What should I disable? Corrupt index sounds like a possible case, but
how do I fix this?
EXPLAIN ANALYZE
select * from cubesocialnetwork.tmp_autenticazionesocial where
idautenticazionesocial = 1622
"Index Scan using tmpautenticazione on tmp_autenticazionesocial
(cost=0.00..8.27 rows=1 width=530) (actual time=0.078..0.081 rows=1
loops=1)"
" Index Cond: (idautenticazionesocial = 1622)"
"Total runtime: 0.128 ms"
On 07/10/2014 00:15, Jim Nasby wrote:
> On 10/5/14, 10:06 AM, Andy Colson wrote:
>> On 10/05/2014 10:00 AM, Adrian Klaver wrote:
>>> On 10/05/2014 07:37 AM, Tim Mickelson wrote:
>>>> Sorry about that, the precise version is: "PostgreSQL 9.1.9 on
>>>> x86_64-unknown-linux-gnu, compiled by gcc (GCC) 4.4.7 20120313 (Red
>>>> Hat
>>>> 4.4.7-3), 64-bit"
>>>
>>> Well 9.1 is at .14 now, so on general principles it would be a good
>>> idea to upgrade. That being said I do not see anything in the
>>> release notes from .10 to .14 that applies. Though to be truthful I
>>> did not read every line. Before upgrading you could try what Andy
>>> suggested which is to REINDEX(tmpautenticazione). See here for the
>>> REINDEX caveats, and a way to INDEX CONCURRENTLY:
>>>
>>> http://www.postgresql.org/docs/9.1/interactive/sql-reindex.html
>>>
>>>>
>>>>
>>
>> I thought .11 sounded like a good candidate. Especially the part:
>>
>> allowing tuples to escape freezing, causing those rows to become
>> invisible once 2^31 transactions have elapsed
> Those rows should then be invisible to the FK checks as well, so I
> don't think that's it.
>
> My guess is also on a corrupted index. A quick test would be to
> disable index scans and try the delete again (use EXPLAIN ANALYZE to
> make sure the delete is using a sequential scan). If that doesn't
> work, post EXPLAIN ANALYZE output for the commands you're running that
> generate this error.
>