On Fri, 2024-07-19 at 05:57 +0000, Domen Šetar wrote:
> The number of wal files on my postgresql server is rising, because it seams
> that one wal is corrupted. Postgrsql is running normaly. I see this in postgresql
> log file:
>
> 2024-07-19 07:44:12 CEST [2205]: [32288-1] user=,db=,app=,client= DETAIL: The failed archive command was: test ! -f
/var/lib/pgsql/ArchiveDir/000000010000044E0000009D&& cp pg_wal/000000010000044E0000009D
/var/lib/pgsql/ArchiveDir/000000010000044E0000009D
>
> Usualy helped if I deleted wal in ArchiveDir directory. But not this time. Wal is
> copied again from pg_wal to ArchiveDir directory and error message continues.
> What can I do to solve this problem? Is pg_resetwal solution fort his problem? If it is, how to use it?
Don't listen to any advice to run "pg_resetwal".
Only consider switching to the standby if your primary crashes because the disk is full.
You need to determine the cause of the problem.
1. All error messages from "archive_command" end up in the log file.
Search for those, they may help you determine the cause.
2. Is there a file /var/lib/pgsql/ArchiveDir/000000010000044E0000009D ?
If yes, delete it, and the problem should be solved.
3. If there is no such file, it must be the "cp" command that is
failing. In that case, you should definitely see an error message
about that in the log file. Likely causes:
- the permissions are not right (try by running the "cp" command as
user "postgres" manually)
- the target directory does not exist
- the target directory is full
Yours,
Laurenz Albe
