Stephen Childs wrote:
> We have 250 analysis jobs that are doing nothing but pulling data over
> the network. As far as I can see they are making little progress
> themselves and are overloading the network causing other jobs to fail.
>
> Is there any way of seeing how much progress the jobs are making? For
> example, if I knew where on the WN disk they were writing the received
> data to, I could keep an eye on the file growing. How much data does
> each job need to pull before it can do useful work?
if they're RFIO jobs you won't see a file growing on the node. But the
stdout in the job's sandbox should show something like
AthenaEventLoopMgr INFO ===>>> start processing event #417486, run
#105012 8055 events processed so far <<<===
AthenaEventLoopMgr INFO ===>>> done processing event #417486, run
#105012 8056 events processed so far <<<===
showing each processed event as it's streamed.
File stagers should be putting the file somewhere like eg
$TMPDIR/condorg_NMe29253/pilot3/Panda_Pilot_29281_1244644372/PandaJob_1010746396_1244644373/AOD.065523._00410.pool.root.1
John
--
Dr John Bland, Systems Administrator
Room 220, Oliver Lodge
Particle Physics Group, University of Liverpool
Mail: [log in to unmask]
Tel : 0151 794 2911
"I canna change the laws of physics, Captain!"
|