[GE globus] Re: VDT modifications to Globus 2 SEG

Jeff Porter rjporter at lbl.gov
Thu Jul 10 15:58:43 BST 2008


    [ The following text is in the "ISO-8859-1" character set. ]
    [ Your display is set for the "ISO-8859-10" character set.  ]
    [ Some special characters may be displayed incorrectly. ]


Hi Blair & others,

I suspect others have made changes to the code between these Oct 03 and 
Sept 07 dates though I didn't compare VDT version with the LeSC version 
directly.  My changes done in Sept '07 were only targeted to allow WS 
GRAM functionality. 

I do remember seeing the issues you mentioned here already in the code - 
appending ".real" to the stdout file and the often wrong job-state 
evaluation returned from the poll function.  I actually did try to fix 
those but none of the solutions I came up with worked completely.  Since 
my goal was to get GT4 working,  I decided to leave them as is.

I will take another look at the code to see if I can recall why these 
problems exist.  Perhaps others have different solutions.  It's also 
possible that one of my changes triggered a new problem that I didn't 
run into.

thanks, Jeff


On 7/10/08 7:06 AM, Blair Bethwaite wrote:
> On Thu, Jul 10, 2008 at 11:45 PM, Yoshio Tanaka
> <yoshio.tanaka at aist.go.jp> wrote:
>   
>> Which version of the LeSC Globus-SGE adapter is included in VDT?
>>     
>
> Good point Yoshio, I should have looked at that to begin with! The
> headers in sge.pm on Komolongma say:
> # Marko Krznaric
> # London eScience Centre
> # June 2003
> #
> # Contributions by David McBride
> # London eScience Centre
> # Oct 2003
> #
> # Contributions by Jeff Porter
> # Lawrence Berkeley National Laboratory
> # Sept 2007
>
> That's using VDT 1.10.1a. I don't have login access to any other
> Globus 2 machines which I know are using VDT so I can't compare. But I
> know I've received the extra polling output from jobmanager-sge on
> some Open Science Grid resources, which also use VDT.
>
>   
>> I know that there were some bugs in the LeSC adapter hence we use AIST
>> version of the Globus-SGE adapter which has added some new features
>> and fixed bugs in the LeSC version though it was released in 5 years
>> ago.
>>     
>
> Okay, so my guess is that the job stdout/err problems and extra output
> from polling were introduced with Jeff's changes, Jeff? Reporting of
> job state as active rather than error may have already been a problem,
> though needs fixed, hopefully Jeff is our man there.
>
> I've attached the sge.pm from Komolongma for reference.
>
> Cheers,
> -Blair
>
>   



More information about the gridengine-users mailing list