[Brahms-dev-l] CRS batch and crash

From: flemming videbaek <videbaek@rcf.rhic.bnl.gov>
Date: Tue Apr 06 2004 - 18:03:08 EDT
As you may have seen RCF have been working on an upgrade of the CRS system, basing the distribution of jobs and
scheduling on Condor. You can find some information on the RCF web-pages. We i.e. mainly I have been testing this out and have used it for most of the preliminary local reconstruction jobs from this years run (~1000 sequences) . I general I found it be more stable (less jobs lost due to missing files) than the old system. The job description files are close to the old but a few differences appear, and the control system i.e. job submission, and status checking is different. It does have a good GUI interface to find out why jobs may have failed.
Due to this crash has been updated to 1.6.1 , installed on /opt/brahms/new..
A working jsf-creating an submission script can be found in bramreco/run04/auau/63/ltr along with a better (I hope/know) script for TPC local tracking. For the time present the new system can only run from rcrsuser3 .. I suggest we soon switch over to this system exclusively, in fact it is the only system that will be available on the 4 new crs nodes to be installed in the next couple of weeks.
Before it can be use a howto file has to be updated.

regards
    Flemming

----------------------------------------------------------------
Flemming Videbaek
Physics Department
Brookhaven National Laboratory

e-mail: videbaek@bnl.gov
phone: 631-344-4106



_______________________________________________
Brahms-dev-l mailing list
Brahms-dev-l@lists.bnl.gov
http://lists.bnl.gov/mailman/listinfo/brahms-dev-l
Received on Tue Apr 6 17:56:55 2004

This archive was generated by hypermail 2.1.8 : Tue Apr 06 2004 - 17:57:19 EDT