On 09/06/2012 10:15 AM, Daniela Bauer wrote:
> The 'fix' for the tags is to revert to an earlier version. In practice
> you can just change two lines in a perl script. In:
> /usr/libexec/lcg-info-dynamic-software
>
>
> foreach my $sc ( keys %subclusters ) {
> # Readin in $edgpath
> foreach my $tag (&tagsFromDir($edgpath) ) {
> # $subclusters{$sc}{'tags'}{lc($tag)} = $tag ;
> $subclusters{$sc}{'tags'}{$tag} = 1 ;
> }
> # Readin from $glitepath
> foreach my $tag (&tagsFromDir("$glitepath/$sc") ) {
> # $subclusters{$sc}{'tags'}{lc($tag)} = $tag ;
> $subclusters{$sc}{'tags'}{$tag} = 1 ;
> }
> }
>
> (This is the working version, with the released version commented out.)
>
Thanks. I guess I'd better put that patch into my build system.
Other things to watch out for are:
1) Re-Yaiming splatted my CREAMDB, even though it had jobs in it
already. If that turns out to be real, then it's horrible.
2) GlueCEStateWaitingJobs is not transmitted properly e.g.
ldapsearch -LLL -x -h lcg-bdii.gridpp.ac.uk:2170 -b o=grid
'(GlueCEUniqueID=hepgrid6*)' | grep StateWaitingJobs
Result: GlueCEStateWaitingJobs: 725
ldapsearch -LLL -x -h lcg-bdii.gridpp.ac.uk:2170 -b o=grid
'(GlueCEUniqueID=hepgrid5*)' | grep StateWaitingJobs
Result: GlueCEStateWaitingJobs: 444444
3) In the CREAM logs, a new exit code, [exitCode=W], crops up, e.g.
06 Sep 2012 12:03:12,252
org.glite.ce.creamapi.jobmanagement.cmdexecutor.AbstractJobExecutor -
JOB CREAM788167268 STATUS CHANGED: RUNNING => DONE-OK
[failureReason=reason=0] [exitCode=W] [localUser=prdhne05]
[gridJobId=https://lb02.grid.cyf-kr.edu.pl:9000/Xy1o6jq5xvfUMuTgeJsrfw]
[workerNode=r21-n01.ph.liv.ac.uk]
[delegationId=13468516952E875921wms012Egrid2Ecyf2Dkr2Eedu2Epl]
What's all that about, then? What does code W mean??
Cheers,
Steve
--
Steve Jones [log in to unmask]
System Administrator office: 220
High Energy Physics Division tel (int): 42334
Oliver Lodge Laboratory tel (ext): +44 (0)151 794 2334
University of Liverpool http://www.liv.ac.uk/physics/hep/
