[Swift-devel] SGE's qstat output parsing failure

[Luiz Gadelha]

It worked fine with the 0.93RC5 binary distribution. I compiled Swift from SVN but I still get the parsing failure error, probably I'm not getting the right cog sources. I got cog and swift from:

https://cogkit.svn.sourceforge.net/svnroot/cogkit/tags/swift-0.93
https://svn.ci.uchicago.edu/svn/vdl2/branches/release-0.93

I tried https://svn.ci.uchicago.edu/svn/vdl2/branches/release-0.93.1 also but it gave me the same error.

Regards,

Luiz

On Nov 23, 2011, at 6:43 PM, Michael Wilde wrote:

> Luiz, O wonder: can you switch to using the latest release candidate (RC) of 0.93? Or better yet, build 0.93 form the latest rev in the 0.93 SVN branch, to pick up David's latest fixes to the SGE provider?
> 
> David is currently working on fixing problems that are affecting us and other users on Ranger (an SGE machine at TACC) and it would be good if we can verify that these latest fixes to the SGE provider work for your installation as well.
> 
> Thanks,
> 
> - Mike
> 
> 
> ----- Original Message -----
>> From: "David Kelly" <davidk at ci.uchicago.edu>
>> To: "Luiz Gadelha" <lgadelha at lncc.br>
>> Cc: "Swift Devel" <swift-devel at ci.uchicago.edu>
>> Sent: Wednesday, November 23, 2011 12:59:49 PM
>> Subject: Re: [Swift-devel] SGE's qstat output parsing failure
>> Thanks Luiz. I've seen some similar issues with 0.92 when parsing
>> qstat output. qstat output tends to change between versions which can
>> cause parsing problems like this. For 0.93, we will be using the XML
>> output to help standardize things a bit.
>> 
>> Regards,
>> David
>> 
>> ----- Original Message -----
>>> From: "Luiz Gadelha" <lgadelha at lncc.br>
>>> To: "Swift Devel" <swift-devel at ci.uchicago.edu>
>>> Sent: Wednesday, November 23, 2011 12:27:03 PM
>>> Subject: [Swift-devel] SGE's qstat output parsing failure
>>> Hi,
>>> 
>>> There is a small bug in the parsing of SGE's qstat output lines. For
>>> the following qstat output:
>>> 
>>> job-ID prior name user state submit/start at queue slots ja-task-ID
>>> -----------------------------------------------------------------------------------------------------------------
>>> 543309 0.50500 null lgadelha r 11/22/2011 14:40:38
>>> linux.q at cmm11bl5.hpc.lncc.br 1
>>> 543310 0.50500 null lgadelha r 11/22/2011 14:40:38
>>> linux.q at cmm11bl5.hpc.lncc.br 1
>>> ...
>>> 
>>> QueuePoller incorrectly assumes that the job identifier is in the
>>> same
>>> position as "job-ID" in the header, which is not the case above,
>>> where
>>> it finds a space character and returns an empty string as the job
>>> identifier, Swift fails with a "Failed to parse qstat line" message.
>>> The attached log file is from a simple test script run that presents
>>> this problem.
>>> 
>>> The attached patch fixed the problem in my environment.
>>> 
>>> Regards,
>>> 
>>> Luiz
>>> 
>>> --
>>> Luiz Gadelha
>>> http://www.lncc.br/~lgadelha
>>> 
>>> 
>>> 
>>> 
>>> 
>>> 
>>> 
>>> 
>>> 
>>> _______________________________________________
>>> Swift-devel mailing list
>>> Swift-devel at ci.uchicago.edu
>>> https://lists.ci.uchicago.edu/cgi-bin/mailman/listinfo/swift-devel
>> _______________________________________________
>> Swift-devel mailing list
>> Swift-devel at ci.uchicago.edu
>> https://lists.ci.uchicago.edu/cgi-bin/mailman/listinfo/swift-devel
> 
> -- 
> Michael Wilde
> Computation Institute, University of Chicago
> Mathematics and Computer Science Division
> Argonne National Laboratory
> 

--
Luiz Gadelha
http://www.lncc.br/~lgadelha