Re: BUG #13999: pg_upgrade failing with FATAL: unrecognized configuration parameter "unix_socket_directory"

From: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
To: rsimeon(at)ebms(dot)com
Cc: pgsql-bugs(at)postgresql(dot)org
Subject: Re: BUG #13999: pg_upgrade failing with FATAL: unrecognized configuration parameter "unix_socket_directory"
Date: 2016-03-04 15:48:53
Message-ID: 6322.1457106533@sss.pgh.pa.us
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-bugs

rsimeon(at)ebms(dot)com writes:
> When running pg_upgrade it appears that the call to the old pg_ctl binary
> has wrong parameters set. Can I upload a screenshot for you?

If it's complaining about "unix_socket_directory", the reason must be that
pg_upgrade is guessing wrong about whether the old server expects that
parameter to be named unix_socket_directory or unix_socket_directories.

Am I right in guessing that you are using Red Hat's packaging rather than
straight community sources? In the community version, that parameter
change happened in 9.3, but I recall that Red Hat back-ported the
unix_socket_directories feature into an earlier version in order to have
a reasonable way of coping with SELinux restrictions. So what this is
sounding like is that they need to carry a patch that adjusts pg_upgrade's
start_postmaster() function to have the right Red-Hat-specific cutoff
version for which way to spell the parameter.

In short, you need to file this issue with Red Hat not the PG community.

regards, tom lane

In response to

Browse pgsql-bugs by date

  From Date Subject
Next Message itparanoia 2016-03-04 16:00:48 BUG #14000: initdb installs in 9.4 by default
Previous Message Alvaro Herrera 2016-03-04 15:36:32 Re: BUG #13844: Logical decoding bug with subxact + row locking