RE: Inability to reliably monitor long-running parallel queries

From: Mark W. Farnham <mwf_at_rsiz.com>
Date: Fri, 2 Nov 2012 17:48:02 -0400
Message-ID: <0c4a01cdb943$bc380df0$34a829d0$_at_rsiz.com>

Kafka would be proud. Well, okay, he wouldn't actually be proud, but he might experience some intellectual amusement that reporting "done with error" for a query that is still running and which may in fact complete without error is defined as "not a bug." Damn, I don't see a tuxedo disguising this one as a feature!

mwf

-----Original Message-----
From: oracle-l-bounce_at_freelists.org [mailto:oracle-l-bounce_at_freelists.org] On Behalf Of Adric Norris
Sent: Friday, November 02, 2012 3:15 PM
To: oracle-l_at_freelists.org
Subject: Re: Inability to reliably monitor long-running parallel queries

I've noticed this as well, although I hadn't had an opportunity to troubleshoot this issue. I'm glad that you've cleared up the mystery, but definitely share your disappointment with the "not a bug" designation. The feature would be *far* more useful without this limitation. On Fri, Nov 2, 2012 at 11:36 AM, Tornblad, John <JTornblad_at_emdeon.com>wrote:

> It turns out that the issue for a parallel query is MMON is watching
> *all* of the parallel servers / slaves and if *any* are idle for more
> than an internally hard-coded 30 minutes... the finger of death is
> applied and the monitoring of the query is terminated... even though
> other parallel servers are active on the same query. I am not sure if
> the query coordinator is included, it seems very likely the QC is idle
> for big queries for more than 30 minutes (easily).
>
> Just wondering if anyone else has run into this issue, I cannot be the
> only customer who has noticed this. Despite "not a bug" declarations
> it seems like this is a bug / misfeature / unwelcome aggressiveness
> from MMON.
>

--
http://www.freelists.org/webpage/oracle-l


--
http://www.freelists.org/webpage/oracle-l

Received on Fri Nov 02 2012 - 22:48:02 CET