[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Condor-users] efficiency question



Thanks for the replies.


On Thu, May 26, 2011 at 7:05 AM, Matthew Farrellee <matt@xxxxxxxxxx> wrote:
For counts of idle jobs, you can use condor_status -sched, understanding it is lazily updated.

Best,


matt


On 05/10/2011 12:14 AM, Rita wrote:
What is the best way to get idle jobs?

I always do condor_q -global -direct schedd (I am using quill and I want
to bypass it). I understand that can be costly but was wondering if
there are tricks with the collector.

condor_status -claimed is nice but there is no way to get idle jobs.




On Tue, Apr 12, 2011 at 7:27 PM, Rita <rmorgan466@xxxxxxxxx
<mailto:rmorgan466@xxxxxxxxx>> wrote:

   Thanks everyone for your replies.

   I am going to wait for 7.6.2 and upgrade. I decided to skip 7.4.x
   series.



   On Tue, Apr 12, 2011 at 3:05 PM, Matthew Farrellee <matt@xxxxxxxxxx
   <mailto:matt@xxxxxxxxxx>> wrote:

       On 04/08/2011 08:04 PM, Rita wrote:

           3) The most recent negotiation cycle and how many were
           negotiated
           Parse a log file?


       http://spinningmatt.wordpress.com/2010/12/05/update-on-negotiation-cycle-statistics/

       In 7.6.0 you'll hopefully get all the stats you care about from
       condor_status -negotiator -long. Let us know if some are missing.

       Before 7.6.0 you'll have to parse logs,

       http://spinningmatt.wordpress.com/2010/03/29/negotiation-cycle-statistics/



           4) Top 10 users who are requesting the most memory and their
           jobs are
           running now
           Check for request_memory?


       This would be a condor_q -format and some post-processing. An
       RFE may be to add sum of ImageSize/RequestMemory/something to
       Submitter ads, visible from condor_status -submitter -long.



           5) Top 10 users who are requesting the most CPU and their
           jobs are
           running now


       Same as (4).



           Basically, it boils down to using condor_status and
           condor_q. I haven't
           found a way to query the negotiator deamon for statistics.


       condor_status -negotiator -long



           BTW, I plan on running these commands on a 5 second basis.


       That is very frequent. You should expect condor_q that processes
       all jobs that often to perturb your system.


       Best,


       matt




   --
   --- Get your facts first, then you can distort them as you please.--




--
--- Get your facts first, then you can distort them as you please.--




--
--- Get your facts first, then you can distort them as you please.--