[DRMAA-WG] Torque/PBS DRMAA - $drmaa_incr_ph$ issue
hrabri.rajic at intel.com
Tue May 8 11:33:54 CDT 2007
Need to search harder :-)
Job working directory
are all mentioned explicitly. Based on the old discussion, that was
likely to be the least common denominator in the native array job
submission routines. That needs to be made more explicit in the spec.
Going beyond it would require a fresh research of the current DRM
systems and also DRMAA user feedback.
>From: drmaa-wg-bounces at ogf.org [mailto:drmaa-wg-bounces at ogf.org] On
>Of Rajic, Hrabri
>Sent: Tuesday, May 08, 2007 8:24 AM
>To: DRMAA Working Group
>Subject: Re: [DRMAA-WG] Torque/PBS DRMAA - $drmaa_incr_ph$ issue
>The restriction is proving to be difficult to locate, at least for me.
>In either case, section 3.2.4 is in a need of update to reflect the
>implied restriction or refer to its text.
>>From: drmaa-wg-bounces at ogf.org [mailto:drmaa-wg-bounces at ogf.org] On
>>Of Peter Troeger
>>Sent: Friday, May 04, 2007 10:27 AM
>>To: DRMAA Working Group
>>Subject: Re: [DRMAA-WG] Torque/PBS DRMAA - $drmaa_incr_ph$ issue
>>> To ease handling of bulk jobs, we recommend adding the ability to
>>> $drmaa_incr_ph$ not only in file paths but also in program
>>> arguments and
>>> environment variables.
>>> The spec is not explicit regarding the program arguments part and
>>> implies that it could be embedded into any attribute, but to be 100%
>>> certain we would need to check the previous records to see if and
>>> the restrictions are.
>>> We also need to check what is in the current implementations.
>>> Could the
>>> implementers, please respond to this thread?
>>The DRMAA specification permits the variable usage explicitly in the
>>according attribute descriptions, which only implies the non-usage
>>for the other attributes. The IDL spec is more precise and lists up
>>the allowed attributes for PARAMETRIC_INDEX. The Condor
>>implementation allows this variable only in the path-related
>>I agree to take this as important feature request, and will create
>>the tracker item when the initial discussion is finished.
>>Here is the list of job template string / string list attributes,
>>which could also get support for this variable:
>>Already allowed for:
>> attribute string workingDirectory;
>> attribute string inputPath;
>> attribute string outputPath;
>> attribute string errorPath;
>>Could also be allowed for:
>> attribute string remoteCommand;
>> attribute OrderedStringList args;
>> attribute Dictionary jobEnvironment;
>> attribute string jobName;
>> attribute string nativeSpecification;
>>Makes maybe no sense for:
>> attribute string jobCategory (since this is based on a predefined
>>list of keywords)
>> attribute StringList email (since this must be a valid eMail
>> drmaa-wg mailing list
>> drmaa-wg at ogf.org
> drmaa-wg mailing list
> drmaa-wg at ogf.org
More information about the drmaa-wg