Stephen writes:
> * Adam Sjøgren (asjo@xxxxxxxxxxxx) wrote:
>>
>> We have 60 processes (workers) running on different machines accessing
>> the database, that all grab jobs from a queue and update rows in a table
>> after doing some calculations (which vary in time from <1s to perhaps a
>> minute, many of them fast).
>>
>> Sometimes new database logins slow down, from usually taking <0.05s to
>> taking minutes. This is for psql as a normal user using Kerberos, for
>> psql as the postgres superuser, for the web-application logging into the
>> database, for everything.
>
> When in doubt, blame DNS.
I'd love to! However I don't see any DNS outages on our local network
correlating with whether I run 60 workers or 5 workers.
> Alternatively, in your case, the issue might be the KDC taking forever
> to issue a ticket for the service.
If that was the cause, logging in as the 'postgres' superuser (not using
Kerberos) locally on the server should be fast regardless, right?
> (though you might check if you have log_hostnames on..).
It's off:
$ grep hostname /etc/postgresql/11/main/postgresql.conf
#log_hostname = off
Thanks for replying!
Best regards,
Adam
--
"Probabilistic algorithms don't appeal to me. (This Adam Sjøgren
is a question of aesthetics, not practicality.) So asjo@xxxxxxxxxxxx
later, I figured out how to remove the probability
and turn it into a deterministic algorithm."
