Re: [Condor-users] [Fwd: Condor and Weka] (short version)

[Srinivasan <srinivasan.manikantan@xxxxxxxxxxxx>]

hi,
"transfer_input_files" parameter just transfers your jar file to the execute nodes but doesnt make it available in CLASSPATH
put the jars also in "jar_files" parameter this should do the magic

regards,
Srinivasan.



On Tue, 2006-02-28 at 12:14 +0100, Gino wrote:
> 
> -------- Original Message --------
> From: 	- Tue Feb 28 11:54:34 2006
> X-Mozilla-Status: 	0001
> X-Mozilla-Status2: 	00800000
> Message-ID: 	<44042BE6.2050309@xxxxxxxx>
> Date: 	Tue, 28 Feb 2006 11:54:30 +0100
> From: 	Gino <ginofiliberti2@xxxxxxxx>
> User-Agent: 	Debian Thunderbird 1.0.7 (X11/20051017)
> X-Accept-Language: 	en-us, en
> MIME-Version: 	1.0
> To: 	Condor-Users Mail List <condor-users@xxxxxxxxxxx>
> Subject: 	Condor and Weka
> References: 	<200602250003.k1P03oNH028140@xxxxxxxxxxxxxxxxx>
> In-Reply-To: 	<200602250003.k1P03oNH028140@xxxxxxxxxxxxxxxxx>
> Content-Type: 	text/plain; charset=ISO-8859-1; format=flowed
> Content-Transfer-Encoding: 	7bit
> 
> 
> 
> Good morning to everybody.
> I'm writing here in order to find a solution to a Condor problem, that 
> I'm encountering. It's still for my degree thesis...
> I searched the web and the manual to find out a solution but maybe my 
> situation it's a bit too complex and the problem can resides in many 
> aspects.
> I implemented a condor pool with 9 personal computers:
> 
> 7 linux machines for execute-submit,
> 1 linux machine for the central manager,
> 1 windows 2000 machine for execute-submit.
> 
> With this situaion I tried the "Hello world program" in the standard 
> universe, than an octave-process, and now I HAVE TO implement (for 
> further expansion of the pool) a weka-based solution.
> Weka is a program written in java...
> Now I'm in trouble because weka doesn't work.
> I tried differnet solutions.
> 
> #First tentative
> Universe    = java
> Executable    = weka/classifiers/bayes/BayesNet.class
> Output        = Weka1-1bis.out
> Error        = Weka1-1bis.error
> Log        = Weka1-1bis.log
> should_transfer_files = YES
> when_to_transfer_output = ON_EXIT
> transfer_input_files= alarm_10000_shuffled_01.arff,weka.jar
> Arguments    = weka.classifiers.bayes.BayesNet.class -D -B 
> -t=alarm_10000_shuffled_01.arff -Q 
> weka.classifiers.bayes.net.search.local.K2mo -- -P 100 -Z 0 -M -N -O 0 
> -E weka.classifiers.bayes.net.estimate.SimpleEstimator -- -A 1.0
> Queue
> #First Error Output
> java.lang.ClassNotFoundException: weka.classifiers.bayes.BayesNet.class
>    at java.net.URLClassLoader$1.run(Unknown Source)
>    at java.security.AccessController.doPrivileged(Native Method)
>    at java.net.URLClassLoader.findClass(Unknown Source)
>    at java.lang.ClassLoader.loadClass(Unknown Source)
>    at sun.misc.Launcher$AppClassLoader.loadClass(Unknown Source)
>    at java.lang.ClassLoader.loadClass(Unknown Source)
>    at java.lang.ClassLoader.loadClassInternal(Unknown Source)
>    at java.lang.Class.forName0(Native Method)
>    at java.lang.Class.forName(Unknown Source)
>    at CondorJavaWrapper.main(CondorJavaWrapper.java:101)
> 
> #Second tentative
> #Prima chiamata della vm per alarm_10000_shuffled_01.arff
> Universe    = java
> Executable    = weka/classifiers/bayes/BayesNet.class
> Output        = Weka1-1ter.out
> Error        = Weka1-1ter.error
> Log        = Weka1-1ter.log
> should_transfer_files = YES
> when_to_transfer_output = ON_EXIT
> transfer_input_files= alarm_10000_shuffled_01.arff,weka.jar
> Arguments    = BayesNet.class -D -B -t=alarm_10000_shuffled_01.arff -Q 
> weka.classifiers.bayes.net.search.local.K2mo -- -P 100 -Z 0 -M -N -O 0 
> -E weka.classifiers.bayes.net.estimate.SimpleEstimator -- -A 1.0
> java_vm_args = -Xmx1200m
> Queue
> #Second Error Output
> java.lang.ClassNotFoundException: BayesNet.class
>    at java.net.URLClassLoader$1.run(Unknown Source)
>    at java.security.AccessController.doPrivileged(Native Method)
>    at java.net.URLClassLoader.findClass(Unknown Source)
>    at java.lang.ClassLoader.loadClass(Unknown Source)
>    at sun.misc.Launcher$AppClassLoader.loadClass(Unknown Source)
>    at java.lang.ClassLoader.loadClass(Unknown Source)
>    at java.lang.ClassLoader.loadClassInternal(Unknown Source)
>    at java.lang.Class.forName0(Native Method)
>    at java.lang.Class.forName(Unknown Source)
>    at CondorJavaWrapper.main(CondorJavaWrapper.java:101)
> 
> #Third tentative
> #Prima chiamata della vm per alarm_10000_shuffled_01.arff
> Universe    = java
> Executable    = weka/classifiers/bayes/BayesNet.class
> Output        = Weka1-1.out
> Error        = Weka1-1.error
> Log        = Weka1-1.log
> requirements    = "LINUX"
> should_transfer_files = YES
> when_to_transfer_output = ON_EXIT
> transfer_input_files= alarm_10000_shuffled_01.arff,weka.jar
> Arguments    = ..weka.classifiers.bayes.BayesNet -D -B 
> -t=alarm_10000_shuffled_01.arff -Q 
> weka.classifiers.bayes.net.search.local.K2mo -- -P 100 -Z 0 -M -N -O 0 
> -E weka.classifiers.bayes.net.estimate.SimpleEstimator -- -A 1.0
> java_vm_args = -Xmx1200m
> Queue
> #This simply doesn't start :)
> 
> So it seams that the first two solution start over the windows machine, 
> and maybe the problem it's due to a compatibility problem between the 
> linux-java-way of calling classes and the way in wich windows call the 
> classes..., or it's just my error in building the job submission file... 
> I've no idea. Is there in the windows condor a sort of condor_config 
> file that need to be set? And wich is my error in configuring the linux 
> condor: why the latest tentative doesn't start? :'((((
> 
> Thank you very much for any hint that could make weka and condor work, 
> in any configuration. I must only show at my discussion (in 20 days) 
> that it works, not important how.
> 
> I posted in another message a condor_config that I edited on the linux submit-execute machines, but I was moderated because the previous messages was too long.
> 
> _______________________________________________
> Condor-users mailing list
> Condor-users@xxxxxxxxxxx
> https://lists.cs.wisc.edu/mailman/listinfo/condor-users
>