[Bioclusters] inconsistent output from simple BLAST job
landman at scalableinformatics.com
Mon Nov 22 16:44:45 EST 2004
Could you tell us what your NFS mount options are for the node? Could
be a case of async writes and bad flushing of output.
peter_webb at agilent.com wrote:
>This may be a BLAST question, a Linux question, or a networking question...
>We're running a small cluster (Supermicro dual Xeon, RedHat Enterprise, SGE, gigE switch interconnect). We observed occasional inconsistent output from BLAST jobs. Inconsistent means that a few hits in the file are missing, not necessarily at the beginning or end of the file.
>I've tracked it down to writing output from the process running on the nodes to a network file system instead of the local disk. This happens even when I log into the node and run the BLAST job directly, so I don't believe it has anything to do with SGE. The problem is intermittent; it will run the test job with identical output hundreds of times, then fail a few times, then run happily again.
>It looks for all the world as if one line of output (using -m 8 in BLAST) has vanished into the ether, without anybody complaining about it.
>Any ideas as to what is up, or suggestions to narrow things down further, would be very welcome.
>Bioclusters maillist - Bioclusters at bioinformatics.org
Joseph Landman, Ph.D
Founder and CEO
Scalable Informatics LLC,
email: landman at scalableinformatics.com
web : http://www.scalableinformatics.com
phone: +1 734 612 4615
More information about the Bioclusters