Dan schrieb am 09.08.2010 um 18:04 (-0400):
> Hello,
> I just wanted to provide some closure on this "issue", we found out
> that there was a PIX firewall between the Perl app server and the
> Oracle database, and once we removed this, the problem did go away.
Good to see the culprit revealed (a
;ll aim to fill our row cache with around 10 per go. */
> > >> /* Using 10 means any 'runt' packets will have less impact. */
> > >>
> > >>
> > >> int txfr_size = 10 * 1460; /* desired transfer/cache size */
> > >>
> > >>
> > >
and or type of the constant my throw the bug
as well. Might just be one of those majic munbers that pad the results just
enough to make the query run fine.
Hope this helps a little
John Scoles
> Date: Fri, 9 Jul 2010 17:47:24 -0400
> Subject: Re: "sql*net message from client&qu
txfr_size / est_width; /* (maybe 1 or 0) */
>>
>>
>>
>> /* To ensure good performance with large rows (near or larger */
>> /* than our target transfer size) we set a minimum cache size. */
>>
>>
>> if (cache_rows < 6) /* is cache a
= (cache_rows > 0) ? 6 : 4;
>}
> if (cache_rows > 32767) /* keep within Oracle's limits */
> cache_rows = 32767;
>
>
>
>
>
> so the above might be buggering you up if your packet size is no longer
> 1460?
>
>
>
> This might be it.
>
>
}
if (cache_rows > 32767) /* keep within Oracle's limits */
cache_rows = 32767;
so the above might be buggering you up if your packet size is no longer 1460?
This might be it.
Cheers
John Scoles
> Date: Fri, 9 Jul 2010 10:15:03 -0400
> Subject: Re: "sql*net messa
Hello,
Apologies, that was DBI 1.607, not .67!
with RowCacheSize '0', it hangs on the same block, with '1', it seems to
go through ok as with '-1'.
As far as changes go, that is part of the mystery, haven't been able to
find any changes around the time when this started to happen. Yes,
The $dbh->{RowCacheSize} = -1; is funny as you are telling the box to only
use 1 byte for RowCach
Try it with '0' and see what happens?
On Thu, Jul 8, 2010 at 7:20 PM, John Scoles wrote:
> Hmm DBI 0.67 that is old and the fact you get the same error with the
> latest DBI and DBD::Oracle would
Hmm DBI 0.67 that is old and the fact you get the same error with the latest
DBI and DBD::Oracle would lead me to think there may be something wrong
internally in your DB or its storage medium, corrupt block perhaps or bad
sector??
The code differences between 1.61 and 0.67 are vast as would be t
Hello,
We do not have any quotas implemented on our Oracle instance. The simple
select works fine for that hanging block on sqlplus, or if i only execute it
for that block in the perl script instead of iterating through them all. I
just tested running for that same block 185 times instead of all
there are various quotas in Oracle that could stop a query. However,
they do tend to kill the guilty session rather than just hang like
that. If your simple select ran multiple times for the same 10k block,
would it still hang?
Cheers
Martin
On 08/07/2010 17:07, Howard, Chris wrote:
If it
If it always stops at the same place, it makes me think
of a resource problem, something like a quota?
I don't remember of there are select quotas.
Does it do the same if run as sysdba or some other
well-endowed database user?
> -Original Message-
> From: Dan [mailto:dkele...@gmail.com
12 matches
Mail list logo