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: 2021-06-08T03:24:31Z
Lists: pgsql-hackers

Attachments

On Mon, Jun 07, 2021 at 10:38:03AM -0400, Tom Lane wrote:
> Hmm.  We do include "-lpgcommon -lpgport" when building the ecpg test
> programs on Unix, so I'd assumed that the MSVC scripts did the same.
> Is there a good reason not to make them do so?

I was looking at that this morning, and yes we need to add more
references here.  Actually, adding only libpgport.lib allows the
compilation and the tests to work, but I agree to add also
libpgcommon.lib so as we don't fall into the same compilation trap
again in the future.

Now, I also see that using pgwin32_setenv() instead of
src/port/setenv.c causes cl to be confused once we update
ecpg_regression.proj because it cannot find setenv().  Bringing the
question, why is it necessary to have both setenv.c and
pgwin32_setenv() on HEAD?  setenv.c should be enough once you have the
fallback implementation of putenv() available.

Attached is the patch I am finishing with, that also brings all this
stuff closer to what I did in 12 and 13 for hamerkop.  The failing
test is passing for me now with MSVC and GSSAPI builds.

Thoughts?
--
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.