News & Announcements
Memory Use Reporting for Biowulf Jobs (Biowulf)
Date: 29 May 2012 11:05:42From: steven fellini (sfellini@NIH.GOV)
Three weeks ago, node memory limits began to be enforced on
Biowulf compiute nodes (http://helix.nih.gov/News/index.php?381).
This results in the automatic deletion of jobs exceeding those
limits.
To better allow users to know when their jobs have been deleted
due to exceeded memory limits, two ways of reporting have been
implemented, effective immediately.
1. Memory use is now reported in the standard output file (".o")
for each job. If the memory used exceeds the memory limit
of the node, the following message appears:
============================================================================
|| NOTE: this job was likely deleted by the batch system due to exceeding
available memory. ||
============================================================================
2. The 'jobcheck' command has been enhanced and can now report
all jobs which have exited with non-zero exit status for a
particular time period.
See the Biowulf User Guide for details on using 'jobcheck':
http://biowulf.nih.gov/user_guide.html#jobcheck

