Re: PostgreSQL Replication Lag – WALs Streaming but Not Being Applied

Поиск
Список
Период
Сортировка
Искать

Re: PostgreSQL Replication Lag – WALs Streaming but Not Being Applied

От:
Simon Musila - Localhost <smusila1@gmail.com>
Дата:
Вложения:
Hello,
max_standby_streaming_delay=30000ms
max_standby_archive_delay=30000ms

I had changed at some point but still failed.

Do you have any activity recorded in pg_stat_activity on the standby ? Yes

For the replicas displayed: only one doesn't fail.

Regards,
Musila

On Tue, Aug 18, 2026 at 9:15 PM Avinash Vallarapu <avinash.vallarapu@gmail.com> wrote:
Hi,

Please see what the below parameters are set to (On Standby).

max_standby_streaming_delay
max_standby_archive_delay

Do you have any activity recorded in pg_stat_activity on the standby ?



On Tue, Aug 18, 2026 at 2:06 PM Gyanuba Gyanuba <gyanubagyanuba@gmail.com> wrote:
Wait pending 

On Tue, Aug 18, 2026, 11:34 PM Simon Musila - Localhost <smusila1@gmail.com> wrote:

Hello,

I would appreciate some guidance on an issue we are experiencing with PostgreSQL streaming replication.

The standby appears to be receiving/streaming WALs and is up to date with the primary, but the WAL records are not being applied to the standby, resulting in an increasing replication/replay lag.

We have confirmed that:

  • WALs are being streamed from the primary to the standby.
  • The WAL receiver appears to be active.
  • The standby is receiving the WALs, but the replay/apply process is not keeping up.
  • pg_stat_database_conflicts shows no conflicts.
  • pg_is_wal_replay_paused() returns false.
  • recovery_min_apply_delay is 0.
  • hot_standby_feedback is disabled.

My questions are:

  1. What could cause WALs to be successfully streamed to the standby but not applied/replayed?
  2. What are the best steps to identify what the WAL replay process is waiting on?
  3. Is restarting the replication/standby the only practical solution, or is there a way to safely resume WAL replay without restarting?
  4. Are there specific PostgreSQL views, wait events, or system metrics that you would recommend checking to identify the bottleneck?

I would appreciate any guidance on troubleshooting this and identifying the underlying cause rather than simply restarting the replication whenever this occurs.

NB: This happens on a daily basis!

Thank you in advance.

Regards,
Musila



--
Regards,
Avinash Vallarapu
HexaCluster Corp.
FAQ