Re: pg_upgrade: when the --old-bindir requires "additional" $libdir/ plugins?

On 04/18/2018 07:07 AM, Pavel Raiskup wrote:
> Hi all,
>
> with a huge delay, I've seen this question [1].
>
> . and it seems like the hstore.so was somewhat intimately integrated into
> OP's database so the '/usr/bin/pg_dump --schema-only --binary-upgrade
> --format=custom' called through 'pg_upgrade' failed with:
>
> pg_dump: [archiver (db)] query failed: ERROR: could not access file
> "$libdir/hstore": No such file or directory

From [1] it seems there is confusion in the system about where $libdir
is. Did anyone run pg_config to try to track this down?

>
> Which means that the dump from old datadir, with old server (without
> hstore.so packaged) failed. But playing with hstore.so a bit, the upgrade
> always worked smoothly for me even without the "old" hstore.so

I am not following the above. If you played with hstore.so how could it
not be around? Or does this have something to do with 'old'?

>
> Googling around, I only found that there were some incompatible changes
> historically in hstore.so causing pg_upgrade failure; but this is
> different issue -- in the issue I talk about now the module was not found
> at all, even though it is usually even not searched for.
>
> So the question is what circumstances could cause this issue in
> particular? Is it expected in general that, to make the pg_upgrade run
> smooth, all the "loaded" server modules are built against and available
> for the --old-bindir server? Don't we have some reproducer for this
> kind of issues?
>
> [1] https://ask.fedoraproject.org/en/question/69316/postgresql-doesnt-work-after-upgrade-to-f22/
> [2] https://bugzilla.redhat.com/show_bug.cgi?id=1557490
>
> Thank for hints,
> Pavel
>
>
>
>
>

--
Adrian Klaver
adrian(dot)klaver(at)aklaver(dot)com