Fw: CRS job notification

• Previous message: Eun-Joo Kim: "some plots from MRST0"
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]

----- Original Message -----
From: "Tony W. Chan" <tony@voyager.rhic.bnl.gov>
To: <mcbreen@bnl.gov>; <ejkim@bnl.gov>; <chujo@bnl.gov>; <momchil@bnl.gov>;
<burt@bnl.gov>; <belt@bnl.gov>; <didenko@bnl.gov>; <jeromel@bnl.gov>
Cc: <throwe@bnl.gov>; <tomw@bnl.gov>
Sent: Monday, February 10, 2003 5:27 PM
Subject: CRS job notification


> Hi,                            Feb. 10, 2003
>
> After lots of debugging, I put into production
> a enhanced CRS job notification protocol. There
> are now 4 new categories of job failures that
> will notify whoever has been designated in the
> JDF file (i.e., bramreco, phnxreco, phobreco,
> starreco). In principle, it will go to either
> rcrsuser1 or rcrsuser2 (when available). The
> default (if database is missing that info)
> is rcrsuser1. If you get a message with one
> of the following key words, it means the
> faulty job has been cleaned up, and you should
> resubmit it.
>
> This type of notification will only occur twice
> a day per experiment, since I use a cron job to
> look for these discrepancies. I limit it to twice
> a day to avoid overloading the database with
> requests.
>
> Please pass this message to the person(s) submitting
> jobs for your respective experiment to notify them
> of this update. Thanks.
>
> Cheers,
>
> Tony
>
> *****************************************************
>
> 1) transfer failed --> job in database, but not in
>                        the assigned node. One or
>                        more of the input and/or
>                        JDF file(s) did not transfer
>                        successfully to CRS node
>
> 2) msg. failed     --> job status says "executing",
>                        but "ps" shows job not running.
>                        Assume failed to "transfer out"
>                        due to message passing failure.
>
> 3) db failed (1)   --> job in node (and # of "running"
>                        jobs in db > 0), but partial db
>                        info is missing. Assume db update
>                        failed.
>
> 4) db failed (2)   --> job in node (and # of "running"
>                        jobs in db = 0), but partial db
>                        info is missing. Assume db update
>                        failed.
>
>
>

• Next message: Eun-Joo Kim: "drift velocity"
• Previous message: Eun-Joo Kim: "some plots from MRST0"
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]

This archive was generated by hypermail 2.1.5 : Tue Feb 11 2003 - 08:56:02 EST