On Sat, Jul 25, 2015 at 10:59:27AM -0400, Tom Lane wrote:
> http://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=axolotl&dt=2015-07-24%2020%3A29%3A18
>
> What evidently happened there is that "pg_ctl start" gave up waiting for
> the postmaster to start too soon. The postmaster log appears to contain
>
> LOG: database system was shut down at 2015-07-24 16:45:40 EDT
> FATAL: the database system is starting up
> LOG: MultiXact member wraparound protections are now enabled
> LOG: database system is ready to accept connections
>
> which indicates that it did successfully come up, but not till after one
> "PQping" probe from pg_ctl, which was rejected with "still starting up".
For the record, that was not a PQping() probe. Since "connection to database
failed: FATAL: the database system is starting up" appeared in the buildfarm
log, it was pg_upgrade attempting to connect after its pg_ctl had given up.
> Meanwhile we've got this log output from pg_ctl:
>
> waiting for server to start........ stopped waiting
> pg_ctl: could not start server
> Examine the log output.
>
> Counting the dots indicates that pg_ctl gave up after precisely 5 seconds.
Thanks for improving that.
