On Tue, Oct 22, 2019 at 4:51 PM Pavel Stehule <pavel.stehule@gmail.com> wrote:
>
> út 22. 10. 2019 v 5:09 odesílatel Amit Kapila <amit.kapila16@gmail.com> napsal:
>>
>>
>> CountOtherDBBackends is called from other places as well, so I don't
>> think it is advisable to change the sleep time in that function.
>> Also, I don't want to add a parameter for it. I think you have a
>> point that in some cases we might end up sleeping for 100ms when we
>> could do with less sleeping time, but I think it is true to some
>> extent today as well. I think we can anyway change it in the future
>> if there is a problem with the sleep timing, but for now, I think we
>> can just call CountOtherDBBackends after sending SIGTERM and call it
>> good. You might want to add a futuristic note in the code.
>>
>
> ok.
>
> I removed sleeping from TerminateOtherDBBackends().
>
> If you want to change any logic there, please, do it without any hesitations. Maybe I don't see, what you think.
>
Fair enough, I will see if I need to change anything. In the
meantime, can you look into thread related to CountDBSubscriptions[1]?
I think the problem reported there will be more serious after your
patch, so it is better if we can fix it before this patch.
[1] - https://www.postgresql.org/message-id/CAA4eK1%2BqhLkCYG2oy9xug9ur_j%3DG2wQNRYAyd%2B-kZfZ1z42pLw%40mail.gmail.com
--
With Regards,
Amit Kapila.
EnterpriseDB: http://www.enterprisedb.com