Hi Tom,
thanks for the hint.
In the background there was an extreme old monitoring script running.
But I didn't know about it. Now it is fixed.
Regards,
Frank
Am 08.11.2018 um 15:48 schrieb Tom Lane:
> =?utf-8?q?PG_Bug_reporting_form?= <noreply@postgresql.org> writes:
>> After some times, many errors are logged about not existing fields.
>> < 2018-11-08 14:21:59.820 CET >ERROR: column "procpid" does not exist at
>> character 132
>> < 2018-11-08 14:21:59.820 CET >STATEMENT: SELECT datname, datid, usename,
>> client_addr, '' AS state, COALESCE(ROUND(EXTRACT(epoch FROM
>> now()-query_start)),0) AS seconds,
>> procpid as pid, regexp_replace(current_query, E'[\n\r\u2028]+', '
>> ', 'g' ) AS current_query FROM pg_stat_activity WHERE (query_start IS NOT
>> NULL AND current_query NOT LIKE '<IDLE>%') ORDER BY query_start, procpid
>> DESC;
>
> The procpid and current_query columns disappeared from pg_stat_activity in
> Postgres 9.2 (or more accurately, they were renamed because they changed
> meaning a bit). Apparently you're using some very old monitoring software
> that hasn't been updated to know about that.
>
> regards, tom lane
>
