Alvaro Herrera <alvherre@commandprompt.com> writes:
> Manuel Sugawara wrote:
>
> Hi Manuel,
>
>> The funny thing is that there was no open transactions, even after
>> restarting the cluster the same message was logged. Today, the
>> database stopped working as expected:
>>
>> ERROR: database is shut down to avoid wraparound data loss in database "postgres"
>> HINT: Stop the postmaster and use a standalone backend to VACUUM in "postgres"
>
> I suggest you look for temp tables that have not been reclaimed. We've
> had a couple of reports where leftover temp tables have stopped the
> frozen-xid counter from advancing. (They would have a very old
> relfrozenxid.)
In each database executed:
select relname, age(relfrozenxid) from pg_class where relnamespace = '10406'::oid;
(note that 10406 is the oid of the pg_temp_1 namespace) none of them
showed temp tables, Is this the correct way?, Any other idea?.
Regards,
Manuel.