BUG #13525: Database crashed with several error messages "Could not open file "pg_multixact/members/"

From: kumarsathishms(at)gmail(dot)com
To: pgsql-bugs(at)postgresql(dot)org
Subject: BUG #13525: Database crashed with several error messages "Could not open file "pg_multixact/members/"
Date: 2015-07-29 12:13:50
Message-ID: 20150729121350.9130.28427@wrigleys.postgresql.org
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-bugs

The following bug has been logged on the website:

Bug reference: 13525
Logged by: Could not open file "pg_multixact/members/
Email address: kumarsathishms(at)gmail(dot)com
PostgreSQL version: 9.3.5
Operating system: Ubuntu 12.04
Description:

Hi Team,

Recently our postgresql db crashed giving the error message Could not open
file "pg_multixact/members/<xxxx>" for the second time.

Architecture -
Master(read+write) and slave(read-only) database - Hot standby.
DB size : 159GB
Note: Both master + slave had same problem.
What we did to to recover from the situation -

Action-1: Drop and restore the DB on same server using PITR. Result : Still
the same error.
Action-2: Restore base and data using PITR. Result:Poor performance in DB
with loggings "WARNING: oldest multixact is far in the past"
Action-3: Remove Posgresql-9.3.5 and install postgresql-9.3.9 restore the
complete DB using pg_restore and did not use PITR. Worked well.

After lot of read-ups, the files were wrapped around and lost over time. So,
How can we avoid such situations in future?. Can this be controlled with
configurations to avoid the wraparounds?.

Any help on this topic would be really great to solve the issue. Let me know
if some more details are needed to analyze the cause.

Thanks a lot !

Responses

Browse pgsql-bugs by date

  From Date Subject
Next Message Stuart Bishop 2015-07-29 13:18:48 postgres --describe-config missing metadata
Previous Message Andres Freund 2015-07-29 11:13:26 Re: BUG #13523: Unexplained deadlocks (possible race condition)