Tom Lane wrote:
> Anyway, it looks to me like the next interesting thing that happens
> after those messages is pgstat_init(), which will try to open a UDP
> port on address "localhost". If there's something munged about your
> DNS setup, perhaps that could fail (though I'm not sure why it would
> fail without any error message).
There runs no DNS server on this box. I use the DNS server from my ISP.
epauli# cat /etc/resolv.conf
search adsl.cistron.nl
nameserver 62.216.31.50
nameserver 62.216.31.60
> You might try turning off the stats collector altogether --- set
> "stats_start_collector" off in postgresql.conf (and make sure
> that stats_command_string, stats_row_level, stats_block_level are
> not turned on). Does it get any further if you do that?
No. The error messages are the same with that options off.
> Also, I'm not aware of any reason that the postmaster would exit
> without printing an error message --- unless it crashed completely,
> and in that case it should leave a core file. Do you see any core
> file? If so, can you get a stack trace from it?
Yes. In /usr/local/pgsql there is a postgres.core
I've put a strace on my webspace.
http://epauli.dyndns.org/strace
Very trange is that the server will start without any problem when i
copy a 3 weeks old backup to /usr/local/pgsql/
Later today i will look further to the problem and make a strace from
pg_ctl with the setup from 3 weeks ago.
Thanks for the help!
--
Edwin Pauli