Rainer Tammer <pgsql@spg.schulergroup.com> writes:
> It did run the server with auto vacuum disabled for ~ 24h - no server
> shutdown.
> After re-enabling auto vacuum the server dies in less then 9 hours:
This is just speculation, but it fits the reported facts: somewhere
you've got something that tries to limit the resources consumed by
a background process, and when it decides the limit is exhausted
then it sends SIGINT. Without autovacuum waking up every so often,
the time for PG-as-a-whole to hit the hypothetical limit would be
much longer.
> Would it be worth adding additional code before every signal to trace
> the source ID and the target PID as well as the source/target process name?
I will be *mightily* astonished if the signal is coming from
within PG. The more so since it appears to be happening while
the instance is entirely idle.
regards, tom lane