[GE users] qlicserver seems to double subtract

olesen Mark.Olesen at emconTechnologies.com
Tue Nov 3 12:02:12 GMT 2009


    [ The following text is in the "utf-8" character set. ]
    [ Your display is set for the "ISO-8859-10" character set.  ]
    [ Some characters may be displayed incorrectly. ]

On Mon, 2009-11-02 at 10:38 -0800, biscisking wrote:
> I take it back.  The abaqus definition was also present in my
> qlicserver.config file, and type="job" was not present.  If it appears
> in both qlicserver.config and qlicserver, what happens?

The values in qlicserver.config have precedence. If different resources
are specified (eg, 'abaqus' in the config file and 'nastran' in the
code), they will be merged together. If you always want to ignore the
hard-coded default values, either add type="overwrite" to the
<resources ..> tag (see
http://olesenm.github.com/flex-grid/config-file.html) or else simply
remove the hard-coded values (eg, comment it out, rename the entry from
'-defaultConfig' to '-defaultConfigIGNORE' or something).

> In any case, a colleague did some testing for me of the type="job"
> setting.  He's running SGE6.1, so he's got consumable=YES and is
> scaling his license requests.  This is what he found:

If he has consumable=YES in his GridEngine configuration and type="job"
for qlicserver the results are going to be something really strange.
Don't do that - it might calculate something, but nothing useful.

> If I remove the type=job it looks like this, which is correct (what I would expect to see):
>
> resource      total  limit extern intern  wait  free
> --------      -----  ----- ------ ------  ----  ----
> abaqus           25      .      .     21    16     4
>
> A qstat ?F then shows gc:abaqus=4, which is what I would expect to
> see. Interestingly, I do not see and messages in qstat ?j <jobid> that
> mentions why the job is queued but it is definitely queued because
> there are not enough licenses it seems. I get the same message about
> the PE but the values don?t seem to map over. If I have a 16 core
> ?r?unning and an 8 core queued it says:
>
> cannot run in PE "8ppn" because it only offers 16 slots

Yup. The message is a bit confusing. But sometimes there is some mention
about the PE not providing enough slots, since (I believe) it has been
prefiltered by the '-l' request.

> it at least seems to be working correctly, despite what the scheduler
> reports.

Good. How about the original problem on 6.2u3?

Between the various posts, I'm now a bit lost about the current status.

/mark

This e-mail message and any attachments may contain legally privileged, confidential or proprietary Information, or information otherwise protected by law of EMCON Technologies, its affiliates, or third parties. This notice serves as marking of its "Confidential" status as defined in any confidentiality agreements concerning the sender and recipient. If you are not the intended recipient(s), or the employee or agent responsible for delivery of this message to the intended recipient(s), you are hereby notified that any dissemination, distribution or copying of this e-mail message is strictly prohibited. 
If you have received this message in error, please immediately notify the sender and delete this e-mail message from your computer.

------------------------------------------------------
http://gridengine.sunsource.net/ds/viewMessage.do?dsForumId=38&dsMessageId=224814

To unsubscribe from this discussion, e-mail: [users-unsubscribe at gridengine.sunsource.net].



More information about the gridengine-users mailing list