On Thu, May 28, 2015 at 8:03 AM, Robert Haas <robertmhaas@gmail.com> wrote:
>> Steve, is there any chance we can get your pg_controldata output and a
>> list of all the files in pg_clog?
>
> Err, make that pg_multixact/members, which I assume is at issue here.
> You didn't show us the DETAIL line from this message, which would
> presumably clarify:
>
> FATAL: could not access status of transaction 1
And I'm still wrong, probably. The new code in 9.4.2 cares about
being able to look at an *offsets* file to find the corresponding
member offset. So most likely it is an offsets file that is missing
here. The question is, how are we ending up with an offsets file that
is referenced by the control file but not actually present on disk?
It seems like it would be good to compare the pg_controldata output to
what is actually present in pg_multixact/offsets (hopefully that's the
right directory, now that I'm on my third try) and try to understand
what is going on here.
--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company