Re: [Condor-users] Help with Java program...on Condor run Cluster

["Andrey Kaliazin" <A.Kaliazin@xxxxxxxxxxx>]

Hello Roy

We have one user here who run jobs similar to yours on our mixed
Windows/Linux
Condor pool. Here is his weka.sub file -

#
# weka.sub 
#
universe	= java
output	= weka.output
error		= weka.error
log		= weka.log

should_transfer_files 	= YES
when_to_transfer_output = ON_EXIT
TRANSFER_FILES  = ALWAYS
executable=/home/daltons/weka-3-4-10/weka.jar
jar_files=/home/daltons/weka-3-4-10/weka.jar
arguments=weka.classifiers.trees.J48 -t weather.arff
transfer_input_files=weather.arff
queue
#
# weka.sub end 
#

You should of course change  the paths to your files accordingly
Mark the proper universe for this job - java - not vanilla

cheers

Andrey Kaliazin
Senior Server Engineer (cluster computing)
Information Systems Aston (ISA)
Aston University, Aston Triangle,
Birmingham, B4 7ET 
Tel: 0121 204 3465 
 

> -----Original Message-----
> From: condor-users-bounces@xxxxxxxxxxx 
> [mailto:condor-users-bounces@xxxxxxxxxxx] On Behalf Of roy 
> hill (IGER-WP)
> Sent: Monday, March 26, 2007 2:24 PM
> To: Condor-Users Mail List
> Subject: [Condor-users] Help with Java program...on Condor run Cluster
> 
> Dear All, 
> 
> Not sure if you can help started trying to run Weka on a 
> Linux Condor Cluster. 
> 
> Somewhere along the route of submitting a job from the submit 
> machine to the node "condor_exec.exe" is getting into the 
> command for the node to execute. any ideas where I should 
> look for this. I'm confused but as a beginner probably not 
> surprising anyhow any help would be appreciated.
> 
> 
> Below is the command I'm using as a test. 
>   
> "weka_condor weka.classifiers.trees.J48 -t /data/wekadata/iris.arff" 
> 
> Best regards, 
> Roy. 
> 
> ------------------------------ 
> 
> Log dump. 
> 
> 000 (496.000.000) 03/26 09:54:57 Job submitted from host: 
> <xxx.xxx.xxx.xxx:32773> 
> ... 
> 001 (496.000.000) 03/26 09:55:01 Job executing on host: 
> <169.254.56.2:32769> 
> ... 
> 007 (496.000.000) 03/26 09:55:01 Shadow exception! 
>         Error from starter on node1.igerclust1: Failed to 
> execute 'java condor_exec.exe weka.classifiers.trees.J48 -t 
> /data/wekadata/iris.arff': No such file or directory
> 
>         0  -  Run Bytes Sent By Job 
>         0  -  Run Bytes Received By Job 
> ... 
> 001 (496.000.000) 03/26 09:55:03 Job executing on host: 
> <169.254.56.2:32769> 
> ... 
> 007 (496.000.000) 03/26 09:55:03 Shadow exception! 
>         Error from starter on node1.igerclust1: Failed to 
> execute 'java condor_exec.exe weka.classifiers.trees.J48 -t 
> /data/wekadata/iris.arff': No such file or directory
> 
>         0  -  Run Bytes Sent By Job 
>         0  -  Run Bytes Received By Job 
> 
> --------------------------------------------------------- 
> # File weka_condor .. 
> ########################## 
> #   weka_condor script 
> #   Set up SDF file for Condor and run it 
> ########################## 
> # 
> #echo "Job started "`date` 
> export WEKAHOME=/usr/bbsrc_common/weka-3-4-10 
> export WEKADATA=/data/wekadata 
> export CLASSPATH=$CLASSPATH:$WEKAHOME/weka.jar 
> allargs="$*" 
> echo "Weka submitted "`date` 
> # run Condor with the basic SDF + arguments param 
> condor_submit -a "arguments=$allargs" 
> /usr/bbsrc_common/EMBOSS/bin/weka_SDF 
> #echo "Job done "`date` 
> 
> 
> 
> --------------------------------------------- 
> # File weka_SDF 
> 
> ########################## 
> #   Example of running weka 
> #   use condor_submit -a "arguments= ..." thisfile 
> ########################## 
> executable=java 
> error=weka_$(Cluster).error 
> output=weka_$(Cluster).stdout 
> log=weka_$(Cluster).log 
> getenv=True 
> copy_to_spool=False 
> universe=vanilla 
> notification=never 
> transfer_executable=False 
> should_transfer_files=YES 
> when_to_transfer_output = ON_EXIT 
> queue 
> 
> 
> 
> 
> 
> 
> 
>