[GE users] wrong records in accounting file

Sebastian Stark stark at tuebingen.mpg.de
Thu Nov 22 15:44:10 GMT 2007


(resending to users@ because I got no response from dev@)

Hi,

did anybody ever find too high values for cpu usage in accounting  
records? I am talking of values ~10000 times higher than expected/ 
possible.

We are trying to find a pattern for jobs with wrong values like "all  
qlogin jobs" or "all array jobs" but we did not find anything yet, so  
we thought it is about time to ask if anybody else encountered these  
erroneous entries.

This is SGE 6.1 checked out from TV61_BRANCH in July.


example of a BAD record:

all.q:node145:kyb:seeger:splin29-6-45:349936:sge: 
0:1191600130:1191600137:1191600297:100:137:160:0:0:0.000000:0:0:0:0:698: 
0:0:0.000000:0:0:0:0:43:1:defaultproject:AGBS:NONE: 
1:0:18446744233.889999:173740430.807745:0.000000:-u seeger -l  
arch=lx26-amd64,h_cpu=720,h_vmem=1G,nicetohave=FALSE:0.000000:NONE: 
19279872.000000

(cpu=18446744233.889999)


example of a GOOD record:

all.q:node147:ebio:sschmidt:betweens.nr.no_lethals:340987:sge: 
0:1191509436:1191509894:1191510503:0:0:609:495:115:0.000000:0:0:0:0:9668 
786:0:0:0.000000:0:0:0:0:31579:13519:defaultproject:ABT2:NONE: 
1:432:610.000000:37.349377:0.000000:-U ABT2 -u sschmidt -l  
h_vmem=4G,nicetohave=FALSE:0.000000:NONE:249806848.000000

(cpu=610.000000)


Sebastian


-- 
Max Planck Institute for Biological Cybernetics
Homepage: http://www.kyb.tue.mpg.de/~stark
vCard:    http://www.kyb.tue.mpg.de/bs/people/stark/sstark.vcf

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe at gridengine.sunsource.net
For additional commands, e-mail: users-help at gridengine.sunsource.net




More information about the gridengine-users mailing list