Thread: pgsql: Implement pg_wal_replay_wait() stored procedure
Implement pg_wal_replay_wait() stored procedure pg_wal_replay_wait() is to be used on standby and specifies waiting for the specific WAL location to be replayed before starting the transaction. This option is useful when the user makes some data changes on primary and needs a guarantee to see these changes on standby. The queue of waiters is stored in the shared memory array sorted by LSN. During replay of WAL waiters whose LSNs are already replayed are deleted from the shared memory array and woken up by setting of their latches. pg_wal_replay_wait() needs to wait without any snapshot held. Otherwise, the snapshot could prevent the replay of WAL records implying a kind of self-deadlock. This is why it is only possible to implement pg_wal_replay_wait() as a procedure working in a non-atomic context, not a function. Catversion is bumped. Discussion: https://postgr.es/m/eb12f9b03851bb2583adab5df9579b4b%40postgrespro.ru Author: Kartyshov Ivan, Alexander Korotkov Reviewed-by: Michael Paquier, Peter Eisentraut, Dilip Kumar, Amit Kapila Reviewed-by: Alexander Lakhin, Bharath Rupireddy, Euler Taveira Branch ------ master Details ------- https://git.postgresql.org/pg/commitdiff/06c418e163e913966e17cb2d3fb1c5f8a8d58308 Modified Files -------------- doc/src/sgml/func.sgml | 113 ++++++++ src/backend/access/transam/xlog.c | 7 + src/backend/access/transam/xlogrecovery.c | 11 + src/backend/catalog/system_functions.sql | 3 + src/backend/commands/Makefile | 3 +- src/backend/commands/meson.build | 1 + src/backend/commands/waitlsn.c | 348 ++++++++++++++++++++++++ src/backend/storage/ipc/ipci.c | 7 + src/backend/storage/lmgr/proc.c | 6 + src/backend/utils/activity/wait_event_names.txt | 1 + src/include/catalog/catversion.h | 2 +- src/include/catalog/pg_proc.dat | 5 + src/include/commands/waitlsn.h | 43 +++ src/test/recovery/meson.build | 1 + src/test/recovery/t/043_wal_replay_wait.pl | 97 +++++++ src/tools/pgindent/typedefs.list | 2 + 16 files changed, 648 insertions(+), 2 deletions(-)
On Tue, Apr 2, 2024 at 10:58 PM Alexander Korotkov <akorotkov@postgresql.org> wrote: > Implement pg_wal_replay_wait() stored procedure I'm trying to figure out if this failure could be related to this commit... https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=culicidae&dt=2024-04-02%2020%3A24%3A55 ------ Regards, Alexander Korotkov
On Wed, Apr 3, 2024 at 9:42 AM Alexander Korotkov <aekorotkov@gmail.com> wrote: > On Tue, Apr 2, 2024 at 10:58 PM Alexander Korotkov > <akorotkov@postgresql.org> wrote: > > Implement pg_wal_replay_wait() stored procedure > > I'm trying to figure out if this failure could be related to this commit... > https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=culicidae&dt=2024-04-02%2020%3A24%3A55 I think you might need to move WaitLSNShmemInit() down to CreateOrAttachShmemStructs(), otherwise -DEXEC_BACKEND build dies with a NULL pointer. (Huh, why doesn't that animal show a backtrace?)
On Wed, 3 Apr 2024 at 09:42, Alexander Korotkov <aekorotkov@gmail.com> wrote: > I'm trying to figure out if this failure could be related to this commit... > https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=culicidae&dt=2024-04-02%2020%3A24%3A55 Yeah, I think it is. The problem is that in WaitLSNSetLatches() waitLSN is NULL, so SpinLockAcquire(&waitLSN->mutex); segfaults. This shared memory segment is initialized in WaitLSNShmemInit() called via CreateSharedMemoryAndSemaphores(). If you look at main() in main.c EXEC_BACKEND builds call SubPostmasterMain(), which has no call to CreateSharedMemoryAndSemaphores(), so waitLSN isn't initialized. David
Hi, On Wed, Apr 3, 2024 at 4:58 AM Alexander Korotkov <akorotkov@postgresql.org> wrote: > > Implement pg_wal_replay_wait() stored procedure > > pg_wal_replay_wait() is to be used on standby and specifies waiting for > the specific WAL location to be replayed before starting the transaction. > This option is useful when the user makes some data changes on primary and > needs a guarantee to see these changes on standby. > > The queue of waiters is stored in the shared memory array sorted by LSN. > During replay of WAL waiters whose LSNs are already replayed are deleted from > the shared memory array and woken up by setting of their latches. > > pg_wal_replay_wait() needs to wait without any snapshot held. Otherwise, > the snapshot could prevent the replay of WAL records implying a kind of > self-deadlock. This is why it is only possible to implement > pg_wal_replay_wait() as a procedure working in a non-atomic context, > not a function. > > Catversion is bumped. > mamba complaints the build error[1]: waitlsn.c: In function 'WaitForLSN': waitlsn.c:275:24: error: 'endtime' may be used uninitialized in this function [-Werror=maybe-uninitialized] 275 | delay_ms = (endtime - GetCurrentTimestamp()) / 1000; | ~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~ Regards, [1] https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=mamba&dt=2024-04-03%2005%3A33%3A24 -- Masahiko Sawada Amazon Web Services: https://aws.amazon.com