Horton

Re: hung backends stuck in spinlock heavy endless loop

Merlin Moncure <mmoncure@gmail.com>

View thread View on postgr.es
From: Merlin Moncure <mmoncure@gmail.com>
To: Heikki Linnakangas <hlinnakangas@vmware.com>
Cc: Peter Geoghegan <pg@heroku.com>, Andres Freund <andres@2ndquadrant.com>, Tom Lane <tgl@sss.pgh.pa.us>, PostgreSQL-development <pgsql-hackers@postgresql.org>
Date: 2015-01-15T14:02:28Z
Lists: pgsql-hackers 
On Thu, Jan 15, 2015 at 6:04 AM, Heikki Linnakangas
<hlinnakangas@vmware.com> wrote:
> On 01/15/2015 03:23 AM, Peter Geoghegan wrote:
>>
>> So now the question is: how did that inconsistency arise? It didn't
>> necessarily arise at the time of the (presumed) split of block 2 to
>> create 9. It could be that the opaque area was changed by something
>> else, some time later. I'll investigate more.
>
>
> Merlin, could you re-run the test with a WAL archive (if you don't have one
> already), and then run pg_xlogdump, filtering it to show only the changes to
> the index? That should show us how the index got to be the way it is. Also,
> if you could post a copy of the raw relation file for pg_class_oid_index; I
> assume it's not too large.
>
> Something like:
>
> pg_xlogdump -r Btree -p walarchive/ -s 0/20035D0 | grep 11917
>
> 11917 is the relfilenode of pg_class_oid_index on a freshly initdb'd
> cluster. In case it's not the same on your system, you can use oid2name to
> find it out.

I'm on it.  Will try this first, then patch removal.

Question: Coming in this morning I did an immediate restart and logged
into the database and queried pg_class via index.   Everything was
fine, and the leftright verify returns nothing.  How did it repair
itself without a reindex?

merlin

Commits

  1. Improve hash_create's API for selecting simple-binary-key hash functions.

  2. Fix WAL-logging of B-tree "unlink halfdead page" operation.

  3. Fix backup-block numbering in redo of b-tree split.

  4. Cleanup of new b-tree page deletion code.

  5. Fix bug in the new B-tree incomplete-split code.

  6. Fix bug in new B-tree page deletion code.

  7. Fix rm_desc routine of b-tree page delete records.

  8. Make the handling of interrupted B-tree page splits more robust.

  9. Fix race condition in B-tree page deletion.