Line data Source code
1 : /*-------------------------------------------------------------------------
2 : * tablesync.c
3 : * PostgreSQL logical replication: initial table data synchronization
4 : *
5 : * Copyright (c) 2012-2026, PostgreSQL Global Development Group
6 : *
7 : * IDENTIFICATION
8 : * src/backend/replication/logical/tablesync.c
9 : *
10 : * NOTES
11 : * This file contains code for initial table data synchronization for
12 : * logical replication.
13 : *
14 : * The initial data synchronization is done separately for each table,
15 : * in a separate apply worker that only fetches the initial snapshot data
16 : * from the publisher and then synchronizes the position in the stream with
17 : * the leader apply worker.
18 : *
19 : * There are several reasons for doing the synchronization this way:
20 : * - It allows us to parallelize the initial data synchronization
21 : * which lowers the time needed for it to happen.
22 : * - The initial synchronization does not have to hold the xid and LSN
23 : * for the time it takes to copy data of all tables, causing less
24 : * bloat and lower disk consumption compared to doing the
25 : * synchronization in a single process for the whole database.
26 : * - It allows us to synchronize any tables added after the initial
27 : * synchronization has finished.
28 : *
29 : * The stream position synchronization works in multiple steps:
30 : * - Apply worker requests a tablesync worker to start, setting the new
31 : * table state to INIT.
32 : * - Tablesync worker starts; changes table state from INIT to DATASYNC while
33 : * copying.
34 : * - Tablesync worker does initial table copy; there is a FINISHEDCOPY (sync
35 : * worker specific) state to indicate when the copy phase has completed, so
36 : * if the worker crashes with this (non-memory) state then the copy will not
37 : * be re-attempted.
38 : * - Tablesync worker then sets table state to SYNCWAIT; waits for state change.
39 : * - Apply worker periodically checks for tables in SYNCWAIT state. When
40 : * any appear, it sets the table state to CATCHUP and starts loop-waiting
41 : * until either the table state is set to SYNCDONE or the sync worker
42 : * exits.
43 : * - After the sync worker has seen the state change to CATCHUP, it will
44 : * read the stream and apply changes (acting like an apply worker) until
45 : * it catches up to the specified stream position. Then it sets the
46 : * state to SYNCDONE. There might be zero changes applied between
47 : * CATCHUP and SYNCDONE, because the sync worker might be ahead of the
48 : * apply worker.
49 : * - Once the state is set to SYNCDONE, the apply will continue tracking
50 : * the table until it reaches the SYNCDONE stream position, at which
51 : * point it sets state to READY and stops tracking. Again, there might
52 : * be zero changes in between.
53 : *
54 : * So the state progression is always: INIT -> DATASYNC -> FINISHEDCOPY
55 : * -> SYNCWAIT -> CATCHUP -> SYNCDONE -> READY.
56 : *
57 : * The catalog pg_subscription_rel is used to keep information about
58 : * subscribed tables and their state. The catalog holds all states
59 : * except SYNCWAIT and CATCHUP which are only in shared memory.
60 : *
61 : * Example flows look like this:
62 : * - Apply is in front:
63 : * sync:8
64 : * -> set in catalog FINISHEDCOPY
65 : * -> set in memory SYNCWAIT
66 : * apply:10
67 : * -> set in memory CATCHUP
68 : * -> enter wait-loop
69 : * sync:10
70 : * -> set in catalog SYNCDONE
71 : * -> exit
72 : * apply:10
73 : * -> exit wait-loop
74 : * -> continue rep
75 : * apply:11
76 : * -> set in catalog READY
77 : *
78 : * - Sync is in front:
79 : * sync:10
80 : * -> set in catalog FINISHEDCOPY
81 : * -> set in memory SYNCWAIT
82 : * apply:8
83 : * -> set in memory CATCHUP
84 : * -> continue per-table filtering
85 : * sync:10
86 : * -> set in catalog SYNCDONE
87 : * -> exit
88 : * apply:10
89 : * -> set in catalog READY
90 : * -> stop per-table filtering
91 : * -> continue rep
92 : *-------------------------------------------------------------------------
93 : */
94 :
95 : #include "postgres.h"
96 :
97 : #include "access/table.h"
98 : #include "access/xact.h"
99 : #include "catalog/indexing.h"
100 : #include "catalog/pg_subscription_rel.h"
101 : #include "catalog/pg_type.h"
102 : #include "commands/copy.h"
103 : #include "miscadmin.h"
104 : #include "nodes/makefuncs.h"
105 : #include "parser/parse_relation.h"
106 : #include "pgstat.h"
107 : #include "replication/logicallauncher.h"
108 : #include "replication/logicalrelation.h"
109 : #include "replication/logicalworker.h"
110 : #include "replication/origin.h"
111 : #include "replication/slot.h"
112 : #include "replication/walreceiver.h"
113 : #include "replication/worker_internal.h"
114 : #include "storage/ipc.h"
115 : #include "storage/latch.h"
116 : #include "storage/lmgr.h"
117 : #include "utils/acl.h"
118 : #include "utils/array.h"
119 : #include "utils/builtins.h"
120 : #include "utils/lsyscache.h"
121 : #include "utils/rls.h"
122 : #include "utils/snapmgr.h"
123 : #include "utils/syscache.h"
124 : #include "utils/usercontext.h"
125 : #include "utils/wait_event.h"
126 :
127 : List *table_states_not_ready = NIL;
128 :
129 : static StringInfo copybuf = NULL;
130 :
131 : /*
132 : * Wait until the relation sync state is set in the catalog to the expected
133 : * one; return true when it happens.
134 : *
135 : * Returns false if the table sync worker or the table itself have
136 : * disappeared, or the table state has been reset.
137 : *
138 : * Currently, this is used in the apply worker when transitioning from
139 : * CATCHUP state to SYNCDONE.
140 : */
141 : static bool
142 192 : wait_for_table_state_change(Oid relid, char expected_state)
143 : {
144 : char state;
145 :
146 : for (;;)
147 223 : {
148 : LogicalRepWorker *worker;
149 : XLogRecPtr statelsn;
150 :
151 415 : CHECK_FOR_INTERRUPTS();
152 :
153 415 : InvalidateCatalogSnapshot();
154 415 : state = GetSubscriptionRelState(MyLogicalRepWorker->subid,
155 : relid, &statelsn);
156 :
157 415 : if (state == SUBREL_STATE_UNKNOWN)
158 0 : break;
159 :
160 415 : if (state == expected_state)
161 192 : return true;
162 :
163 : /* Check if the sync worker is still running and bail if not. */
164 223 : LWLockAcquire(LogicalRepWorkerLock, LW_SHARED);
165 223 : worker = logicalrep_worker_find(WORKERTYPE_TABLESYNC,
166 223 : MyLogicalRepWorker->subid, relid,
167 : false);
168 223 : LWLockRelease(LogicalRepWorkerLock);
169 223 : if (!worker)
170 0 : break;
171 :
172 223 : (void) WaitLatch(MyLatch,
173 : WL_LATCH_SET | WL_TIMEOUT | WL_EXIT_ON_PM_DEATH,
174 : 1000L, WAIT_EVENT_LOGICAL_SYNC_STATE_CHANGE);
175 :
176 223 : ResetLatch(MyLatch);
177 : }
178 :
179 0 : return false;
180 : }
181 :
182 : /*
183 : * Wait until the apply worker changes the state of our synchronization
184 : * worker to the expected one.
185 : *
186 : * Used when transitioning from SYNCWAIT state to CATCHUP.
187 : *
188 : * Returns false if the apply worker has disappeared.
189 : */
190 : static bool
191 194 : wait_for_worker_state_change(char expected_state)
192 : {
193 : int rc;
194 :
195 : for (;;)
196 194 : {
197 : LogicalRepWorker *worker;
198 :
199 388 : CHECK_FOR_INTERRUPTS();
200 :
201 : /*
202 : * Done if already in correct state. (We assume this fetch is atomic
203 : * enough to not give a misleading answer if we do it with no lock.)
204 : */
205 388 : if (MyLogicalRepWorker->relstate == expected_state)
206 194 : return true;
207 :
208 : /*
209 : * Bail out if the apply worker has died, else signal it we're
210 : * waiting.
211 : */
212 194 : LWLockAcquire(LogicalRepWorkerLock, LW_SHARED);
213 194 : worker = logicalrep_worker_find(WORKERTYPE_APPLY,
214 194 : MyLogicalRepWorker->subid, InvalidOid,
215 : false);
216 194 : if (worker && worker->proc)
217 194 : logicalrep_worker_wakeup_ptr(worker);
218 194 : LWLockRelease(LogicalRepWorkerLock);
219 194 : if (!worker)
220 0 : break;
221 :
222 : /*
223 : * Wait. We expect to get a latch signal back from the apply worker,
224 : * but use a timeout in case it dies without sending one.
225 : */
226 194 : rc = WaitLatch(MyLatch,
227 : WL_LATCH_SET | WL_TIMEOUT | WL_EXIT_ON_PM_DEATH,
228 : 1000L, WAIT_EVENT_LOGICAL_SYNC_STATE_CHANGE);
229 :
230 194 : if (rc & WL_LATCH_SET)
231 194 : ResetLatch(MyLatch);
232 : }
233 :
234 0 : return false;
235 : }
236 :
237 : /*
238 : * Handle table synchronization cooperation from the synchronization
239 : * worker.
240 : *
241 : * If the sync worker is in CATCHUP state and reached (or passed) the
242 : * predetermined synchronization point in the WAL stream, mark the table as
243 : * SYNCDONE and finish.
244 : */
245 : void
246 228 : ProcessSyncingTablesForSync(XLogRecPtr current_lsn)
247 : {
248 228 : SpinLockAcquire(&MyLogicalRepWorker->relmutex);
249 :
250 228 : if (MyLogicalRepWorker->relstate == SUBREL_STATE_CATCHUP &&
251 228 : current_lsn >= MyLogicalRepWorker->relstate_lsn)
252 : {
253 : TimeLineID tli;
254 194 : char syncslotname[NAMEDATALEN] = {0};
255 194 : char originname[NAMEDATALEN] = {0};
256 :
257 194 : MyLogicalRepWorker->relstate = SUBREL_STATE_SYNCDONE;
258 194 : MyLogicalRepWorker->relstate_lsn = current_lsn;
259 :
260 194 : SpinLockRelease(&MyLogicalRepWorker->relmutex);
261 :
262 : /*
263 : * UpdateSubscriptionRelState must be called within a transaction.
264 : */
265 194 : if (!IsTransactionState())
266 194 : StartTransactionCommand();
267 :
268 194 : UpdateSubscriptionRelState(MyLogicalRepWorker->subid,
269 194 : MyLogicalRepWorker->relid,
270 194 : MyLogicalRepWorker->relstate,
271 194 : MyLogicalRepWorker->relstate_lsn,
272 : false);
273 :
274 : /*
275 : * End streaming so that LogRepWorkerWalRcvConn can be used to drop
276 : * the slot.
277 : */
278 194 : walrcv_endstreaming(LogRepWorkerWalRcvConn, &tli);
279 :
280 : /*
281 : * Cleanup the tablesync slot.
282 : *
283 : * This has to be done after updating the state because otherwise if
284 : * there is an error while doing the database operations we won't be
285 : * able to rollback dropped slot.
286 : */
287 194 : ReplicationSlotNameForTablesync(MyLogicalRepWorker->subid,
288 194 : MyLogicalRepWorker->relid,
289 : syncslotname,
290 : sizeof(syncslotname));
291 :
292 : /*
293 : * It is important to give an error if we are unable to drop the slot,
294 : * otherwise, it won't be dropped till the corresponding subscription
295 : * is dropped. So passing missing_ok = false.
296 : */
297 194 : ReplicationSlotDropAtPubNode(LogRepWorkerWalRcvConn, syncslotname, false);
298 :
299 194 : CommitTransactionCommand();
300 194 : pgstat_report_stat(false);
301 :
302 : /*
303 : * Start a new transaction to clean up the tablesync origin tracking.
304 : * This transaction will be ended within the FinishSyncWorker(). Now,
305 : * even, if we fail to remove this here, the apply worker will ensure
306 : * to clean it up afterward.
307 : *
308 : * We need to do this after the table state is set to SYNCDONE.
309 : * Otherwise, if an error occurs while performing the database
310 : * operation, the worker will be restarted and the in-memory state of
311 : * replication progress (remote_lsn) won't be rolled-back which would
312 : * have been cleared before restart. So, the restarted worker will use
313 : * invalid replication progress state resulting in replay of
314 : * transactions that have already been applied.
315 : */
316 194 : StartTransactionCommand();
317 :
318 194 : ReplicationOriginNameForLogicalRep(MyLogicalRepWorker->subid,
319 194 : MyLogicalRepWorker->relid,
320 : originname,
321 : sizeof(originname));
322 :
323 : /*
324 : * Resetting the origin session removes the ownership of the slot.
325 : * This is needed to allow the origin to be dropped.
326 : */
327 194 : replorigin_session_reset();
328 194 : replorigin_xact_clear(true);
329 :
330 : /*
331 : * Drop the tablesync's origin tracking if exists.
332 : *
333 : * There is a chance that the user is concurrently performing refresh
334 : * for the subscription where we remove the table state and its origin
335 : * or the apply worker would have removed this origin. So passing
336 : * missing_ok = true.
337 : */
338 194 : replorigin_drop_by_name(originname, true, false);
339 :
340 194 : FinishSyncWorker();
341 : }
342 : else
343 34 : SpinLockRelease(&MyLogicalRepWorker->relmutex);
344 34 : }
345 :
346 : /*
347 : * Handle table synchronization cooperation from the apply worker.
348 : *
349 : * Walk over all subscription tables that are individually tracked by the
350 : * apply process (currently, all that have state other than
351 : * SUBREL_STATE_READY) and manage synchronization for them.
352 : *
353 : * If there are tables that need synchronizing and are not being synchronized
354 : * yet, start sync workers for them (if there are free slots for sync
355 : * workers). To prevent starting the sync worker for the same relation at a
356 : * high frequency after a failure, we store its last start time with each sync
357 : * state info. We start the sync worker for the same relation after waiting
358 : * at least wal_retrieve_retry_interval.
359 : *
360 : * For tables that are being synchronized already, check if sync workers
361 : * either need action from the apply worker or have finished. This is the
362 : * SYNCWAIT to CATCHUP transition.
363 : *
364 : * If the synchronization position is reached (SYNCDONE), then the table can
365 : * be marked as READY and is no longer tracked.
366 : */
367 : void
368 4336 : ProcessSyncingTablesForApply(XLogRecPtr current_lsn)
369 : {
370 : struct tablesync_start_time_mapping
371 : {
372 : Oid relid;
373 : TimestampTz last_start_time;
374 : };
375 : static HTAB *last_start_times = NULL;
376 : ListCell *lc;
377 : bool started_tx;
378 4336 : bool should_exit = false;
379 4336 : Relation rel = NULL;
380 :
381 : Assert(!IsTransactionState());
382 :
383 : /* We need up-to-date sync state info for subscription tables here. */
384 4336 : FetchRelationStates(NULL, NULL, &started_tx);
385 :
386 : /*
387 : * Prepare a hash table for tracking last start times of workers, to avoid
388 : * immediate restarts. We don't need it if there are no tables that need
389 : * syncing.
390 : */
391 4336 : if (table_states_not_ready != NIL && !last_start_times)
392 126 : {
393 : HASHCTL ctl;
394 :
395 126 : ctl.keysize = sizeof(Oid);
396 126 : ctl.entrysize = sizeof(struct tablesync_start_time_mapping);
397 126 : last_start_times = hash_create("Logical replication table sync worker start times",
398 : 256, &ctl, HASH_ELEM | HASH_BLOBS);
399 : }
400 :
401 : /*
402 : * Clean up the hash table when we're done with all tables (just to
403 : * release the bit of memory).
404 : */
405 4210 : else if (table_states_not_ready == NIL && last_start_times)
406 : {
407 98 : hash_destroy(last_start_times);
408 98 : last_start_times = NULL;
409 : }
410 :
411 : /*
412 : * Process all tables that are being synchronized.
413 : */
414 6101 : foreach(lc, table_states_not_ready)
415 : {
416 1765 : SubscriptionRelState *rstate = (SubscriptionRelState *) lfirst(lc);
417 :
418 1765 : if (!started_tx)
419 : {
420 325 : StartTransactionCommand();
421 325 : started_tx = true;
422 : }
423 :
424 : Assert(get_rel_relkind(rstate->relid) != RELKIND_SEQUENCE);
425 :
426 1765 : if (rstate->state == SUBREL_STATE_SYNCDONE)
427 : {
428 : /*
429 : * Apply has caught up to the position where the table sync has
430 : * finished. Mark the table as ready so that the apply will just
431 : * continue to replicate it normally.
432 : */
433 191 : if (current_lsn >= rstate->lsn)
434 : {
435 : char originname[NAMEDATALEN];
436 :
437 191 : rstate->state = SUBREL_STATE_READY;
438 191 : rstate->lsn = current_lsn;
439 :
440 : /*
441 : * Remove the tablesync origin tracking if exists.
442 : *
443 : * There is a chance that the user is concurrently performing
444 : * refresh for the subscription where we remove the table
445 : * state and its origin or the tablesync worker would have
446 : * already removed this origin. We can't rely on tablesync
447 : * worker to remove the origin tracking as if there is any
448 : * error while dropping we won't restart it to drop the
449 : * origin. So passing missing_ok = true.
450 : *
451 : * Lock the subscription and origin in the same order as we
452 : * are doing during DDL commands to avoid deadlocks. See
453 : * AlterSubscription_refresh.
454 : */
455 191 : LockSharedObject(SubscriptionRelationId, MyLogicalRepWorker->subid,
456 : 0, AccessShareLock);
457 :
458 191 : if (!rel)
459 185 : rel = table_open(SubscriptionRelRelationId, RowExclusiveLock);
460 :
461 191 : ReplicationOriginNameForLogicalRep(MyLogicalRepWorker->subid,
462 : rstate->relid,
463 : originname,
464 : sizeof(originname));
465 191 : replorigin_drop_by_name(originname, true, false);
466 :
467 : /*
468 : * Update the state to READY only after the origin cleanup.
469 : */
470 191 : UpdateSubscriptionRelState(MyLogicalRepWorker->subid,
471 191 : rstate->relid, rstate->state,
472 : rstate->lsn, true);
473 : }
474 : }
475 : else
476 : {
477 : LogicalRepWorker *syncworker;
478 :
479 : /*
480 : * Look for a sync worker for this relation.
481 : */
482 1574 : LWLockAcquire(LogicalRepWorkerLock, LW_SHARED);
483 :
484 1574 : syncworker = logicalrep_worker_find(WORKERTYPE_TABLESYNC,
485 1574 : MyLogicalRepWorker->subid,
486 : rstate->relid, false);
487 :
488 1574 : if (syncworker)
489 : {
490 : /* Found one, update our copy of its state */
491 716 : SpinLockAcquire(&syncworker->relmutex);
492 716 : rstate->state = syncworker->relstate;
493 716 : rstate->lsn = syncworker->relstate_lsn;
494 716 : if (rstate->state == SUBREL_STATE_SYNCWAIT)
495 : {
496 : /*
497 : * Sync worker is waiting for apply. Tell sync worker it
498 : * can catchup now.
499 : */
500 192 : syncworker->relstate = SUBREL_STATE_CATCHUP;
501 192 : syncworker->relstate_lsn =
502 192 : Max(syncworker->relstate_lsn, current_lsn);
503 : }
504 716 : SpinLockRelease(&syncworker->relmutex);
505 :
506 : /* If we told worker to catch up, wait for it. */
507 716 : if (rstate->state == SUBREL_STATE_SYNCWAIT)
508 : {
509 : /* Signal the sync worker, as it may be waiting for us. */
510 192 : if (syncworker->proc)
511 192 : logicalrep_worker_wakeup_ptr(syncworker);
512 :
513 : /* Now safe to release the LWLock */
514 192 : LWLockRelease(LogicalRepWorkerLock);
515 :
516 192 : if (started_tx)
517 : {
518 : /*
519 : * We must commit the existing transaction to release
520 : * the existing locks before entering a busy loop.
521 : * This is required to avoid any undetected deadlocks
522 : * due to any existing lock as deadlock detector won't
523 : * be able to detect the waits on the latch.
524 : *
525 : * Also close any tables prior to the commit.
526 : */
527 192 : if (rel)
528 : {
529 29 : table_close(rel, NoLock);
530 29 : rel = NULL;
531 : }
532 192 : CommitTransactionCommand();
533 192 : pgstat_report_stat(false);
534 : }
535 :
536 : /*
537 : * Enter busy loop and wait for synchronization worker to
538 : * reach expected state (or die trying).
539 : */
540 192 : StartTransactionCommand();
541 192 : started_tx = true;
542 :
543 192 : wait_for_table_state_change(rstate->relid,
544 : SUBREL_STATE_SYNCDONE);
545 : }
546 : else
547 524 : LWLockRelease(LogicalRepWorkerLock);
548 : }
549 : else
550 : {
551 : /*
552 : * If there is no sync worker for this table yet, count
553 : * running sync workers for this subscription, while we have
554 : * the lock.
555 : */
556 : int nsyncworkers =
557 858 : logicalrep_sync_worker_count(MyLogicalRepWorker->subid);
558 : struct tablesync_start_time_mapping *hentry;
559 : bool found;
560 :
561 : /* Now safe to release the LWLock */
562 858 : LWLockRelease(LogicalRepWorkerLock);
563 :
564 858 : hentry = hash_search(last_start_times, &rstate->relid,
565 : HASH_ENTER, &found);
566 858 : if (!found)
567 199 : hentry->last_start_time = 0;
568 :
569 858 : launch_sync_worker(WORKERTYPE_TABLESYNC, nsyncworkers,
570 : rstate->relid, &hentry->last_start_time);
571 : }
572 : }
573 : }
574 :
575 : /* Close table if opened */
576 4336 : if (rel)
577 156 : table_close(rel, NoLock);
578 :
579 :
580 4336 : if (started_tx)
581 : {
582 : /*
583 : * Even when the two_phase mode is requested by the user, it remains
584 : * as 'pending' until all tablesyncs have reached READY state.
585 : *
586 : * When this happens, we restart the apply worker and (if the
587 : * conditions are still ok) then the two_phase tri-state will become
588 : * 'enabled' at that time.
589 : *
590 : * Note: If the subscription has no tables then leave the state as
591 : * PENDING, which allows ALTER SUBSCRIPTION ... REFRESH PUBLICATION to
592 : * work.
593 : */
594 1106 : if (MySubscription->twophasestate == LOGICALREP_TWOPHASE_STATE_PENDING)
595 : {
596 40 : CommandCounterIncrement(); /* make updates visible */
597 40 : if (AllTablesyncsReady())
598 : {
599 6 : ereport(LOG,
600 : (errmsg("logical replication apply worker for subscription \"%s\" will restart so that two_phase can be enabled",
601 : MySubscription->name)));
602 6 : should_exit = true;
603 : }
604 : }
605 :
606 1106 : CommitTransactionCommand();
607 1106 : pgstat_report_stat(true);
608 : }
609 :
610 4336 : if (should_exit)
611 : {
612 : /*
613 : * Reset the last-start time for this worker so that the launcher will
614 : * restart it without waiting for wal_retrieve_retry_interval.
615 : */
616 6 : ApplyLauncherForgetWorkerStartTime(MySubscription->oid);
617 :
618 6 : proc_exit(0);
619 : }
620 4330 : }
621 :
622 : /*
623 : * Create list of columns for COPY based on logical relation mapping.
624 : */
625 : static List *
626 206 : make_copy_attnamelist(LogicalRepRelMapEntry *rel)
627 : {
628 206 : List *attnamelist = NIL;
629 : int i;
630 :
631 546 : for (i = 0; i < rel->remoterel.natts; i++)
632 : {
633 340 : attnamelist = lappend(attnamelist,
634 340 : makeString(rel->remoterel.attnames[i]));
635 : }
636 :
637 :
638 206 : return attnamelist;
639 : }
640 :
641 : /*
642 : * Data source callback for the COPY FROM, which reads from the remote
643 : * connection and passes the data back to our local COPY.
644 : */
645 : static int
646 15061 : copy_read_data(void *outbuf, int minread, int maxread)
647 : {
648 15061 : int bytesread = 0;
649 : int avail;
650 :
651 : /* If there are some leftover data from previous read, use it. */
652 15061 : avail = copybuf->len - copybuf->cursor;
653 15061 : if (avail)
654 : {
655 0 : if (avail > maxread)
656 0 : avail = maxread;
657 0 : memcpy(outbuf, ©buf->data[copybuf->cursor], avail);
658 0 : copybuf->cursor += avail;
659 0 : maxread -= avail;
660 0 : bytesread += avail;
661 : }
662 :
663 15063 : while (maxread > 0 && bytesread < minread)
664 : {
665 15063 : pgsocket fd = PGINVALID_SOCKET;
666 : int len;
667 15063 : char *buf = NULL;
668 :
669 : for (;;)
670 : {
671 : /* Try read the data. */
672 15063 : len = walrcv_receive(LogRepWorkerWalRcvConn, &buf, &fd);
673 :
674 15063 : CHECK_FOR_INTERRUPTS();
675 :
676 15063 : if (len == 0)
677 2 : break;
678 15061 : else if (len < 0)
679 15061 : return bytesread;
680 : else
681 : {
682 : /* Process the data */
683 14857 : copybuf->data = buf;
684 14857 : copybuf->len = len;
685 14857 : copybuf->cursor = 0;
686 :
687 14857 : avail = copybuf->len - copybuf->cursor;
688 14857 : if (avail > maxread)
689 0 : avail = maxread;
690 14857 : memcpy(outbuf, ©buf->data[copybuf->cursor], avail);
691 14857 : outbuf = (char *) outbuf + avail;
692 14857 : copybuf->cursor += avail;
693 14857 : maxread -= avail;
694 14857 : bytesread += avail;
695 : }
696 :
697 14857 : if (maxread <= 0 || bytesread >= minread)
698 14857 : return bytesread;
699 : }
700 :
701 : /*
702 : * Wait for more data or latch.
703 : */
704 2 : (void) WaitLatchOrSocket(MyLatch,
705 : WL_SOCKET_READABLE | WL_LATCH_SET |
706 : WL_TIMEOUT | WL_EXIT_ON_PM_DEATH,
707 : fd, 1000L, WAIT_EVENT_LOGICAL_SYNC_DATA);
708 :
709 2 : ResetLatch(MyLatch);
710 : }
711 :
712 0 : return bytesread;
713 : }
714 :
715 :
716 : /*
717 : * Get information about remote relation in similar fashion the RELATION
718 : * message provides during replication.
719 : *
720 : * This function also returns (a) the relation qualifications to be used in
721 : * the COPY command, and (b) whether the remote relation has published any
722 : * generated column.
723 : */
724 : static void
725 209 : fetch_remote_table_info(char *nspname, char *relname, LogicalRepRelation *lrel,
726 : List **qual, bool *gencol_published)
727 : {
728 : WalRcvExecResult *res;
729 : StringInfoData cmd;
730 : TupleTableSlot *slot;
731 209 : Oid tableRow[] = {OIDOID, CHAROID, CHAROID};
732 209 : Oid attrRow[] = {INT2OID, TEXTOID, OIDOID, BOOLOID, BOOLOID};
733 209 : Oid qualRow[] = {TEXTOID};
734 : bool isnull;
735 : int natt;
736 209 : StringInfo pub_names = NULL;
737 209 : Bitmapset *included_cols = NULL;
738 209 : int server_version = walrcv_server_version(LogRepWorkerWalRcvConn);
739 :
740 209 : lrel->nspname = nspname;
741 209 : lrel->relname = relname;
742 :
743 : /* First fetch Oid and replica identity. */
744 209 : initStringInfo(&cmd);
745 209 : appendStringInfo(&cmd, "SELECT c.oid, c.relreplident, c.relkind"
746 : " FROM pg_catalog.pg_class c"
747 : " INNER JOIN pg_catalog.pg_namespace n"
748 : " ON (c.relnamespace = n.oid)"
749 : " WHERE n.nspname = %s"
750 : " AND c.relname = %s",
751 : quote_literal_cstr(nspname),
752 : quote_literal_cstr(relname));
753 209 : res = walrcv_exec(LogRepWorkerWalRcvConn, cmd.data,
754 : lengthof(tableRow), tableRow);
755 :
756 209 : if (res->status != WALRCV_OK_TUPLES)
757 0 : ereport(ERROR,
758 : (errcode(ERRCODE_CONNECTION_FAILURE),
759 : errmsg("could not fetch table info for table \"%s.%s\" from publisher: %s",
760 : nspname, relname, res->err)));
761 :
762 209 : slot = MakeSingleTupleTableSlot(res->tupledesc, &TTSOpsMinimalTuple);
763 209 : if (!tuplestore_gettupleslot(res->tuplestore, true, false, slot))
764 1 : ereport(ERROR,
765 : (errcode(ERRCODE_UNDEFINED_OBJECT),
766 : errmsg("table \"%s.%s\" not found on publisher",
767 : nspname, relname)));
768 :
769 208 : lrel->remoteid = DatumGetObjectId(slot_getattr(slot, 1, &isnull));
770 : Assert(!isnull);
771 208 : lrel->replident = DatumGetChar(slot_getattr(slot, 2, &isnull));
772 : Assert(!isnull);
773 208 : lrel->relkind = DatumGetChar(slot_getattr(slot, 3, &isnull));
774 : Assert(!isnull);
775 :
776 208 : ExecDropSingleTupleTableSlot(slot);
777 208 : walrcv_clear_result(res);
778 :
779 :
780 : /*
781 : * Get column lists for each relation.
782 : *
783 : * We need to do this before fetching info about column names and types,
784 : * so that we can skip columns that should not be replicated.
785 : */
786 208 : if (server_version >= 150000)
787 : {
788 : WalRcvExecResult *pubres;
789 : TupleTableSlot *tslot;
790 208 : Oid attrsRow[] = {INT2VECTOROID};
791 :
792 : /* Build the pub_names comma-separated string. */
793 208 : pub_names = makeStringInfo();
794 208 : GetPublicationsStr(MySubscription->publications, pub_names, true);
795 :
796 : /*
797 : * Fetch info about column lists for the relation (from all the
798 : * publications).
799 : */
800 208 : resetStringInfo(&cmd);
801 208 : appendStringInfo(&cmd,
802 : "SELECT DISTINCT"
803 : " (CASE WHEN (array_length(gpt.attrs, 1) = c.relnatts)"
804 : " THEN NULL ELSE gpt.attrs END)"
805 : " FROM pg_publication p,"
806 : " LATERAL pg_get_publication_tables(p.pubname) gpt,"
807 : " pg_class c"
808 : " WHERE gpt.relid = %u AND c.oid = gpt.relid"
809 : " AND p.pubname IN ( %s )",
810 : lrel->remoteid,
811 : pub_names->data);
812 :
813 208 : pubres = walrcv_exec(LogRepWorkerWalRcvConn, cmd.data,
814 : lengthof(attrsRow), attrsRow);
815 :
816 208 : if (pubres->status != WALRCV_OK_TUPLES)
817 0 : ereport(ERROR,
818 : (errcode(ERRCODE_CONNECTION_FAILURE),
819 : errmsg("could not fetch column list info for table \"%s.%s\" from publisher: %s",
820 : nspname, relname, pubres->err)));
821 :
822 : /*
823 : * We don't support the case where the column list is different for
824 : * the same table when combining publications. See comments atop
825 : * fetch_relation_list. So there should be only one row returned.
826 : * Although we already checked this when creating the subscription, we
827 : * still need to check here in case the column list was changed after
828 : * creating the subscription and before the sync worker is started.
829 : */
830 208 : if (tuplestore_tuple_count(pubres->tuplestore) > 1)
831 0 : ereport(ERROR,
832 : errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
833 : errmsg("cannot use different column lists for table \"%s.%s\" in different publications",
834 : nspname, relname));
835 :
836 : /*
837 : * Get the column list and build a single bitmap with the attnums.
838 : *
839 : * If we find a NULL value, it means all the columns should be
840 : * replicated.
841 : */
842 208 : tslot = MakeSingleTupleTableSlot(pubres->tupledesc, &TTSOpsMinimalTuple);
843 208 : if (tuplestore_gettupleslot(pubres->tuplestore, true, false, tslot))
844 : {
845 208 : Datum cfval = slot_getattr(tslot, 1, &isnull);
846 :
847 208 : if (!isnull)
848 : {
849 : ArrayType *arr;
850 : int nelems;
851 : int16 *elems;
852 :
853 22 : arr = DatumGetArrayTypeP(cfval);
854 22 : nelems = ARR_DIMS(arr)[0];
855 22 : elems = (int16 *) ARR_DATA_PTR(arr);
856 :
857 59 : for (natt = 0; natt < nelems; natt++)
858 37 : included_cols = bms_add_member(included_cols, elems[natt]);
859 : }
860 :
861 208 : ExecClearTuple(tslot);
862 : }
863 208 : ExecDropSingleTupleTableSlot(tslot);
864 :
865 208 : walrcv_clear_result(pubres);
866 : }
867 :
868 : /*
869 : * Now fetch column names and types.
870 : */
871 208 : resetStringInfo(&cmd);
872 208 : appendStringInfoString(&cmd,
873 : "SELECT a.attnum,"
874 : " a.attname,"
875 : " a.atttypid,"
876 : " a.attnum = ANY(i.indkey)");
877 :
878 : /* Generated columns can be replicated since version 18. */
879 208 : if (server_version >= 180000)
880 208 : appendStringInfoString(&cmd, ", a.attgenerated != ''");
881 :
882 416 : appendStringInfo(&cmd,
883 : " FROM pg_catalog.pg_attribute a"
884 : " LEFT JOIN pg_catalog.pg_index i"
885 : " ON (i.indexrelid = pg_get_replica_identity_index(%u))"
886 : " WHERE a.attnum > 0::pg_catalog.int2"
887 : " AND NOT a.attisdropped %s"
888 : " AND a.attrelid = %u"
889 : " ORDER BY a.attnum",
890 : lrel->remoteid,
891 208 : (server_version >= 120000 && server_version < 180000 ?
892 : "AND a.attgenerated = ''" : ""),
893 : lrel->remoteid);
894 208 : res = walrcv_exec(LogRepWorkerWalRcvConn, cmd.data,
895 : server_version >= 180000 ? lengthof(attrRow) : lengthof(attrRow) - 1, attrRow);
896 :
897 208 : if (res->status != WALRCV_OK_TUPLES)
898 0 : ereport(ERROR,
899 : (errcode(ERRCODE_CONNECTION_FAILURE),
900 : errmsg("could not fetch table info for table \"%s.%s\" from publisher: %s",
901 : nspname, relname, res->err)));
902 :
903 : /* We don't know the number of rows coming, so allocate enough space. */
904 208 : lrel->attnames = palloc0_array(char *, MaxTupleAttributeNumber);
905 208 : lrel->atttyps = palloc0_array(Oid, MaxTupleAttributeNumber);
906 208 : lrel->attkeys = NULL;
907 :
908 : /*
909 : * Store the columns as a list of names. Ignore those that are not
910 : * present in the column list, if there is one.
911 : */
912 208 : natt = 0;
913 208 : slot = MakeSingleTupleTableSlot(res->tupledesc, &TTSOpsMinimalTuple);
914 585 : while (tuplestore_gettupleslot(res->tuplestore, true, false, slot))
915 : {
916 : char *rel_colname;
917 : AttrNumber attnum;
918 :
919 377 : attnum = DatumGetInt16(slot_getattr(slot, 1, &isnull));
920 : Assert(!isnull);
921 :
922 : /* If the column is not in the column list, skip it. */
923 377 : if (included_cols != NULL && !bms_is_member(attnum, included_cols))
924 : {
925 31 : ExecClearTuple(slot);
926 31 : continue;
927 : }
928 :
929 346 : rel_colname = TextDatumGetCString(slot_getattr(slot, 2, &isnull));
930 : Assert(!isnull);
931 :
932 346 : lrel->attnames[natt] = rel_colname;
933 346 : lrel->atttyps[natt] = DatumGetObjectId(slot_getattr(slot, 3, &isnull));
934 : Assert(!isnull);
935 :
936 346 : if (DatumGetBool(slot_getattr(slot, 4, &isnull)))
937 111 : lrel->attkeys = bms_add_member(lrel->attkeys, natt);
938 :
939 : /* Remember if the remote table has published any generated column. */
940 346 : if (server_version >= 180000 && !(*gencol_published))
941 : {
942 346 : *gencol_published = DatumGetBool(slot_getattr(slot, 5, &isnull));
943 : Assert(!isnull);
944 : }
945 :
946 : /* Should never happen. */
947 346 : if (++natt >= MaxTupleAttributeNumber)
948 0 : elog(ERROR, "too many columns in remote table \"%s.%s\"",
949 : nspname, relname);
950 :
951 346 : ExecClearTuple(slot);
952 : }
953 208 : ExecDropSingleTupleTableSlot(slot);
954 :
955 208 : lrel->natts = natt;
956 :
957 208 : walrcv_clear_result(res);
958 :
959 : /*
960 : * Get relation's row filter expressions. DISTINCT avoids the same
961 : * expression of a table in multiple publications from being included
962 : * multiple times in the final expression.
963 : *
964 : * We need to copy the row even if it matches just one of the
965 : * publications, so we later combine all the quals with OR.
966 : *
967 : * For initial synchronization, row filtering can be ignored in following
968 : * cases:
969 : *
970 : * 1) one of the subscribed publications for the table hasn't specified
971 : * any row filter
972 : *
973 : * 2) one of the subscribed publications has puballtables set to true
974 : *
975 : * 3) one of the subscribed publications is declared as TABLES IN SCHEMA
976 : * that includes this relation
977 : */
978 208 : if (server_version >= 150000)
979 : {
980 : /* Reuse the already-built pub_names. */
981 : Assert(pub_names != NULL);
982 :
983 : /* Check for row filters. */
984 208 : resetStringInfo(&cmd);
985 208 : appendStringInfo(&cmd,
986 : "SELECT DISTINCT pg_get_expr(gpt.qual, gpt.relid)"
987 : " FROM pg_publication p,"
988 : " LATERAL pg_get_publication_tables(p.pubname) gpt"
989 : " WHERE gpt.relid = %u"
990 : " AND p.pubname IN ( %s )",
991 : lrel->remoteid,
992 : pub_names->data);
993 :
994 208 : res = walrcv_exec(LogRepWorkerWalRcvConn, cmd.data, 1, qualRow);
995 :
996 208 : if (res->status != WALRCV_OK_TUPLES)
997 0 : ereport(ERROR,
998 : (errmsg("could not fetch table WHERE clause info for table \"%s.%s\" from publisher: %s",
999 : nspname, relname, res->err)));
1000 :
1001 : /*
1002 : * Multiple row filter expressions for the same table will be combined
1003 : * by COPY using OR. If any of the filter expressions for this table
1004 : * are null, it means the whole table will be copied. In this case it
1005 : * is not necessary to construct a unified row filter expression at
1006 : * all.
1007 : */
1008 208 : slot = MakeSingleTupleTableSlot(res->tupledesc, &TTSOpsMinimalTuple);
1009 223 : while (tuplestore_gettupleslot(res->tuplestore, true, false, slot))
1010 : {
1011 212 : Datum rf = slot_getattr(slot, 1, &isnull);
1012 :
1013 212 : if (!isnull)
1014 15 : *qual = lappend(*qual, makeString(TextDatumGetCString(rf)));
1015 : else
1016 : {
1017 : /* Ignore filters and cleanup as necessary. */
1018 197 : if (*qual)
1019 : {
1020 3 : list_free_deep(*qual);
1021 3 : *qual = NIL;
1022 : }
1023 197 : break;
1024 : }
1025 :
1026 15 : ExecClearTuple(slot);
1027 : }
1028 208 : ExecDropSingleTupleTableSlot(slot);
1029 :
1030 208 : walrcv_clear_result(res);
1031 208 : destroyStringInfo(pub_names);
1032 : }
1033 :
1034 208 : pfree(cmd.data);
1035 208 : }
1036 :
1037 : /*
1038 : * Copy existing data of a table from publisher.
1039 : *
1040 : * Caller is responsible for locking the local relation.
1041 : */
1042 : static void
1043 209 : copy_table(Relation rel)
1044 : {
1045 : LogicalRepRelMapEntry *relmapentry;
1046 : LogicalRepRelation lrel;
1047 209 : List *qual = NIL;
1048 : WalRcvExecResult *res;
1049 : StringInfoData cmd;
1050 : CopyFromState cstate;
1051 : List *attnamelist;
1052 : ParseState *pstate;
1053 209 : List *options = NIL;
1054 209 : bool gencol_published = false;
1055 :
1056 : /* Get the publisher relation info. */
1057 209 : fetch_remote_table_info(get_namespace_name(RelationGetNamespace(rel)),
1058 209 : RelationGetRelationName(rel), &lrel, &qual,
1059 : &gencol_published);
1060 :
1061 : /* Put the relation into relmap. */
1062 208 : logicalrep_relmap_update(&lrel);
1063 :
1064 : /* Map the publisher relation to local one. */
1065 208 : relmapentry = logicalrep_rel_open(lrel.remoteid, NoLock);
1066 : Assert(rel == relmapentry->localrel);
1067 :
1068 : /* Start copy on the publisher. */
1069 206 : initStringInfo(&cmd);
1070 :
1071 : /* Regular or partitioned table with no row filter or generated columns */
1072 206 : if ((lrel.relkind == RELKIND_RELATION || lrel.relkind == RELKIND_PARTITIONED_TABLE)
1073 206 : && qual == NIL && !gencol_published)
1074 : {
1075 192 : appendStringInfo(&cmd, "COPY %s",
1076 192 : quote_qualified_identifier(lrel.nspname, lrel.relname));
1077 :
1078 : /* If the table has columns, then specify the columns */
1079 192 : if (lrel.natts)
1080 : {
1081 191 : appendStringInfoString(&cmd, " (");
1082 :
1083 : /*
1084 : * XXX Do we need to list the columns in all cases? Maybe we're
1085 : * replicating all columns?
1086 : */
1087 510 : for (int i = 0; i < lrel.natts; i++)
1088 : {
1089 319 : if (i > 0)
1090 128 : appendStringInfoString(&cmd, ", ");
1091 :
1092 319 : appendStringInfoString(&cmd, quote_identifier(lrel.attnames[i]));
1093 : }
1094 :
1095 191 : appendStringInfoChar(&cmd, ')');
1096 : }
1097 :
1098 192 : appendStringInfoString(&cmd, " TO STDOUT");
1099 : }
1100 : else
1101 : {
1102 : /*
1103 : * For non-tables and tables with row filters, we need to do COPY
1104 : * (SELECT ...), but we can't just do SELECT * because we may need to
1105 : * copy only subset of columns including generated columns. For tables
1106 : * with any row filters, build a SELECT query with OR'ed row filters
1107 : * for COPY.
1108 : *
1109 : * We also need to use this same COPY (SELECT ...) syntax when
1110 : * generated columns are published, because copy of generated columns
1111 : * is not supported by the normal COPY.
1112 : */
1113 14 : appendStringInfoString(&cmd, "COPY (SELECT ");
1114 35 : for (int i = 0; i < lrel.natts; i++)
1115 : {
1116 21 : appendStringInfoString(&cmd, quote_identifier(lrel.attnames[i]));
1117 21 : if (i < lrel.natts - 1)
1118 7 : appendStringInfoString(&cmd, ", ");
1119 : }
1120 :
1121 14 : appendStringInfoString(&cmd, " FROM ");
1122 :
1123 : /*
1124 : * For regular tables, make sure we don't copy data from a child that
1125 : * inherits the named table as those will be copied separately.
1126 : */
1127 14 : if (lrel.relkind == RELKIND_RELATION)
1128 11 : appendStringInfoString(&cmd, "ONLY ");
1129 :
1130 14 : appendStringInfoString(&cmd, quote_qualified_identifier(lrel.nspname, lrel.relname));
1131 : /* list of OR'ed filters */
1132 14 : if (qual != NIL)
1133 : {
1134 : ListCell *lc;
1135 11 : char *q = strVal(linitial(qual));
1136 :
1137 11 : appendStringInfo(&cmd, " WHERE %s", q);
1138 12 : for_each_from(lc, qual, 1)
1139 : {
1140 1 : q = strVal(lfirst(lc));
1141 1 : appendStringInfo(&cmd, " OR %s", q);
1142 : }
1143 11 : list_free_deep(qual);
1144 : }
1145 :
1146 14 : appendStringInfoString(&cmd, ") TO STDOUT");
1147 : }
1148 :
1149 : /*
1150 : * Prior to v16, initial table synchronization will use text format even
1151 : * if the binary option is enabled for a subscription.
1152 : */
1153 206 : if (walrcv_server_version(LogRepWorkerWalRcvConn) >= 160000 &&
1154 206 : MySubscription->binary)
1155 : {
1156 5 : appendStringInfoString(&cmd, " WITH (FORMAT binary)");
1157 5 : options = list_make1(makeDefElem("format",
1158 : (Node *) makeString("binary"), -1));
1159 : }
1160 :
1161 206 : res = walrcv_exec(LogRepWorkerWalRcvConn, cmd.data, 0, NULL);
1162 206 : pfree(cmd.data);
1163 206 : if (res->status != WALRCV_OK_COPY_OUT)
1164 0 : ereport(ERROR,
1165 : (errcode(ERRCODE_CONNECTION_FAILURE),
1166 : errmsg("could not start initial contents copy for table \"%s.%s\": %s",
1167 : lrel.nspname, lrel.relname, res->err)));
1168 206 : walrcv_clear_result(res);
1169 :
1170 206 : copybuf = makeStringInfo();
1171 :
1172 206 : pstate = make_parsestate(NULL);
1173 206 : (void) addRangeTableEntryForRelation(pstate, rel, AccessShareLock,
1174 : NULL, false, false);
1175 :
1176 206 : attnamelist = make_copy_attnamelist(relmapentry);
1177 206 : cstate = BeginCopyFrom(pstate, rel, NULL, NULL, false, copy_read_data, attnamelist, options);
1178 :
1179 : /* Do the copy */
1180 205 : (void) CopyFrom(cstate);
1181 :
1182 194 : logicalrep_rel_close(relmapentry, NoLock);
1183 194 : }
1184 :
1185 : /*
1186 : * Determine the tablesync slot name.
1187 : *
1188 : * The name must not exceed NAMEDATALEN - 1 because of remote node constraints
1189 : * on slot name length. We append system_identifier to avoid slot_name
1190 : * collision with subscriptions in other clusters. With the current scheme
1191 : * pg_%u_sync_%u_UINT64_FORMAT (3 + 10 + 6 + 10 + 20 + '\0'), the maximum
1192 : * length of slot_name will be 50.
1193 : *
1194 : * The returned slot name is stored in the supplied buffer (syncslotname) with
1195 : * the given size.
1196 : *
1197 : * Note: We don't use the subscription slot name as part of tablesync slot name
1198 : * because we are responsible for cleaning up these slots and it could become
1199 : * impossible to recalculate what name to cleanup if the subscription slot name
1200 : * had changed.
1201 : */
1202 : void
1203 408 : ReplicationSlotNameForTablesync(Oid suboid, Oid relid,
1204 : char *syncslotname, Size szslot)
1205 : {
1206 408 : snprintf(syncslotname, szslot, "pg_%u_sync_%u_" UINT64_FORMAT, suboid,
1207 : relid, GetSystemIdentifier());
1208 408 : }
1209 :
1210 : /*
1211 : * Start syncing the table in the sync worker.
1212 : *
1213 : * If nothing needs to be done to sync the table, we exit the worker without
1214 : * any further action.
1215 : *
1216 : * The returned slot name is palloc'ed in current memory context.
1217 : */
1218 : static char *
1219 210 : LogicalRepSyncTableStart(XLogRecPtr *origin_startpos)
1220 : {
1221 : char *slotname;
1222 : char *err;
1223 : char relstate;
1224 : XLogRecPtr relstate_lsn;
1225 : Relation rel;
1226 : AclResult aclresult;
1227 : WalRcvExecResult *res;
1228 : char originname[NAMEDATALEN];
1229 : ReplOriginId originid;
1230 : UserContext ucxt;
1231 : bool must_use_password;
1232 : bool run_as_owner;
1233 :
1234 : /* Check the state of the table synchronization. */
1235 210 : StartTransactionCommand();
1236 210 : relstate = GetSubscriptionRelState(MyLogicalRepWorker->subid,
1237 210 : MyLogicalRepWorker->relid,
1238 : &relstate_lsn);
1239 210 : CommitTransactionCommand();
1240 :
1241 : /* Is the use of a password mandatory? */
1242 415 : must_use_password = MySubscription->passwordrequired &&
1243 205 : !MySubscription->ownersuperuser;
1244 :
1245 210 : SpinLockAcquire(&MyLogicalRepWorker->relmutex);
1246 210 : MyLogicalRepWorker->relstate = relstate;
1247 210 : MyLogicalRepWorker->relstate_lsn = relstate_lsn;
1248 210 : SpinLockRelease(&MyLogicalRepWorker->relmutex);
1249 :
1250 : /*
1251 : * If synchronization is already done or no longer necessary, exit now
1252 : * that we've updated shared memory state.
1253 : */
1254 210 : switch (relstate)
1255 : {
1256 0 : case SUBREL_STATE_SYNCDONE:
1257 : case SUBREL_STATE_READY:
1258 : case SUBREL_STATE_UNKNOWN:
1259 0 : FinishSyncWorker(); /* doesn't return */
1260 : }
1261 :
1262 : /* Calculate the name of the tablesync slot. */
1263 210 : slotname = (char *) palloc(NAMEDATALEN);
1264 210 : ReplicationSlotNameForTablesync(MySubscription->oid,
1265 210 : MyLogicalRepWorker->relid,
1266 : slotname,
1267 : NAMEDATALEN);
1268 :
1269 : /*
1270 : * Here we use the slot name instead of the subscription name as the
1271 : * application_name, so that it is different from the leader apply worker,
1272 : * so that synchronous replication can distinguish them.
1273 : */
1274 210 : LogRepWorkerWalRcvConn =
1275 210 : walrcv_connect(MySubscription->conninfo, true, true,
1276 : must_use_password,
1277 : slotname, &err);
1278 210 : if (LogRepWorkerWalRcvConn == NULL)
1279 0 : ereport(ERROR,
1280 : (errcode(ERRCODE_CONNECTION_FAILURE),
1281 : errmsg("table synchronization worker for subscription \"%s\" could not connect to the publisher: %s",
1282 : MySubscription->name, err)));
1283 :
1284 : Assert(MyLogicalRepWorker->relstate == SUBREL_STATE_INIT ||
1285 : MyLogicalRepWorker->relstate == SUBREL_STATE_DATASYNC ||
1286 : MyLogicalRepWorker->relstate == SUBREL_STATE_FINISHEDCOPY);
1287 :
1288 : /* Assign the origin tracking record name. */
1289 210 : ReplicationOriginNameForLogicalRep(MySubscription->oid,
1290 210 : MyLogicalRepWorker->relid,
1291 : originname,
1292 : sizeof(originname));
1293 :
1294 210 : if (MyLogicalRepWorker->relstate == SUBREL_STATE_DATASYNC)
1295 : {
1296 : /*
1297 : * We have previously errored out before finishing the copy so the
1298 : * replication slot might exist. We want to remove the slot if it
1299 : * already exists and proceed.
1300 : *
1301 : * XXX We could also instead try to drop the slot, last time we failed
1302 : * but for that, we might need to clean up the copy state as it might
1303 : * be in the middle of fetching the rows. Also, if there is a network
1304 : * breakdown then it wouldn't have succeeded so trying it next time
1305 : * seems like a better bet.
1306 : */
1307 11 : ReplicationSlotDropAtPubNode(LogRepWorkerWalRcvConn, slotname, true);
1308 : }
1309 199 : else if (MyLogicalRepWorker->relstate == SUBREL_STATE_FINISHEDCOPY)
1310 : {
1311 : /*
1312 : * The COPY phase was previously done, but tablesync then crashed
1313 : * before it was able to finish normally.
1314 : */
1315 0 : StartTransactionCommand();
1316 :
1317 : /*
1318 : * The origin tracking name must already exist. It was created first
1319 : * time this tablesync was launched.
1320 : */
1321 0 : originid = replorigin_by_name(originname, false);
1322 0 : replorigin_session_setup(originid, 0);
1323 0 : replorigin_xact_state.origin = originid;
1324 0 : *origin_startpos = replorigin_session_get_progress(false);
1325 :
1326 0 : CommitTransactionCommand();
1327 :
1328 0 : goto copy_table_done;
1329 : }
1330 :
1331 210 : SpinLockAcquire(&MyLogicalRepWorker->relmutex);
1332 210 : MyLogicalRepWorker->relstate = SUBREL_STATE_DATASYNC;
1333 210 : MyLogicalRepWorker->relstate_lsn = InvalidXLogRecPtr;
1334 210 : SpinLockRelease(&MyLogicalRepWorker->relmutex);
1335 :
1336 : /*
1337 : * Update the state, create the replication origin, and make them visible
1338 : * to others.
1339 : */
1340 210 : StartTransactionCommand();
1341 210 : UpdateSubscriptionRelState(MyLogicalRepWorker->subid,
1342 210 : MyLogicalRepWorker->relid,
1343 210 : MyLogicalRepWorker->relstate,
1344 210 : MyLogicalRepWorker->relstate_lsn,
1345 : false);
1346 :
1347 : /*
1348 : * Create the replication origin in a separate transaction from the one
1349 : * that sets up the origin in shared memory. This prevents the risk that
1350 : * changes to the origin in shared memory cannot be rolled back if the
1351 : * transaction aborts.
1352 : */
1353 209 : originid = replorigin_by_name(originname, true);
1354 209 : if (!OidIsValid(originid))
1355 198 : originid = replorigin_create(originname);
1356 :
1357 209 : CommitTransactionCommand();
1358 209 : pgstat_report_stat(true);
1359 :
1360 209 : StartTransactionCommand();
1361 :
1362 : /*
1363 : * Use a standard write lock here. It might be better to disallow access
1364 : * to the table while it's being synchronized. But we don't want to block
1365 : * the main apply process from working and it has to open the relation in
1366 : * RowExclusiveLock when remapping remote relation id to local one.
1367 : */
1368 209 : rel = table_open(MyLogicalRepWorker->relid, RowExclusiveLock);
1369 :
1370 : /*
1371 : * Start a transaction in the remote node in REPEATABLE READ mode. This
1372 : * ensures that both the replication slot we create (see below) and the
1373 : * COPY are consistent with each other.
1374 : */
1375 209 : res = walrcv_exec(LogRepWorkerWalRcvConn,
1376 : "BEGIN READ ONLY ISOLATION LEVEL REPEATABLE READ",
1377 : 0, NULL);
1378 209 : if (res->status != WALRCV_OK_COMMAND)
1379 0 : ereport(ERROR,
1380 : (errcode(ERRCODE_CONNECTION_FAILURE),
1381 : errmsg("table copy could not start transaction on publisher: %s",
1382 : res->err)));
1383 209 : walrcv_clear_result(res);
1384 :
1385 : /*
1386 : * Create a new permanent logical decoding slot. This slot will be used
1387 : * for the catchup phase after COPY is done, so tell it to use the
1388 : * snapshot to make the final data consistent.
1389 : */
1390 209 : walrcv_create_slot(LogRepWorkerWalRcvConn,
1391 : slotname, false /* permanent */ , false /* two_phase */ ,
1392 : MySubscription->failover,
1393 : CRS_USE_SNAPSHOT, origin_startpos);
1394 :
1395 : /*
1396 : * Advance the origin to the LSN got from walrcv_create_slot and then set
1397 : * up the origin. The advancement is WAL logged for the purpose of
1398 : * recovery. Locks are to prevent the replication origin from vanishing
1399 : * while advancing.
1400 : *
1401 : * The purpose of doing these before the copy is to avoid doing the copy
1402 : * again due to any error in advancing or setting up origin tracking.
1403 : */
1404 209 : LockRelationOid(ReplicationOriginRelationId, RowExclusiveLock);
1405 209 : replorigin_advance(originid, *origin_startpos, InvalidXLogRecPtr,
1406 : true /* go backward */ , true /* WAL log */ );
1407 209 : UnlockRelationOid(ReplicationOriginRelationId, RowExclusiveLock);
1408 :
1409 209 : replorigin_session_setup(originid, 0);
1410 209 : replorigin_xact_state.origin = originid;
1411 :
1412 : /*
1413 : * If the user did not opt to run as the owner of the subscription
1414 : * ('run_as_owner'), then copy the table as the owner of the table.
1415 : */
1416 209 : run_as_owner = MySubscription->runasowner;
1417 209 : if (!run_as_owner)
1418 208 : SwitchToUntrustedUser(rel->rd_rel->relowner, &ucxt);
1419 :
1420 : /*
1421 : * Check that our table sync worker has permission to insert into the
1422 : * target table.
1423 : */
1424 209 : aclresult = pg_class_aclcheck(RelationGetRelid(rel), GetUserId(),
1425 : ACL_INSERT);
1426 209 : if (aclresult != ACLCHECK_OK)
1427 0 : aclcheck_error(aclresult,
1428 0 : get_relkind_objtype(rel->rd_rel->relkind),
1429 0 : RelationGetRelationName(rel));
1430 :
1431 : /*
1432 : * COPY FROM does not honor RLS policies. That is not a problem for
1433 : * subscriptions owned by roles with BYPASSRLS privilege (or superuser,
1434 : * who has it implicitly), but other roles should not be able to
1435 : * circumvent RLS. Disallow logical replication into RLS enabled
1436 : * relations for such roles.
1437 : */
1438 209 : if (check_enable_rls(RelationGetRelid(rel), InvalidOid, false) == RLS_ENABLED)
1439 0 : ereport(ERROR,
1440 : (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
1441 : errmsg("user \"%s\" cannot replicate into relation with row-level security enabled: \"%s\"",
1442 : GetUserNameFromId(GetUserId(), true),
1443 : RelationGetRelationName(rel))));
1444 :
1445 : /* Now do the initial data copy */
1446 209 : PushActiveSnapshot(GetTransactionSnapshot());
1447 209 : copy_table(rel);
1448 194 : PopActiveSnapshot();
1449 :
1450 194 : res = walrcv_exec(LogRepWorkerWalRcvConn, "COMMIT", 0, NULL);
1451 194 : if (res->status != WALRCV_OK_COMMAND)
1452 0 : ereport(ERROR,
1453 : (errcode(ERRCODE_CONNECTION_FAILURE),
1454 : errmsg("table copy could not finish transaction on publisher: %s",
1455 : res->err)));
1456 194 : walrcv_clear_result(res);
1457 :
1458 194 : if (!run_as_owner)
1459 193 : RestoreUserContext(&ucxt);
1460 :
1461 194 : table_close(rel, NoLock);
1462 :
1463 : /* Make the copy visible. */
1464 194 : CommandCounterIncrement();
1465 :
1466 : /*
1467 : * Update the persisted state to indicate the COPY phase is done; make it
1468 : * visible to others.
1469 : */
1470 194 : UpdateSubscriptionRelState(MyLogicalRepWorker->subid,
1471 194 : MyLogicalRepWorker->relid,
1472 : SUBREL_STATE_FINISHEDCOPY,
1473 194 : MyLogicalRepWorker->relstate_lsn,
1474 : false);
1475 :
1476 194 : CommitTransactionCommand();
1477 :
1478 194 : copy_table_done:
1479 :
1480 194 : elog(DEBUG1,
1481 : "LogicalRepSyncTableStart: '%s' origin_startpos lsn %X/%08X",
1482 : originname, LSN_FORMAT_ARGS(*origin_startpos));
1483 :
1484 : /*
1485 : * We are done with the initial data synchronization, update the state.
1486 : */
1487 194 : SpinLockAcquire(&MyLogicalRepWorker->relmutex);
1488 194 : MyLogicalRepWorker->relstate = SUBREL_STATE_SYNCWAIT;
1489 194 : MyLogicalRepWorker->relstate_lsn = *origin_startpos;
1490 194 : SpinLockRelease(&MyLogicalRepWorker->relmutex);
1491 :
1492 : /*
1493 : * Finally, wait until the leader apply worker tells us to catch up and
1494 : * then return to let LogicalRepApplyLoop do it.
1495 : */
1496 194 : wait_for_worker_state_change(SUBREL_STATE_CATCHUP);
1497 194 : return slotname;
1498 : }
1499 :
1500 : /*
1501 : * Execute the initial sync with error handling. Disable the subscription,
1502 : * if it's required.
1503 : *
1504 : * Allocate the slot name in long-lived context on return. Note that we don't
1505 : * handle FATAL errors which are probably because of system resource error and
1506 : * are not repeatable.
1507 : */
1508 : static void
1509 210 : start_table_sync(XLogRecPtr *origin_startpos, char **slotname)
1510 : {
1511 210 : char *sync_slotname = NULL;
1512 :
1513 : Assert(am_tablesync_worker());
1514 :
1515 210 : PG_TRY();
1516 : {
1517 : /* Call initial sync. */
1518 210 : sync_slotname = LogicalRepSyncTableStart(origin_startpos);
1519 : }
1520 15 : PG_CATCH();
1521 : {
1522 15 : if (MySubscription->disableonerr)
1523 1 : DisableSubscriptionAndExit();
1524 : else
1525 : {
1526 : /*
1527 : * Report the worker failed during table synchronization. Abort
1528 : * the current transaction so that the stats message is sent in an
1529 : * idle state.
1530 : */
1531 14 : AbortOutOfAnyTransaction();
1532 14 : pgstat_report_subscription_error(MySubscription->oid);
1533 :
1534 14 : PG_RE_THROW();
1535 : }
1536 : }
1537 194 : PG_END_TRY();
1538 :
1539 : /* allocate slot name in long-lived context */
1540 194 : *slotname = MemoryContextStrdup(ApplyContext, sync_slotname);
1541 194 : pfree(sync_slotname);
1542 194 : }
1543 :
1544 : /*
1545 : * Runs the tablesync worker.
1546 : *
1547 : * It starts syncing tables. After a successful sync, sets streaming options
1548 : * and starts streaming to catchup with apply worker.
1549 : */
1550 : static void
1551 210 : run_tablesync_worker(void)
1552 : {
1553 : char originname[NAMEDATALEN];
1554 210 : XLogRecPtr origin_startpos = InvalidXLogRecPtr;
1555 210 : char *slotname = NULL;
1556 : WalRcvStreamOptions options;
1557 :
1558 210 : start_table_sync(&origin_startpos, &slotname);
1559 :
1560 194 : ReplicationOriginNameForLogicalRep(MySubscription->oid,
1561 194 : MyLogicalRepWorker->relid,
1562 : originname,
1563 : sizeof(originname));
1564 :
1565 194 : set_apply_error_context_origin(originname);
1566 :
1567 194 : set_stream_options(&options, slotname, &origin_startpos);
1568 :
1569 194 : walrcv_startstreaming(LogRepWorkerWalRcvConn, &options);
1570 :
1571 : /* Apply the changes till we catchup with the apply worker. */
1572 194 : start_apply(origin_startpos);
1573 0 : }
1574 :
1575 : /* Logical Replication Tablesync worker entry point */
1576 : void
1577 210 : TableSyncWorkerMain(Datum main_arg)
1578 : {
1579 210 : int worker_slot = DatumGetInt32(main_arg);
1580 :
1581 210 : SetupApplyOrSyncWorker(worker_slot);
1582 :
1583 210 : run_tablesync_worker();
1584 :
1585 0 : FinishSyncWorker();
1586 : }
1587 :
1588 : /*
1589 : * If the subscription has no tables then return false.
1590 : *
1591 : * Otherwise, are all tablesyncs READY?
1592 : *
1593 : * Note: This function is not suitable to be called from outside of apply or
1594 : * tablesync workers because MySubscription needs to be already initialized.
1595 : */
1596 : bool
1597 221 : AllTablesyncsReady(void)
1598 : {
1599 : bool started_tx;
1600 : bool has_tables;
1601 :
1602 : /* We need up-to-date sync state info for subscription tables here. */
1603 221 : FetchRelationStates(&has_tables, NULL, &started_tx);
1604 :
1605 221 : if (started_tx)
1606 : {
1607 13 : CommitTransactionCommand();
1608 13 : pgstat_report_stat(true);
1609 : }
1610 :
1611 : /*
1612 : * Return false when there are no tables in subscription or not all tables
1613 : * are in ready state; true otherwise.
1614 : */
1615 221 : return has_tables && (table_states_not_ready == NIL);
1616 : }
1617 :
1618 : /*
1619 : * Return whether the subscription currently has any tables.
1620 : *
1621 : * Note: Unlike HasSubscriptionTables(), this function relies on cached
1622 : * information for subscription tables. Additionally, it should not be
1623 : * invoked outside of apply or tablesync workers, as MySubscription must be
1624 : * initialized first.
1625 : */
1626 : bool
1627 140 : HasSubscriptionTablesCached(void)
1628 : {
1629 : bool started_tx;
1630 : bool has_tables;
1631 :
1632 : /* We need up-to-date subscription tables info here */
1633 140 : FetchRelationStates(&has_tables, NULL, &started_tx);
1634 :
1635 140 : if (started_tx)
1636 : {
1637 1 : CommitTransactionCommand();
1638 1 : pgstat_report_stat(true);
1639 : }
1640 :
1641 140 : return has_tables;
1642 : }
1643 :
1644 : /*
1645 : * Update the two_phase state of the specified subscription in pg_subscription.
1646 : */
1647 : void
1648 8 : UpdateTwoPhaseState(Oid suboid, char new_state)
1649 : {
1650 : Relation rel;
1651 : HeapTuple tup;
1652 : bool nulls[Natts_pg_subscription];
1653 : bool replaces[Natts_pg_subscription];
1654 : Datum values[Natts_pg_subscription];
1655 :
1656 : Assert(new_state == LOGICALREP_TWOPHASE_STATE_DISABLED ||
1657 : new_state == LOGICALREP_TWOPHASE_STATE_PENDING ||
1658 : new_state == LOGICALREP_TWOPHASE_STATE_ENABLED);
1659 :
1660 8 : rel = table_open(SubscriptionRelationId, RowExclusiveLock);
1661 8 : tup = SearchSysCacheCopy1(SUBSCRIPTIONOID, ObjectIdGetDatum(suboid));
1662 8 : if (!HeapTupleIsValid(tup))
1663 0 : elog(ERROR,
1664 : "cache lookup failed for subscription oid %u",
1665 : suboid);
1666 :
1667 : /* Form a new tuple. */
1668 8 : memset(values, 0, sizeof(values));
1669 8 : memset(nulls, false, sizeof(nulls));
1670 8 : memset(replaces, false, sizeof(replaces));
1671 :
1672 : /* And update/set two_phase state */
1673 8 : values[Anum_pg_subscription_subtwophasestate - 1] = CharGetDatum(new_state);
1674 8 : replaces[Anum_pg_subscription_subtwophasestate - 1] = true;
1675 :
1676 8 : tup = heap_modify_tuple(tup, RelationGetDescr(rel),
1677 : values, nulls, replaces);
1678 8 : CatalogTupleUpdate(rel, &tup->t_self, tup);
1679 :
1680 8 : heap_freetuple(tup);
1681 8 : table_close(rel, RowExclusiveLock);
1682 8 : }
|
