[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Condor-users] Submiting jobs from Globus to a Condor pools. should_transfer_files



On 11/20/2009 06:42 AM, cmasmas cmasmas wrote:
> Hello everyone:
> 
> As I said in a previous mail, I'm trying to set up a Condor pool managed by
> Globus.
> 
> First of all, if I include this lines to a job template, the job will run
> OK.
> 
> *should_transfer_files = YES
> when_to_transfer_output = ON_EXIT
> 
> *Else, the job becomes idle. Here is the output for condor_q -better:
> 
> *WARNING:  Be advised:
>    No resources matched request's constraints
> 
> The Requirements expression for your job is:
> 
> ( target.Arch == "INTEL" ) && ( target.OpSys == "LINUX" ) &&
> ( target.Disk >= DiskUsage ) && ( ( target.Memory * 1024 ) >= ImageSize ) &&
> ( TARGET.FileSystemDomain == MY.FileSystemDomain )
> 
>     Condition                         Machines Matched    Suggestion
>     ---------                         ----------------    ----------
> 1   ( TARGET.FileSystemDomain == "xxx.es" )0                   MODIFY TO "
> CNode2.xxx.es"
> 2   ( target.Arch == "INTEL" )        2
> 3   ( target.OpSys == "LINUX" )       2
> 4   ( target.Disk >= 7 )              2
> 5   ( ( 1024 * target.Memory ) >= 7 ) 2         *
> 
> 
> The problem is that I don't know how to specify the "should_transfer_files"
> attribute in a Globus job template, so jobs always become idle.
> 
> So, two questions (answering one of them would be enough):
> a) Is there a way to get the jobs running without specifying "should tranfer
> files"?
> b) How can I specify that attribute in a Globus job template?
> 
> 
> This is an example template:
> 
> *<job>
>         <executable>/bin/hostname</executable>
>         <argument>-f</argument>
>         <stdout>${GLOBUS_USER_HOME}/hostname.stdout</stdout>
>         <stderr>${GLOBUS_USER_HOME}/hostname.stderr</stderr>
>  </job>*
> 
> Thanks in advance.
> 

(a) - For some attribute, e.g. Arch OpSys FileSystemDomain Memory Disk, condor_submit tries to be helpful and provide a meaningful default in your job's Requirements expression. condor_submit will not second guess you though. If you use any of those attributes yourself, condor_submit won't provide a default. For instance,

11:51:17am> chopin:~ $ echo "cmd=/bin/true\nqueue\n" | condor_submit
Submitting job(s)condo.
1 job(s) submitted to cluster 339.

11:51:44am> chopin:~ $ condor_q -l 339 | grep ^Requirements
Requirements = (Arch == "INTEL") && (OpSys == "LINUX") && (Disk >= DiskUsage) && ((Memory * 1024) >= ImageSize) && (TARGET.FileSystemDomain == MY.FileSystemDomain)

11:51:51am> chopin:~ $ echo "cmd=/bin/true\nrequirements=FileSYstemDomain =!= UNDEFINED\nqueue\n" | condor_submit
Submitting job(s).
1 job(s) submitted to cluster 340.

11:52:12am> chopin:~ $ condor_q -l 340 | grep ^Requirements                   Requirements = (FileSYstemDomain =!= UNDEFINED) && (Arch == "INTEL") && (OpSys == "LINUX") && (Disk >= DiskUsage) && ((Memory * 1024) >= ImageSize)

Checking for FileSystemDomain to be defined is logically a nop.

Best,


matt