[Swift-devel] odd scheduling behavior in swift-0.9

[Glen Hocky]

Hi Everyone,
I had been having problems from communicado to ranger so this morning I 
tried running on ranger from ranger as the submit host with the 
following sites file. This time I had >600 jobs go active while only 3-4 
jobs in the queue were active i.e. only 48-64 coasters should have been 
active.

>   <pool handle="ranger">
>     <execution provider="coaster" 
> url="gatekeeper.ranger.tacc.teragrid.org" jobManager="gt2:gt2:SGE"/>
>     <!--<filesystem provider="coaster" 
> url="gt2://gatekeeper.ranger.tacc.teragrid.org"/>-->
>     <!-- <gridftp url="gsiftp://gridftp.ranger.tacc.teragrid.org:2811" 
> /> -->
>     <gridftp url="local://localhost" />
>     <profile namespace="env" 
> key="SWIFT_JOBDIR_PATH">/tmp/hockyg/jobdir</profile>
>     <profile namespace="env" 
> key="SWIFT_JOBDIR_PATH">/tmp/hockyg/jobdir</profile>
>     <profile namespace="globus" key="project">TG-MCB080099N</profile>
>     <profile namespace="globus" key="coastersPerNode">16</profile>
>     <profile namespace="globus" 
> key="coasterWorkerMaxwalltime">05:00:00</profile>
>     <profile namespace="globus" key="maxwalltime">60</profile>
>     <profile namespace="karajan" key="initialScore">50</profile>
>     <profile namespace="karajan" key="jobThrottle">10</profile>
>     <workdirectory>/share/home/01021/hockyg/swiftwork</workdirectory>
>   </pool>
Log file is transfered to here:
ci home: 
/home/hockyg/public_html/swift_logs/ranger_from_ranger/oops-20090427-0845-pc8celi7.log
web:       
/http://www.ci.uchicago.edu/~hockyg/swift_logs/ranger_from_ranger/oops-20090427-0845-pc8celi7.log

coaster logs are in:
ci home: /home/hockyg/oops/swift/coaster_logs/ranger