I have not seen this, so cannot comment, but when I am trying to simulate i do not see issues.
One thing to note,
It seems your wal is on nfs mount , can you rule out any nfs errors if it is nfs.
On Fri, Jun 4, 2021, 6:24 PM Atul Kumar <akumar14871@xxxxxxxxx> wrote:
Hi,
archive_command is 'cp %p /nfslogs/wal/%f'
and no, we are not removing anything from pg_xlog directory.
once old WAL files of pg_xlog directory are archived in
'/nfslogs/wal/' directory then these WAL files are getting generated
with the same name in pg_xlog directory.
my query is Why is this happening ?
please help me with your suggestions.
Regards.
On 6/4/21, Jehan-Guillaume de Rorthais <jgdr@xxxxxxxxxx> wrote:
> On Fri, 4 Jun 2021 15:39:30 +0530
> Atul Kumar <akumar14871@xxxxxxxxx> wrote:
>
>> HI,
>>
>> We have a centos 6 enviornment where postgres 9.6 is running on it.
>>
>> We have strange behavior of WAL files of pg_xlog directory
>>
>> As we have set archive_command to archive WAL files at different
>> location and the archive_command is working fine.
>>
>> So strange behavior is :
>>
>> We have a WAL file say for example "00000001000036CD000000E2" of
>> 01.06.2021 (1st June 2021) that is getting archive successfully at the
>> archive location and once it is archived, this file with same name
>> (00000001000036CD000000E2) is getting generated with the latest
>> timestamp (as today is 04.06.2021).and all old WAL files are behaving
>> in same manner.
>
> What is you archive_command?
>
> I'm not sure I understand correctly, but keep in mind your
> archive_command must be "read only". Do not remove the WAL file after
> archiving
> it.
>
> Regards,
>
