Hi Jamie,
Now I tried to use the old condor_gridmanager fron 6.7.13 - and it
worked to the extend that it could submit jobs to NG-sites, however,
it failed in retrieving the output - here is the GridManager log:
1/25 11:24:00 [1701] resource brenta.ijs.si is now up
1/25 11:24:00 [1701] (4.0) doEvaluateState called: gmState
GM_RECOVER_QUERY, condorState 1
1/25 11:24:01 [1701] (4.0) doEvaluateState called: gmState
GM_RECOVER_QUERY, condorState 1
1/25 11:24:03 [1701] (4.0) doEvaluateState called: gmState
GM_EXIT_INFO, condorState 1
1/25 11:24:03 [1701] ERROR "Bad NORDUGRID_EXIT_INFO Result" at line
4960 in file gahp-client.C
Should I switch back completely to 6.7.13 ? - changelog claims that
6.7.14 can submit to ng and unicore...
Cheers,
Michael
Michael Gronager wrote:
Hi Jamie,
I'm currently testing the nordugrid and lcg submission functionality
with condor-6.7.14 ... I see that nordugrid_gahp is now part of the
distribution ;-)
I can run LCG jobs, however, output fails due to :
Globus error 155: the job manager could not stage out a file
Code 2 Subcode 155
Nordugrid jobs will not execute:
the gridmanager can't handle this job type
Code 0 Subcode 0
Do I still have to replace the gridmanager ? - And if not - when will
the newest gridmanager ship with condor (6.7.15?)
Cheers,
Michael
Btw - below I have listed my submission script to nordugrid:
universe=grid
grid_type=nordugrid
grid_resource=nordugrid morpheus.dcgc.dk
executable=test.sh
WhenToTransferOutput=ON_EXIT
dir=.
output=$(dir)/out.xrsl.$(Cluster).$(Process)
error=$(dir)/err.xrsl.$(Cluster).$(Process)
log=$(dir)/log.xrsl.$(Cluster).$(Process)
queue 1