elein <elein@varlena.com> writes:
> From: Tom Lane <tgl ( at ) sss ( dot ) pgh ( dot ) pa ( dot ) us>
>> Hmmm ... are you using integer timestamps by any chance?
>> It looks to me like TimestampTzPlusMilliseconds() would overflow
>> for such a large timeout.
> Is this a bug or should there be documentation that tells us the max
> of statement timeout should be (what?)
It's a bug. You're supposed to get an error if you set the value too
large:
regression=# set statement_timeout TO '30d';
ERROR: invalid value for parameter "statement_timeout": "30d"
HINT: Value exceeds integer range.
Patch is in CVS already, please check if it fixes it for you.
regards, tom lane