Horton

Re: Multiple hosts in connection string failed to failover in non-hot standby mode

Michael Paquier <michael@paquier.xyz>

View thread View on postgr.es
From: Michael Paquier <michael@paquier.xyz>
To: Tom Lane <tgl@sss.pgh.pa.us>
Cc: Justin Pryzby <pryzby@telsasoft.com>, Hubert Zhang <zhubert@vmware.com>, tsunakawa.takay@fujitsu.com, pgsql-hackers@postgresql.org, Andreas Seltenreich <seltenreich@gmx.de>, Bruce Momjian <bruce@momjian.us>
Date: 2021-06-07T06:43:42Z
Lists: pgsql-hackers 
On Sun, Jun 06, 2021 at 05:27:49PM -0400, Tom Lane wrote:
> It seems like nobody's terribly interested in figuring out why
> pg_GSS_have_cred_cache() is misbehaving on Windows.

I have been investigating that for a couple of hours in total, but
nothing to report yet.

> So I took
> a look at disabling GSSENC in these test cases to try to silence
> hamerkop's test failure that way.  Here's a proposed patch.
> It relies on setenv() being available, but I think that's fine
> because we link the ECPG test programs with libpgport.

No, that's not it.  The compilation of the tests happens when
triggering the tests as of ecpgcheck() in vcregress.pl so I think that
this is going to fail.  This requires at least the addition of a
reference to libpgport in ecpg_regression.proj, perhaps more.
--
Michael

Commits

  1. Clear conn->errorMessage at successful completion of PQconnectdb().

  2. Avoid ECPG test failures in some GSS-capable environments.

  3. Try next host after a "cannot connect now" failure.

  4. Uniformly identify the target host in libpq connection failure reports.

  5. Allow pg_regress.c wrappers to postprocess test result files.

  6. In libpq, always append new error messages to conn->errorMessage.