Re: BUG #9721: Fatal error on startup: no free slots in PMChildFlags array

On 25.03.2014 17:26, Tom Lane wrote:
> Daniel Hahler <postgresql(at)thequod(dot)de> writes:
>> On 25.03.2014 17:02, Tom Lane wrote:
>>> Oh. So actually, "failed to startup" was a complete misdescription, and
>>> what you should've said was "it appeared to be running normally, but I
>>> saw this burst of weird messages in the postmaster log, so I decided to
>>> restart it"?
>
>> Sorry, "failed to startup" was a misdescription, because I did not remember that OpenVZ suspends containers on reboots.
>> I should have said "crashed half an hour after being resumed".
>
>> I have not restarted PostgreSQL though, but started it when noticing that it wasn't running (anymore).
>
> Well, your log doesn't show any evidence to support the claim that
> Postgres wasn't running. We see some child processes having minor trouble
> while exiting, and one that failed while trying to start up; the latter
> being strong evidence that the postmaster was still working at least.

I've received the following error on 11:32:19 from a client app trying to connect:
OperationalError: could not connect to server: Connection refused
Is the server running on host "10.122.x.x" and accepting
TCP/IP connections on port x?

> So in short, you haven't given us a complete description of what happened.
> What led you to decide you needed to restart Postgres?

There was no Postgres process running in 'ps aux', and I only had to start, not restart it.

Relevant entries from the shell history (llog is for syslog/messages):

47 12:32 ps aux
48 12:33 llog
52 12:33 less ~l/daemon.log
55 12:34 less ~l/postgresql/postgresql-9.3-main.log
56 12:34 ps aux
64 12:38 service postgresql start

Regards,
Daniel.

--
http://daniel.hahler.de/