researchers WHERE lab_id = 6;
ABORT;
-- applies to DDL, too
-- we can mix DDL and INSERT
BEGIN;
INSERT INTO labs VALUES (6, 'Bell Labs');
ALTER TABLE labs ADD COLUMN motto text;
COMMIT;
ABORT;
-- whether it occurs first or second
BEGIN;
ALTER TABLE labs ADD COLUMN motto text;
INSERT INTO labs VALUES (6, 'Bell Labs');
COMMIT;
ABORT;
-- but the DDL should correctly roll back
SELECT "Column", "Type", "Modifiers" FROM table_desc WHERE relid='public.labs'::regclass;
@ -289,7 +289,7 @@ ORDER BY nodeport, shardid;
SELECT * FROM run_command_on_workers('drop function reject_large_id()')
ORDER BY nodeport;
-- ALTER TABLE and COPY are compatible if ALTER TABLE precedes COPY
-- ALTER and copy are compatible
BEGIN;
ALTER TABLE labs ADD COLUMN motto text;
\copy labs from stdin delimiter ','
@ -297,12 +297,18 @@ ALTER TABLE labs ADD COLUMN motto text;
\.
ROLLBACK;
-- but not if COPY precedes ALTER TABLE
BEGIN;
\copy labs from stdin delimiter ','
12,fsociety
\.
ALTER TABLE labs ADD COLUMN motto text;
ABORT;
-- cannot perform DDL once a connection is used for multiple shards
BEGIN;
SELECT lab_id FROM researchers WHERE lab_id = 1 AND id = 0;
SELECT lab_id FROM researchers WHERE lab_id = 2 AND id = 0;
ALTER TABLE researchers ADD COLUMN motto text;
ROLLBACK;
-- multi-shard operations can co-exist with DDL in a transactional way
@ -944,3 +950,147 @@ DROP TABLE reference_modifying_xacts, hash_modifying_xacts, hash_modifying_xacts
SELECT * FROM run_command_on_workers('DROP USER test_user');
DROP USER test_user;
-- set up foreign keys to test transactions with co-located and reference tables
BEGIN;
SET LOCAL citus.shard_replication_factor TO 1;
SET LOCAL citus.shard_count TO 4;
CREATE TABLE usergroups (
gid int PRIMARY KEY,
name text
);
SELECT create_reference_table('usergroups');
CREATE TABLE itemgroups (
gid int PRIMARY KEY,
name text
);
SELECT create_reference_table('itemgroups');
CREATE TABLE users (
id int PRIMARY KEY,
name text,
user_group int
);
SELECT create_distributed_table('users', 'id');
CREATE TABLE items (
user_id int REFERENCES users (id) ON DELETE CASCADE,
item_name text,
item_group int
);
SELECT create_distributed_table('items', 'user_id');
-- Table to find values that live in different shards on the same node
SELECT id, shard_name('users', shardid), nodename, nodeport
FROM
pg_dist_shard_placement
JOIN
( SELECT id, get_shard_id_for_distribution_column('users', id) shardid FROM generate_series(1,10) id ) ids
USING (shardid)
ORDER BY
id;
END;
-- the INSERTs into items should see the users
BEGIN;
\COPY users FROM STDIN WITH CSV
1,brian,0
6,metin,0
\.
INSERT INTO items VALUES (1, 'item-1');
INSERT INTO items VALUES (6, 'item-6');
END;
SELECT user_id FROM items ORDER BY user_id;
-- should not be able to open multiple connections per node after INSERTing over one connection
BEGIN;
INSERT INTO users VALUES (2, 'burak');
INSERT INTO users VALUES (3, 'burak');
\COPY items FROM STDIN WITH CSV
2,item-2,0
3,item-3,0
\.
END;
-- cannot perform DDL after a co-located table has been read over 1 connection
BEGIN;
SELECT id FROM users WHERE id = 1;
SELECT id FROM users WHERE id = 6;
ALTER TABLE items ADD COLUMN last_update timestamptz;
END;
-- but the other way around is fine
BEGIN;
ALTER TABLE items ADD COLUMN last_update timestamptz;
SELECT id FROM users JOIN items ON (id = user_id) WHERE id = 1;
SELECT id FROM users JOIN items ON (id = user_id) WHERE id = 6;
END;
BEGIN;
-- establish multiple connections to a node
\COPY users FROM STDIN WITH CSV
2,burak,0
3,burak,0
\.
-- now read from the reference table over each connection
SELECT user_id FROM items JOIN itemgroups ON (item_group = gid) WHERE user_id = 2;
SELECT user_id FROM items JOIN itemgroups ON (item_group = gid) WHERE user_id = 3;
-- perform a DDL command on the reference table
ALTER TABLE itemgroups ADD COLUMN last_update timestamptz;
END;
BEGIN;
-- establish multiple connections to a node
\COPY users FROM STDIN WITH CSV
2,burak,0
3,burak,0
\.
-- read from the reference table over each connection
SELECT user_id FROM items JOIN itemgroups ON (item_group = gid) WHERE user_id = 2;
SELECT user_id FROM items JOIN itemgroups ON (item_group = gid) WHERE user_id = 3;
-- perform a DDL command on a co-located reference table
ALTER TABLE usergroups ADD COLUMN last_update timestamptz;
END;
BEGIN;
-- make a modification over connection 1
INSERT INTO usergroups VALUES (0,'istanbul');
-- copy over connections 1 and 2
\COPY users FROM STDIN WITH CSV
2,burak,0
3,burak,0
\.
-- cannot read modifications made over different connections
SELECT id FROM users JOIN usergroups ON (gid = user_group) WHERE id = 3;
END;
-- make sure we can see cascading deletes
BEGIN;
SELECT master_modify_multiple_shards('DELETE FROM users');
SELECT user_id FROM items JOIN itemgroups ON (item_group = gid) WHERE user_id = 1;
SELECT user_id FROM items JOIN itemgroups ON (item_group = gid) WHERE user_id = 6;
END;
-- test visibility after COPY
INSERT INTO usergroups VALUES (2,'group');
BEGIN;
-- opens two separate connections to node
\COPY users FROM STDIN WITH CSV
2,onder,2
4,murat,2
\.
-- Uses first connection, which wrote the row with id = 2
SELECT * FROM users JOIN usergroups ON (user_group = gid) WHERE id = 2;
-- Should use second connection, which wrote the row with id = 4
SELECT * FROM users JOIN usergroups ON (user_group = gid) WHERE id = 4;
END;
DROP TABLE items, users, itemgroups, usergroups;

6
src/test/regress/sql/multi_mx_modifying_xacts.sql
View File

@ -150,12 +150,6 @@ INSERT INTO labs_mx VALUES (6, 'Bell labs_mx');
SELECT count(*) FROM researchers_mx WHERE lab_id = 6;
ABORT;
-- applies to DDL
BEGIN;
INSERT INTO labs_mx VALUES (6, 'Bell labs_mx');
ALTER TABLE labs_mx ADD COLUMN motto text;
COMMIT;
-- doesn't apply to COPY after modifications
BEGIN;
INSERT INTO labs_mx VALUES (6, 'Bell labs_mx');

4
src/test/regress/sql/multi_reference_table.sql
View File

@ -994,13 +994,13 @@ UPDATE reference_table_test SET value_1 = 10 WHERE value_1 = 2;
COMMIT;
SELECT * FROM reference_table_test;
-- do not allow mixing transactions
-- DML+master_modify_multiple_shards is allowed
BEGIN;
INSERT INTO reference_table_test VALUES (2, 2.0, '2', '2016-12-02');
SELECT master_modify_multiple_shards('DELETE FROM colocated_table_test');
ROLLBACK;
-- Do not allow DDL and modification in the same transaction
-- DDL+DML is allowed
BEGIN;
ALTER TABLE reference_table_test ADD COLUMN value_dummy INT;
INSERT INTO reference_table_test VALUES (2, 2.0, '2', '2016-12-02');