GSI Forum
GSI Helmholtzzentrum für Schwerionenforschung

Home » PANDA » PANDA - Computing » Grid and Infrastructure » test of submit script at gridka
test of submit script at gridka [message #12047] Fri, 17 June 2011 13:26 Go to next message
donghee is currently offline  donghee
Messages: 385
Registered: January 2009
Location: Germnay
first-grade participant
From: *dip.t-dialin.net
Dear panda specially gridka family,

Recently I have got gridka account, of course I want to use it effectively.

Now I'm testing analysis scripts with tdr11 data (psi3770=r202).
In attachment, I put my trial jdl, macro code, bin.sh scripts
and a validation code.
I tried already submitting with 10 subjobs, but I couldn't find output error message in my output directory.

Quote:


[pgdb2.gla.ac.uk:3307] /panda/user/d/dkang/ > masterJob 1168153 -printsite
Jun 17 13:07:35 info Checking the masterjob of 1168153
Jun 17 13:07:36 info The job 1168153 is in status: DONE
It has the following subjobs:
Subjobs in ERROR_V (ikp642.ikp.kfa-juelich.de): 1
Subjobs in ERROR_V (kvit14.KVI.nl): 7
Subjobs in ERROR_V (smigrid02.smi.oeaw.ac.at): 2

In total, there are 10 subjobs




Probably, I made some bug in my submit scripts, or macro code when define the input files and strings etc...
Could you teach me where I care and fix to do correctly.

Thank you in advance,
Donghee


Re: test of submit script at gridka [message #12048 is a reply to message #12047] Fri, 17 June 2011 13:35 Go to previous messageGo to next message
StefanoSpataro is currently offline  StefanoSpataro
Messages: 2736
Registered: June 2005
Location: Torino
first-grade participant

From: *to.infn.it
Hi,
if the jobs are failing, it is not storing the output on the SE.
You can put the output in some folder typing:

registerOutput #jobid


and you can get the #jobid of the subjobs typing:

masterJob 1168153 -printid
Re: test of submit script at gridka [message #12049 is a reply to message #12048] Fri, 17 June 2011 14:21 Go to previous messageGo to next message
donghee is currently offline  donghee
Messages: 385
Registered: January 2009
Location: Germnay
first-grade participant
From: *dip.t-dialin.net
Thank you stefano,

I can read output error!
So problem is that I have to give input file name separately for each spliting file using jdl and bin.sh


You can simply imagine that if you want to access points.root and pid.root files according produced and given individual name, which are produced by pbarprod, then can prepare all corresponding name or string by means of combining jdl with bin.sh.

I think that I define correct name in my two scripts but I cannot read it.
Quote:



Error in <TFile::TFile>: file /panda/user/p/pbarprod/tdr11/output/sig/run202/10/evt_points_stt_r202_10 .root does not exist
[/QUOTE]

I'm login to gridka as a personal user.
Do you have any other idea?

Best wishes,
Donghee

Re: test of submit script at gridka [message #12050 is a reply to message #12048] Fri, 17 June 2011 14:23 Go to previous messageGo to next message
donghee is currently offline  donghee
Messages: 385
Registered: January 2009
Location: Germnay
first-grade participant
From: *dip.t-dialin.net
Why allows only 60 subjobs for personal user in gridka?
Re: test of submit script at gridka [message #12116 is a reply to message #12047] Wed, 22 June 2011 12:10 Go to previous messageGo to next message
donghee is currently offline  donghee
Messages: 385
Registered: January 2009
Location: Germnay
first-grade participant
From: *kph.uni-mainz.de
Dear all gridka user,

If I analyze gridka, for instance, the data on
/panda/user/p/pbarprod/tdr11/output/sig/run701/1/...


Can I use data collection list still at analysis level?
Quote:


find -c output-run701 /panda/user/p/pbarprod/tdr11/output/sig/run701/ evt_p*.root


After preparing list,
then in the jdl and bin script one need to indicate which files belong to which submit job. For output I can do that, but the name of input files are differ and at least two input file(points.root and pid.root) are needed in the analysis script.
How machine can handle different name of input file name?

Please teach me how it works?

Best wishes,
Donghee
Re: test of submit script at gridka [message #12118 is a reply to message #12116] Wed, 22 June 2011 12:26 Go to previous messageGo to next message
StefanoSpataro is currently offline  StefanoSpataro
Messages: 2736
Registered: June 2005
Location: Torino
first-grade participant

From: *to.infn.it
Hi,
first of all, gridka is "Grid Computing Centre Karlsruhe"... maybe you mean "grid".
The way could be, in your shell script, to move the filename in something you can use, i.e.:

mv evt_points_*.root evt_points_stt_evt.root


I am using this for my analysis.
Re: test of submit script at gridka [message #12121 is a reply to message #12116] Wed, 22 June 2011 13:04 Go to previous messageGo to next message
donghee is currently offline  donghee
Messages: 385
Registered: January 2009
Location: Germnay
first-grade participant
From: *kph.uni-mainz.de
Hi all,

Definitely, I cannot do such way in gridka (Grid Computing Centre Karlsruhe) in my mind.
Quote:


mv evt_points_*.root evt_points_stt_evt.root



Because points file and pid files belongs to user for "pbarprod".
and I think that I don't have a great privilege or super power.

If I change the name of file, doesn't affect to next other users?
Probably in gridka with Logical File system one can do safely without worring about it. Because the file is not really change of physical name.
If that is really true... it really strange, but really interesting feature.

Thank you for your teaching!

Donghee
Re: test of submit script at gridka [message #12122 is a reply to message #12121] Wed, 22 June 2011 13:51 Go to previous message
StefanoSpataro is currently offline  StefanoSpataro
Messages: 2736
Registered: June 2005
Location: Torino
first-grade participant

From: *to.infn.it
When youa re analysing a collection, the system is copyng the fiels from the collection into some temp directory, therefore you can modify them as you wish.. the original names will persist on the original place.
However, you can use %alien_counter to propagate such number also itno your analysis macros. If you look at the scripts we are sending you can see how this works... you have onky to use it as input parameter of your analysis macro.
Previous Topic: copy analysis output to local
Next Topic: spliting the job via collection
Goto Forum:
  


Current Time: Tue Dec 03 02:51:15 CET 2024

Total time taken to generate the page: 0.00814 seconds