[torqueusers] tracejob output truncated for large jobs

Marcus R. Epperson mrepper at sandia.gov
Thu Sep 15 12:29:45 MDT 2005


The output from tracejob was getting truncated for large jobs (observed 
at 256-nodes).  Specifically, the default 4k buffer that tracejob uses 
is not enough to hold the exec_host= and Resource_List.neednodes= fields 
in the accounting logs for jobs of this size (at least not on our machines).

The attached patch is a quick fix that increases the buffer to 16k.  It 
is a patch against 1.1.0p6, but should be easy enough to make the same 
change for 1.2.0p6 (or others).

Just thought I'd send this in case anyone else runs into the same problem.

-Marcus
-------------- next part --------------
A non-text attachment was scrubbed...
Name: tracejob-1.1.0p6-bufsize.patch
Type: text/x-patch
Size: 507 bytes
Desc: not available
Url : http://www.supercluster.org/pipermail/torqueusers/attachments/20050915/f77dd4dd/tracejob-1.1.0p6-bufsize.bin


More information about the torqueusers mailing list