[GE users] job not killed when h_cpu limit reached

Reuti reuti at staff.uni-marburg.de
Sat Aug 21 16:05:00 BST 2004


    [ The following text is in the "ISO-8859-1" character set. ]
    [ Your display is set for the "ISO-8859-10" character set.  ]
    [ Some special characters may be displayed incorrectly. ]

Hi again,

one thing more I also want to throw in, is the (strange?) setting of the h_cpu 
limit by SGE. According to the man page of queue_conf it's multiplied by the 
number of slots requested by the job. In fact, I just played around a bit and 
found, it's multiplied by the number of slots taken from one machine instead. I 
mean:

Having h_cpu=20 and a request for four slots will give by a 1-1-1-1 
distribution to four machines on each machine a hard limit of 20, hence 80 in 
total.

When you get 2-2 instead, to each slot the applied limit is 40 and your job 
will consume a total amount of 160.

Is this behavior intended, or can anyone explain why it is this way? (Linux on 
AMD64)

Cheers - Reuti

---------------------------------------------------------------------
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