Re: 9.3.9 and pg_multixact corruption - Mailing list pgsql-hackers

From Alvaro Herrera
Subject Re: 9.3.9 and pg_multixact corruption
Date
Msg-id 20150910213932.GT2912@alvherre.pgsql
Whole thread Raw
In response to 9.3.9 and pg_multixact corruption  (Bernd Helmle <bernd@oopsware.de>)
List pgsql-hackers
Bernd Helmle wrote:
> A customer had a severe issue with a PostgreSQL 9.3.9/sparc64/Solaris 11
> instance.
> 

> 2015-09-08 11:40:59 CEST [27047] FATAL:  could not access status of
> transaction 1068235595
> 2015-09-08 11:40:59 CEST [27047] DETAIL:  Could not seek in file
> "pg_multixact/members/FFFF5FC4" to offset 4294950912: Invalid argument.
> 2015-09-08 11:40:59 CEST [27047] CONTEXT:  xlog redo create mxid 1068235595
> offset 2147483648 nmembers 2: 2896635220 (upd) 2896635510 (keysh) 

I think the math to compute segment number and byte offset of the member
might be bogus here.  The file names in pg_multixact/members is supposed
to go to 14078 (hex) in a 8kB-BLCKSZ build, and of course it goes a bit
higher in builds with smaller page sizes, but nowhere as high as
FFFF5FC4.  And the offset is way too close to 2^32 (exactly 16384 less
than 2^32, to be precise.)

-- 
Álvaro Herrera                http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services



pgsql-hackers by date:

Previous
From: Robert Haas
Date:
Subject: Re: Hooking at standard_join_search (Was: Re: Foreign join pushdown vs EvalPlanQual)
Next
From: Robert Haas
Date:
Subject: Re: Moving SS_finalize_plan processing to the end of planning