Hi all,

we've upgraded maui from 3.2 to 3.3 and noticed some less verbosity in
logs in some kind of jobs (failing ones). Seems that maui log does not
say job-node relation in some "failing" jobs. 


In previous versions we used :

grep 15025090 /var/log/maui.log|grep td

*(all our wn name start with "td", so I grep by td)

to find the node, and now it does not work:

# grep 15025090 /var/log/maui.log|grep  td
#
# grep 15025090 /var/log/maui.log
01/10 11:27:58 MPBSJobLoad(15025090,15025090.pbs03.pic.es,J,TaskList,0)
01/10 11:27:58 MReqCreate(15025090,SrcRQ,DstRQ,DoCreate)
01/10 11:27:58 INFO:     job '15025090' loaded:   1 atprd005    atprd  28800    
   Idle   0 1294655237   [NONE] [NONE] [NONE] >=      0 >=      0 [test] 
1294655273
01/10 11:28:18 INFO:     4 feasible tasks found for job 15025090:0 in partition 
DEFAULT (1 Needed)
01/10 11:28:18 INFO:     tasks located for job 15025090:  1 of 1 required (4 
feasible)
01/10 11:28:18 MJobStart(15025090)
01/10 11:28:18 MRMJobStart(15025090,Msg,SC)
01/10 11:28:18 MPBSJobStart(15025090,base,Msg,SC)
01/10 11:28:19 INFO:     job '15025090' successfully started
01/10 11:28:19 INFO:     starting job '15025090'
01/10 11:29:43 MJobDestroy(15025090)

and this is torque's node log:

$ grep 15025038 /var/spool/pbs/mom_logs/20110110
01/10/2011 11:20:40;0001;   pbs_mom;Svr;pbs_mom;LOG_ERROR::Permission denied 
(13) in TMakeTmpDir, Unable to make job transient directory: 
/home/tmp/15025038.pbs03.pic.es
01/10/2011 11:20:40;0001;   pbs_mom;Svr;pbs_mom;LOG_ERROR::start_exec, cannot 
create temp dir '/home/tmp/15025038.pbs03.pic.es'
01/10/2011 11:20:40;0008;   pbs_mom;Req;send_sisters;sending ABORT to sisters 
for job 15025038.pbs03.pic.es
01/10/2011 11:20:40;0080;   pbs_mom;Job;15025038.pbs03.pic.es;obit sent to 
server


so, as you can see, the job has been dispatched to the node, but maui
is not logging that.

I've tried to increase verbosity until 4, but then, logs are so big
that they waste more than 15GB of disk space (too much).

I've looked to configurable parameters (showconfig) and I don't see
if this has become a configurable feature.

So, anyone knows who to re-enable destination-host info for failing
jobs in maui logs? 


TIA,
Arnau
_______________________________________________
mauiusers mailing list
[email protected]
http://www.supercluster.org/mailman/listinfo/mauiusers

Reply via email to