Hi,
I analyze a strange case with the startup process on replica spend almost all time in waiting state:
"postgres: startup process recovering 0000000100004A3E000000BF waiting"
As a result recovery seriously lagged behind master (in range of hours).
Replica have hot_standby=on and almost always have some long running (in range of minutes-ten minutes) select queries.
I expected to find some not granted locks in pg_locks on the standby server, but select * from pg_locks where not granted; show no results.
Is there any way to look into the startup process and see what it waiting for (less invasive than gdb).
PS: there are no DDL routinely run on the master (so there no need accessexclusivelock on standby to apply ddl).
Kind Regards,
Maksym
--