On Wed, Jan 23, 2008 at 11:22:06AM -0800, elein wrote:
> Running 8.3RC1
>
> I have an sql script where one or more create index statements
> raise a statement timeout message. The statement timeout is
> set to 1d.
>
> The script runs in ~3 hours including the timeout messages.
>
> The script does this:
>
> BEGIN;
> create table temp.xxx ...
> insert into temp.xxx ...
> COMMIT;
> create index one on temp.xxx(col1);
> create index one on temp.xxx(col2);
> create index one on temp.xxx(col3);
> create index one on temp.xxx(col4);
> BEGIN;
> drop table public.xxx;
> alter table temp.xxx set schema public;
> COMMIT;
>
> I have sprinkled the script with show statement_timeout
> and it always returns 1D.
>
> I suspect either a problem with counting statement time
> or another error is using the statement timeout message.
>
> Ideas?
>
> --elein
> [EMAIL PROTECTED]
>
>>From: Tom Lane <tgl ( at ) sss ( dot ) pgh ( dot ) pa ( dot ) us>
>>To: elein <elein ( at ) varlena ( dot ) com>
>>Subject: Re: Statement Timeout Message Incorrect
>>Date: Wed, 23 Jan 2008 15:52:35 -0500
>>elein <elein ( at ) varlena ( dot ) com> writes:
>>> Running 8.3RC1
>>> I have an sql script where one or more create index statements
>>> raise a statement timeout message. The statement timeout is
>>> set to 1d.
>>
>>Hmmm ... are you using integer timestamps by any chance?
>>It looks to me like TimestampTzPlusMilliseconds() would overflow
>>for such a large timeout.
>>
>>> The script runs in ~3 hours including the timeout messages.
>>
>>If I'm doing the math correctly, 1 day would overflow to about
>>8 minutes in microseconds, so if you're seeing individual statements
>>time out after more than that then there's something else going on.
>>
>> regards, tom lane
>>
Is this a bug or should there be documentation that tells us the max
of statement timeout should be (what?)
We did build with integer timestamps on a 64 bit machine with RC1.
Sorry for the weird format. I'm on the digest and couldn't wait to
reply.
--elein
[EMAIL PROTECTED]
---------------------------(end of broadcast)---------------------------
TIP 1: if posting/reading through Usenet, please send an appropriate
subscribe-nomail command to [EMAIL PROTECTED] so that your
message can get through to the mailing list cleanly
