Horton

Re: Synchronizing slots from primary to standby

Bertrand Drouvot <bertranddrouvot.pg@gmail.com>

View thread View on postgr.es
From: Bertrand Drouvot <bertranddrouvot.pg@gmail.com>
To: Amit Kapila <amit.kapila16@gmail.com>
Cc: shveta malik <shveta.malik@gmail.com>, Bharath Rupireddy <bharath.rupireddyforpostgres@gmail.com>, "Zhijie Hou (Fujitsu)" <houzj.fnst@fujitsu.com>, Ajin Cherian <itsajin@gmail.com>, Peter Smith <smithpb2250@gmail.com>, Masahiko Sawada <sawada.mshk@gmail.com>, Dilip Kumar <dilipbalaut@gmail.com>, Nisha Moond <nisha.moond412@gmail.com>, "Hayato Kuroda (Fujitsu)" <kuroda.hayato@fujitsu.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>, Alvaro Herrera <alvherre@alvh.no-ip.org>
Date: 2024-04-06T16:47:31Z
Lists: pgsql-hackers 
Hi,

On Sat, Apr 06, 2024 at 10:13:00AM +0530, Amit Kapila wrote:
> On Fri, Apr 5, 2024 at 8:05 PM Bertrand Drouvot
> <bertranddrouvot.pg@gmail.com> wrote:
> 
> I think the new LSN can be visible only when the corresponding WAL is
> written by XLogWrite(). I don't know what in XLogSetAsyncXactLSN() can
> make it visible. In your experiment below, isn't it possible that in
> the meantime WAL writer has written that WAL due to which you are
> seeing an updated location?

What I did is:

session 1:  select pg_current_wal_lsn();\watch 1
session 2:  select pg_backend_pid();

terminal 1: tail -f logfile | grep -i snap 
terminal 2 : gdb -p <backendpid session 2) -ex 'b LogCurrentRunningXacts' + "continue" once in gdb

session 2: SELECT pg_log_standby_snapshot();

That produces a break in the gdb session, then:

Breakpoint 1, LogCurrentRunningXacts (CurrRunningXacts=0x5774f92f8da0 <CurrentRunningXactsData.13>) at standby.c:1346
1346    {
(gdb) n
1350
 
Then next, next until the DEBUG message is emitted (confirmed in terminal 1).

At this stage the DEBUG message shows the new LSN while session 1 still displays
the previous LSN.

Then once XLogSetAsyncXactLSN() is done in the gdb session (terminal 2) then
session 1 displays the new LSN.

This is reproducible as desired.

With more debugging I can see that when the spinlock is released in XLogSetAsyncXactLSN()
then XLogWrite() is doing its job and then session 1 does see the new value (that
happens in this order, and as you said that's expected).

My point is that while the DEBUG message is emitted session 1 still 
see the old LSN (until the new LSN is vsible). I think that we should emit the
DEBUG message once session 1 can see the new value (If not, I think the timestamp
of the DEBUG message can be missleading during debugging purpose).

> I think I am missing how exactly moving DEBUG2 can confirm the above theory.

I meant to say that instead of seeing:

2024-04-05 04:37:05.074 UTC [3854278][background writer][:0] DEBUG:  snapshot of 0+0 running transaction ids (lsn 0/3000098 oldest xid 740 latest complete 739 next xid 740)
2024-04-05 04:37:05.197 UTC [3866475][client backend][2/4:0] LOG:  statement: SELECT '0/3000060' <= replay_lsn AND state = 'streaming'

We would probably see something like:

2024-04-05 04:37:05.<something> UTC [3866475][client backend][2/4:0] LOG:  statement: SELECT '0/3000060' <= replay_lsn AND state = 'streaming'
2024-04-05 04:37:05.<something>+xx UTC [3854278][background writer][:0] DEBUG:  snapshot of 0+0 running transaction ids (lsn 0/3000098 oldest xid 740 latest complete 739 next xid 740)

And then it would be clear that the query has ran before the new LSN is visible.

> > If the theory is proven then I'm not sure we need the extra complexity of
> > injection point here, maybe just relying on the slots created via SQL API could
> > be enough.
> >
> 
> Yeah, that could be the first step. We can probably add an injection
> point to control the bgwrite behavior and then add tests involving
> walsender performing the decoding. But I think it is important to have
> sufficient tests in this area as I see they are quite helpful in
> uncovering the issues.
>

Yeah agree. 

Regards,

-- 
Bertrand Drouvot
PostgreSQL Contributors Team
RDS Open Source Databases
Amazon Web Services: https://aws.amazon.com

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).