From 59c2961b980e496a11bb24ab05b7f613e46d8f58 Mon Sep 17 00:00:00 2001 From: Zhijie Hou Date: Mon, 11 Aug 2025 16:23:24 +0800 Subject: [PATCH v60 2/2] Resume retaining the information for conflict detection The patch allows the launcher to re-initialized invalidated slot, if at least one apply worker has confirmed that the retention duration is now within the max_conflict_retention_duration. --- doc/src/sgml/ref/create_subscription.sgml | 7 +- src/backend/replication/logical/launcher.c | 83 +++++++++++++---- src/backend/replication/logical/worker.c | 100 ++++++++++++++++----- src/include/replication/worker_internal.h | 7 ++ 4 files changed, 158 insertions(+), 39 deletions(-) diff --git a/doc/src/sgml/ref/create_subscription.sgml b/doc/src/sgml/ref/create_subscription.sgml index 6121c86561e..0e34d136672 100644 --- a/doc/src/sgml/ref/create_subscription.sgml +++ b/doc/src/sgml/ref/create_subscription.sgml @@ -539,9 +539,12 @@ CREATE SUBSCRIPTION subscription_nameretain_dead_tuples is enabled, confirm that the retention duration exceeded the max_conflict_retention_duration set within the - corresponding subscription. To re-enable retention, you can disable - retain_dead_tuples for all subscriptions and + corresponding subscription. To re-enable retention manually, you can + disable retain_dead_tuples for all subscriptions and re-enable it after confirming this replication slot has been dropped. + Alternatively, the retention will be automatically resumed once at + least one apply worker confirms that the retention duration is within + the specified limit. Note that Overall retention will not stop if other subscriptions diff --git a/src/backend/replication/logical/launcher.c b/src/backend/replication/logical/launcher.c index f1dfb51ccf8..b85acd95e6b 100644 --- a/src/backend/replication/logical/launcher.c +++ b/src/backend/replication/logical/launcher.c @@ -102,9 +102,12 @@ static int logicalrep_pa_worker_count(Oid subid); static void logicalrep_launcher_attach_dshmem(void); static void ApplyLauncherSetWorkerStartTime(Oid subid, TimestampTz start_time); static TimestampTz ApplyLauncherGetWorkerStartTime(Oid subid); -static void compute_min_nonremovable_xid(LogicalRepWorker *worker, TransactionId *xmin); +static void compute_min_nonremovable_xid(LogicalRepWorker *worker, + bool can_advance_xmin, + TransactionId *xmin); static bool acquire_conflict_slot_if_exists(void); static void update_conflict_slot_xmin(TransactionId new_xmin); +static void init_conflict_slot_xmin(void); /* @@ -464,6 +467,8 @@ retry: worker->stream_fileset = NULL; worker->leader_pid = is_parallel_apply_worker ? MyProcPid : InvalidPid; worker->parallel_apply = is_parallel_apply_worker; + worker->stop_conflict_info_retention = (retain_dead_tuples && + !TransactionIdIsValid(MyReplicationSlot->data.xmin)); worker->oldest_nonremovable_xid = retain_dead_tuples ? MyReplicationSlot->data.xmin : InvalidTransactionId; @@ -1258,8 +1263,8 @@ ApplyLauncherMain(Datum main_arg) * required for conflict detection among all running apply * workers that enables retain_dead_tuples. */ - if (sub->retaindeadtuples && can_advance_xmin) - compute_min_nonremovable_xid(w, &xmin); + if (sub->retaindeadtuples) + compute_min_nonremovable_xid(w, can_advance_xmin, &xmin); /* worker is running already */ continue; @@ -1369,9 +1374,11 @@ ApplyLauncherMain(Datum main_arg) * in *xmin. */ static void -compute_min_nonremovable_xid(LogicalRepWorker *worker, TransactionId *xmin) +compute_min_nonremovable_xid(LogicalRepWorker *worker, bool can_advance_xmin, + TransactionId *xmin) { TransactionId nonremovable_xid; + bool stop_retention; Assert(worker != NULL); @@ -1383,13 +1390,43 @@ compute_min_nonremovable_xid(LogicalRepWorker *worker, TransactionId *xmin) SpinLockAcquire(&worker->mutex); nonremovable_xid = worker->oldest_nonremovable_xid; + stop_retention = worker->stop_conflict_info_retention; SpinLockRelease(&worker->mutex); /* * Skip collecting oldest_nonremovable_xid for workers that have stopped * conflict retention. */ + if (stop_retention) + return; + + /* + * Initialize slot.xmin as a apply worker resumes retention of information + * critical for conflict detection. + */ + if (!TransactionIdIsValid(MyReplicationSlot->data.xmin)) + init_conflict_slot_xmin(); + + /* + * Assign slot.xmin to the apply worker's oldest_nonremovable_xid if the + * latter is invalid. This ensures the apply worker continues to maintain + * the oldest_nonremovable_xid (see get_candidate_xid). + */ if (!TransactionIdIsValid(nonremovable_xid)) + { + nonremovable_xid = MyReplicationSlot->data.xmin; + + SpinLockAcquire(&worker->mutex); + worker->oldest_nonremovable_xid = nonremovable_xid; + SpinLockRelease(&worker->mutex); + + /* Notify the apply worker to start the next cycle of management */ + LWLockAcquire(LogicalRepWorkerLock, LW_SHARED); + logicalrep_worker_wakeup_ptr(worker); + LWLockRelease(LogicalRepWorkerLock); + } + + if (!can_advance_xmin) return; if (!TransactionIdIsValid(*xmin) || @@ -1451,23 +1488,15 @@ update_conflict_slot_xmin(TransactionId new_xmin) } /* - * Create and acquire the replication slot used to retain information for - * conflict detection, if not yet. + * Initialize the xmin for the conflict detection slot. */ -void -CreateConflictDetectionSlot(void) +static void +init_conflict_slot_xmin(void) { TransactionId xmin_horizon; - /* Exit early, if the replication slot is already created and acquired */ - if (MyReplicationSlot) - return; - - ereport(LOG, - errmsg("creating replication conflict detection slot")); - - ReplicationSlotCreate(CONFLICT_DETECTION_SLOT, false, RS_PERSISTENT, false, - false, false); + Assert(MyReplicationSlot && + !TransactionIdIsValid(MyReplicationSlot->data.xmin)); LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); @@ -1487,6 +1516,26 @@ CreateConflictDetectionSlot(void) ReplicationSlotSave(); } +/* + * Create and acquire the replication slot used to retain information for + * conflict detection, if not yet. + */ +void +CreateConflictDetectionSlot(void) +{ + /* Exit early, if the replication slot is already created and acquired */ + if (MyReplicationSlot) + return; + + ereport(LOG, + errmsg("creating replication conflict detection slot")); + + ReplicationSlotCreate(CONFLICT_DETECTION_SLOT, false, RS_PERSISTENT, false, + false, false); + + init_conflict_slot_xmin(); +} + /* * Is current process the logical replication launcher? */ diff --git a/src/backend/replication/logical/worker.c b/src/backend/replication/logical/worker.c index 5269cc980f2..246938f2011 100644 --- a/src/backend/replication/logical/worker.c +++ b/src/backend/replication/logical/worker.c @@ -418,6 +418,10 @@ typedef struct RetainDeadTuplesData long table_sync_wait_time; /* time spent waiting for table sync * to finish */ + bool wait_for_initial_xid; /* wait for the launcher to initialize + * the apply worker's + * oldest_nonremovable_xid */ + /* * The following fields are used to determine the timing for the next * round of transaction ID advancement. @@ -4376,10 +4380,6 @@ can_advance_nonremovable_xid(RetainDeadTuplesData *rdt_data) if (!MySubscription->retaindeadtuples) return false; - /* No need to advance if we have already stopped retaining */ - if (!TransactionIdIsValid(MyLogicalRepWorker->oldest_nonremovable_xid)) - return false; - return true; } @@ -4417,6 +4417,33 @@ get_candidate_xid(RetainDeadTuplesData *rdt_data) TransactionId oldest_running_xid; TimestampTz now; + /* + * No need to advance if the apply worker has resumed retention but the + * launcher has not yet initialized slot.xmin and assigned it to + * oldest_nonremovable_xid. + * + * It might seem feasible to directly check the conflict detection + * slot.xmin instead of relying on the launcher to assign the worker's + * oldest_nonremovable_xid; however, that could lead to a race condition + * where slot.xmin is set to InvalidTransactionId immediately after the + * check. In such cases, oldest_nonremovable_xid would no longer be + * protected by a replication slot and could become unreliable if a + * wraparound occurs. + */ + if (rdt_data->wait_for_initial_xid) + { + TransactionId nonremovable_xid; + + SpinLockAcquire(&MyLogicalRepWorker->mutex); + nonremovable_xid = MyLogicalRepWorker->oldest_nonremovable_xid; + SpinLockRelease(&MyLogicalRepWorker->mutex); + + if (!TransactionIdIsValid(nonremovable_xid)) + return; + + rdt_data->wait_for_initial_xid = false; + } + /* * Use last_recv_time when applying changes in the loop to avoid * unnecessary system time retrieval. If last_recv_time is not available, @@ -4661,13 +4688,38 @@ wait_for_local_flush(RetainDeadTuplesData *rdt_data) if (last_flushpos < rdt_data->remote_lsn) return; + /* + * If conflict info retention was previously stopped due to a timeout, and + * the time required to advance the non-removable transaction ID has now + * decreased to within acceptable limits, log a message. The next step is + * to wait for the launcher to initialize the oldest_nonremovable_xid. + */ + if (MyLogicalRepWorker->stop_conflict_info_retention) + { + ereport(LOG, + errmsg("logical replication worker for subscription \"%s\" resumes retaining the information for detecting conflicts", + MySubscription->name), + errdetail("The time spent applying changes up to LSN %X/%X is now within the maximum limit of %u ms.", + LSN_FORMAT_ARGS(rdt_data->remote_lsn), + MySubscription->maxconflretention)); + + rdt_data->wait_for_initial_xid = true; + } + /* * Reaching here means the remote WAL position has been received, and all * transactions up to that position on the publisher have been applied and * flushed locally. So, we can advance the non-removable transaction ID. + * + * However, if oldest_nonremovable_xid is invalid, indicating that + * retention was stopped and is now being resumed, refrain from updating + * oldest_nonremovable_xid until the launcher provides an initial value + * (see get_candidate_xid() for details). */ SpinLockAcquire(&MyLogicalRepWorker->mutex); - MyLogicalRepWorker->oldest_nonremovable_xid = rdt_data->candidate_xid; + if (TransactionIdIsValid(MyLogicalRepWorker->oldest_nonremovable_xid)) + MyLogicalRepWorker->oldest_nonremovable_xid = rdt_data->candidate_xid; + MyLogicalRepWorker->stop_conflict_info_retention = false; SpinLockRelease(&MyLogicalRepWorker->mutex); elog(DEBUG2, "confirmed flush up to remote lsn %X/%X: new oldest_nonremovable_xid %u", @@ -4710,9 +4762,8 @@ reset_retention_data_fields(RetainDeadTuplesData *rdt_data) * to InvalidTransactionId, notify the launcher to set the slot.xmin to * InvalidTransactionId as well, and return true. Return false otherwise. * - * Currently, the retention will not resume automatically unless user manually - * disables retain_dead_tuples and re-enables it after confirming that the - * replication slot has been dropped. + * The retention will resume automatically if the worker has confirmed that the + * retention duration is now within the max_conflict_retention_duration. */ static bool should_stop_conflict_info_retention(RetainDeadTuplesData *rdt_data) @@ -4743,18 +4794,26 @@ should_stop_conflict_info_retention(RetainDeadTuplesData *rdt_data) rdt_data->table_sync_wait_time)) return false; - ereport(LOG, - errmsg("logical replication worker for subscription \"%s\" will stop retaining conflict information", - MySubscription->name), - errdetail("The time spent advancing the non-removable transaction ID has exceeded the maximum limit of %u ms.", - MySubscription->maxconflretention)); - - SpinLockAcquire(&MyLogicalRepWorker->mutex); - MyLogicalRepWorker->oldest_nonremovable_xid = InvalidTransactionId; - SpinLockRelease(&MyLogicalRepWorker->mutex); - - /* Notify launcher to update the conflict slot */ - ApplyLauncherWakeup(); + /* + * Log a message and reset relevant data when the worker is about to stop + * retaining conflict information. + */ + if (!MyLogicalRepWorker->stop_conflict_info_retention) + { + ereport(LOG, + errmsg("logical replication worker for subscription \"%s\" will stop retaining the information for detecting conflicts", + MySubscription->name), + errdetail("The time spent advancing the non-removable transaction ID has exceeded the maximum limit of %u ms.", + MySubscription->maxconflretention)); + + SpinLockAcquire(&MyLogicalRepWorker->mutex); + MyLogicalRepWorker->oldest_nonremovable_xid = InvalidTransactionId; + MyLogicalRepWorker->stop_conflict_info_retention = true; + SpinLockRelease(&MyLogicalRepWorker->mutex); + + /* Notify launcher to update the conflict slot */ + ApplyLauncherWakeup(); + } reset_retention_data_fields(rdt_data); @@ -5609,6 +5668,7 @@ InitializeLogRepWorker(void) */ if (am_leader_apply_worker() && MySubscription->retaindeadtuples && + !MyLogicalRepWorker->stop_conflict_info_retention && !TransactionIdIsValid(MyLogicalRepWorker->oldest_nonremovable_xid)) { ereport(LOG, diff --git a/src/include/replication/worker_internal.h b/src/include/replication/worker_internal.h index 9c0c2b8050c..32a1cbb6528 100644 --- a/src/include/replication/worker_internal.h +++ b/src/include/replication/worker_internal.h @@ -105,6 +105,13 @@ typedef struct LogicalRepWorker */ TransactionId oldest_nonremovable_xid; + /* + * Indicates whether the apply worker has stopped retaining information + * useful for conflict detection. This is used only when + * retain_dead_tuples is enabled. + */ + bool stop_conflict_info_retention; + /* Stats. */ XLogRecPtr last_lsn; TimestampTz last_send_time; -- 2.50.1.windows.1