Horton

Re: Synchronizing slots from primary to standby

Peter Smith <smithpb2250@gmail.com>

View thread View on postgr.es
From: Peter Smith <smithpb2250@gmail.com>
To: shveta malik <shveta.malik@gmail.com>
Cc: Bertrand Drouvot <bertranddrouvot.pg@gmail.com>, Amit Kapila <amit.kapila16@gmail.com>, Masahiko Sawada <sawada.mshk@gmail.com>, "Zhijie Hou (Fujitsu)" <houzj.fnst@fujitsu.com>, Dilip Kumar <dilipbalaut@gmail.com>, Nisha Moond <nisha.moond412@gmail.com>, "Hayato Kuroda (Fujitsu)" <kuroda.hayato@fujitsu.com>, Bharath Rupireddy <bharath.rupireddyforpostgres@gmail.com>, Peter Eisentraut <peter.eisentraut@enterprisedb.com>, Bruce Momjian <bruce@momjian.us>, Ashutosh Sharma <ashu.coek88@gmail.com>, Andres Freund <andres@anarazel.de>, pgsql-hackers <pgsql-hackers@postgresql.org>, Ajin Cherian <itsajin@gmail.com>, Alvaro Herrera <alvherre@alvh.no-ip.org>
Date: 2024-01-25T05:08:48Z
Lists: pgsql-hackers 
Here are some review comments for v67-0002.

======
src/backend/replication/logical/slotsync.c

1.
+/* The sleep time (ms) between slot-sync cycles varies dynamically
+ * (within a MIN/MAX range) according to slot activity. See
+ * wait_for_slot_activity() for details.
+ */
+#define MIN_WORKER_NAPTIME_MS  200
+#define MAX_WORKER_NAPTIME_MS  30000 /* 30s */
+
+static long sleep_ms = MIN_WORKER_NAPTIME_MS;

In my previous review for this, I meant for there to be no whitespace
between the #defines and the static long sleep_ms so the prior comment
then clearly belongs to all 3 lines

~~~

2. synchronize_one_slot

+ /*
+ * Sanity check: Make sure that concerned WAL is received and flushed
+ * before syncing slot to target lsn received from the primary server.
+ *
+ * This check should never pass as on the primary server, we have waited
+ * for the standby's confirmation before updating the logical slot.
+ */
+ latestFlushPtr = GetStandbyFlushRecPtr(NULL);
+ if (remote_slot->confirmed_lsn > latestFlushPtr)
+ {
+ ereport(LOG,
+ errmsg("skipping slot synchronization as the received slot sync"
+    " LSN %X/%X for slot \"%s\" is ahead of the standby position %X/%X",
+    LSN_FORMAT_ARGS(remote_slot->confirmed_lsn),
+    remote_slot->name,
+    LSN_FORMAT_ARGS(latestFlushPtr)));
+
+ return false;
+ }

Previously in v65 this was an elog, but now it is an ereport. But
since this is a sanity check condition that "should never pass" wasn't
the elog the more appropriate choice?

~~~

3. synchronize_one_slot

+ /*
+ * We don't want any complicated code while holding a spinlock, so do
+ * namestrcpy() and get_database_oid() outside.
+ */
+ namestrcpy(&plugin_name, remote_slot->plugin);
+ dbid = get_database_oid(remote_slot->database, false);

IMO just simplify the whole comment, here and for the other similar
comment in local_slot_update().

SUGGESTION
/* Avoid expensive operations while holding a spinlock. */

~~~

4. synchronize_slots

+ /* Construct the remote_slot tuple and synchronize each slot locally */
+ tupslot = MakeSingleTupleTableSlot(res->tupledesc, &TTSOpsMinimalTuple);
+ while (tuplestore_gettupleslot(res->tuplestore, true, false, tupslot))
+ {
+ bool isnull;
+ RemoteSlot *remote_slot = palloc0(sizeof(RemoteSlot));
+ Datum d;
+
+ remote_slot->name = TextDatumGetCString(slot_getattr(tupslot, 1, &isnull));
+ Assert(!isnull);
+
+ remote_slot->plugin = TextDatumGetCString(slot_getattr(tupslot, 2, &isnull));
+ Assert(!isnull);
+
+ /*
+ * It is possible to get null values for LSN and Xmin if slot is
+ * invalidated on the primary server, so handle accordingly.
+ */
+ d = slot_getattr(tupslot, 3, &isnull);
+ remote_slot->confirmed_lsn = isnull ? InvalidXLogRecPtr :
+ DatumGetLSN(d);
+
+ d = slot_getattr(tupslot, 4, &isnull);
+ remote_slot->restart_lsn = isnull ? InvalidXLogRecPtr : DatumGetLSN(d);
+
+ d = slot_getattr(tupslot, 5, &isnull);
+ remote_slot->catalog_xmin = isnull ? InvalidTransactionId :
+ DatumGetTransactionId(d);
+
+ remote_slot->two_phase = DatumGetBool(slot_getattr(tupslot, 6, &isnull));
+ Assert(!isnull);
+
+ remote_slot->database = TextDatumGetCString(slot_getattr(tupslot,
+ 7, &isnull));
+ Assert(!isnull);
+
+ d = slot_getattr(tupslot, 8, &isnull);
+ remote_slot->invalidated = isnull ? RS_INVAL_NONE :
+ get_slot_invalidation_cause(TextDatumGetCString(d));

Would it be better to get rid of the hardwired column numbers and then
be able to use the SLOTSYNC_COLUMN_COUNT already defined as a sanity
check?

SUGGESTION
int col = 0;
...
remote_slot->name = TextDatumGetCString(slot_getattr(tupslot, ++col, &isnull));
...
remote_slot->plugin = TextDatumGetCString(slot_getattr(tupslot, ++col,
&isnull));
...
d = slot_getattr(tupslot, ++col, &isnull);
remote_slot->confirmed_lsn = isnull ? InvalidXLogRecPtr : DatumGetLSN(d);
...
d = slot_getattr(tupslot, ++col, &isnull);
remote_slot->restart_lsn = isnull ? InvalidXLogRecPtr : DatumGetLSN(d);
...
d = slot_getattr(tupslot, ++col, &isnull);
remote_slot->catalog_xmin = isnull ? InvalidTransactionId :
DatumGetTransactionId(d);
...
remote_slot->two_phase = DatumGetBool(slot_getattr(tupslot, ++col, &isnull));
...
remote_slot->database = TextDatumGetCString(slot_getattr(tupslot,
++col, &isnull));
...
d = slot_getattr(tupslot, ++col, &isnull);
remote_slot->invalidated = isnull ? RS_INVAL_NONE :
get_slot_invalidation_cause(TextDatumGetCString(d));

/* Sanity check */
Asert(col == SLOTSYNC_COLUMN_COUNT);

~~~

5.
+static char *
+validate_parameters_and_get_dbname(void)
+{
+ char    *dbname;

These are configuration issues, so probably all these ereports could
also set errcode(ERRCODE_INVALID_PARAMETER_VALUE).

======
Kind Regards,
Peter Smith.
Fujitsu Australia

Commits

  1. Doc: Add the new section "Logical Replication Failover".

  2. Fix the review comments and a bug in the slot sync code.

  3. Fix a test in failover slots regression test.

  4. Fix the intermittent buildfarm failures in 040_standby_failover_slots_sync.

  5. Split XLogCtl->LogwrtResult into separate struct members

  6. Ensure that the sync slots reach a consistent state after promotion without losing data.

  7. Introduce a new GUC 'standby_slot_names'.

  8. Fix BF failure introduced by commit b3f6b14cf4.

  9. Fixups for commit 93db6cbda0.

  10. Fix BF failure in commit 93db6cbda0.

  11. Add a new slot sync worker to synchronize logical slots.

  12. Improve ERROR/LOG messages added by commits ddd5f4f54a and 7a424ece48.

  13. Disable autovacuum on primary in 040_standby_failover_slots_sync test.

  14. Fix the incorrect format specifier used in commit 7a424ece48.

  15. Change the LOG level in 040_standby_failover_slots_sync.pl to DEBUG2.

  16. Add more LOG and DEBUG messages for slot synchronization.

  17. Another try to fix BF failure introduced in commit ddd5f4f54a.

  18. Fix BF introduced in commit ddd5f4f54a.

  19. Add a slot synchronization function.

  20. Clean-ups for 776621a5e4 and 7329240437.

  21. Enhance libpqrcv APIs to support slot synchronization.

  22. Add a failover option to subscriptions.

  23. Allow setting failover property in the replication command.

  24. Allow to enable failover property for replication slots via SQL API.

  25. Add macros for looping through a List without a ListCell.

  26. Track conflict_reason in pg_replication_slots.

  27. Allow upgrades to preserve the full subscription's state.

  28. Add support for incremental backup.

  29. Simplify some logic in CreateReplicationSlot()

  30. Fix uninitialized access to InitialRunningXacts during decoding after ERROR.

  31. Flexible options for CREATE_REPLICATION_SLOT.

  32. Empty search_path in logical replication apply worker and walsender.

  33. Fix oldest xmin and LSN computation across repslots after advancing

  34. Require C99 (and thus MSCV 2013 upwards).