[GE users] SGE 6.2 queue cpu external job limits

templedf dan.templeton at sun.com
Fri May 1 19:54:14 BST 2009


If you can reliably count the number of external jobs on a node, you can 
report that number via a custom complex and a load sensor.  You can then 
set an RQS that says "limit hosts {*} to slots=$myslots-$external" to 
debit each machine for the external jobs running on it.  Note that 
you'll also have to monitor or set a myslots complex for each machine 
instead of using slots in the equation, because the value you're setting 
for slots can't be self-referential.  Do keep in mind, though, that the 
complexes are only updated periodically, every 40 seconds by default, so 
if your external jobs are short-lived, you have no hope of keeping up 
with them.

Daniel

emjga wrote:
>
> Folks
>
> I am very new to SGE so forgive me for some stupid Questions.
>
> We have a 20 odd nodes most of which have 8 cpu's
> Therefore we have a queue that allow up to 8 jobs to run (slots) per 
> machine
>
> The problem is that the nodes are all so used by external jobs from 
> outside SGE
> Is it possible to get SGE to check how many CPU's are been used by 
> external jobs
> say 3 and then to only allow 5 SGE jobs to be submitted to that 
> queue/host
>
> At the moment we have load checking on but that will either allow jobs 
> or disallow jobs (np_load_av = 1.75)
> It does not limit the number of job slots.
>
> Any suggestions.
> Thanks
>
> *Matthew Garrett*
> Registered in England and Wales No.811900          
> Registered Office 33 Cavendish Square, London W1G 0PW
> This e-mail and any attachments are intended only for the person or entity
> to whom it is addressed and may contain confidential or privileged
> information.  If you are not the addressee, any disclosure, reproduction,
> copying, distribution, or use of this communication is strictly prohibited.
> If you are not the intended recipient or person responsible for delivering
> this message to the named addressee, please notify us immediately and delete
> this e-mail.
> It is the responsibility of the addressee to scan this email and any
> attachments for computer viruses or other defects.  The sender does not
> accept liability for any loss or damage of any nature, however caused,
> which may result directly or indirectly from this email or any file attached.
>         
>

------------------------------------------------------
http://gridengine.sunsource.net/ds/viewMessage.do?dsForumId=38&dsMessageId=188202

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



More information about the gridengine-users mailing list