Re: Autovacuum of pg_database

Dne 5.5.2016 v 17:43 Alvaro Herrera napsal(a):
> Ondřej Světlík wrote:
>> Dne 5.5.2016 v 17:19 Ondřej Světlík napsal(a):
>>> Dne 5.5.2016 v 17:16 Ondřej Světlík napsal(a):
>>>> Hello again,
>>>>
>>>> we have another strange problem with autovacuum. The process goes well
>>>> until it reaches the table pg_database (in some databases, not all),
>>>> then it loops over it forever  (until I run vacuum full analyze on that
>>>> table, simple vacuum doesn't help):
>>>>
>>>> 2016-05-05 17:13:33 CEST  34818LOG:  automatic vacuum of table
>>>> "somedatabase.pg_catalog.pg_database": index scans: 0
>>>>      pages: 0 removed, 233 remain
>>>>      tuples: 0 removed, 5817 remain, 0 are dead but not yet removable
>>>>      buffer usage: 87 hits, 0 misses, 0 dirtied
>>
>> And now the same goes for pg_authid. Sorry for spamming, I hope this is the
>> last one.
>
> These are all shared catalogs.  There are others, so you may still see
> more.  We got another report for pg_database
> https://www.postgresql.org/message-id/A9D40BB7-CFD6-46AF-A0A1-249F04878A2A%40amazon.com
> so I suppose there really is a bug.  I don't know what's going on there.
>

Let me know if I can help trace the cause.

With regards

Ondřej