[GE users] Array jobs and Usage Accounting.

isakrejda isakrejda at lbl.gov
Tue Mar 17 16:41:52 GMT 2009


    [ 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,

On 3/17/09 9:26 AM, jlb wrote:

On Tue, 17 Mar 2009 at 8:46am, isakrejda wrote



I am running 6.0u11 on my production cluster and we noticed that users
who run array jobs are grossly
overcharged for the running time of their jobs. Looks to me like each
instance carries the load of all
running instances.



This is a longstanding issue (see
<http://gridengine.sunsource.net/issues/show_bug.cgi?id=2298><http://gridengine.sunsource.net/issues/show_bug.cgi?id=2298>) and,
unfortunately, one that doesn't appear any closer to a resolution than it
was almost a year ago.  To avoid it, we moved from fairshare to functional
share, which doesn't have this issue.


I know, I myself dragged it into this time a few times. I vaguely remember being convinced
to upgrade to 6.0u11 to overcome this issue (and it was not solved) and that's why I with hope
check new releases. This time (6.2u1 ) the issue flipped (underreporting vs over-reporting before)...

Since you moved from fair share to functional share, you can probably answer this question:
Is it possible to establish with functional shares
a system that resembles fair share? I have several groups that share the cluster and with fair share
we kept them reasonably happy about sharing. I have not looked into functional shares yet...

Thank You,

iwona





More information about the gridengine-users mailing list