UWS POST PHASE=RUN
Pierre Le Sidaner
pierre.lesidaner at obspm.fr
Wed Nov 5 22:45:42 CET 2014
On 11/05/2014 09:11 PM, Paul Harrison wrote:
> On 2014-11 -05, at 19:22, Walter Landry <wlandry at caltech.edu> wrote:
>
>> Getting rid of PENDING would certainly not be a minor modification to
>> UWS. It would, however, make UWS easier to implement, since it
>> integrates better with existing queuing software (e.g. SLURM, LSF,
>> PBS).
>>
> Can I just say that the way that I implemented my UWS server in the past was that the PENDING state was before the job was sent to any backend cluster service - part of the point of the UWS is to produce a uniform facade onto other job systems - there might not be an exact match for each of their states - but the intention of PENDING in UWS is that it is a "before everything" state, where in the original conception, all that the UWS server knows about the job is the JDL - i.e. only the business parameters, and none of the desired job control parameters.
>
>
> Paul.
>
>
Since Interop 2011 in Naple we have present that UWS 1.0 is proposing 3
ways to send a job witch make confusion for the service providers and
clients.
Also that pending status is not in phase with any usual batch queue or
scheduler do. User should not submit a job until he is ready, and is not
able to modify parameters or time duration when a job is submitted.
During a pending phase nothing should be done (like downloading data)
it's only during run phase.
Pierre
--
-------------------------------------------------------------------------
Pierre Le Sidaner
Observatoire de Paris
Division Informatique de l'Observatoire
Observatoire Virtuel 01 40 51 20 89
61, avenue de l'Observatoire 75014 Paris
mailto:pierre.lesidaner at obspm.fr
http://vo.obspm.fr
--------------------------------------------------------------------------
More information about the grid
mailing list