Hi Szabolcs,
You can change the "giis url" in the gocdb for you site and this will update
the Giis monitor query as well. However there may a delay of up to 30
minutes since this information is cached. I can clear this cache to make
the transition faster, if you let me know when the change is made.
Btw, there are new scripts that check the status of rgma:
http://grid-deployment.web.cern.ch/grid-deployment/documentation/LCG2-Site-T
esting/LCG2-Site-Testing.html#SECTION00060000000000000000
Cheers,
Min
-----Original Message-----
From: LHC Computer Grid - Rollout [mailto:[log in to unmask]] On
Behalf Of Laurence
Sent: Monday, April 11, 2005 12:02 PM
To: [log in to unmask]
Subject: Re: [LCG-ROLLOUT] LCG-2_4_0 problems after upgrading
1. GIP
There is no yaim function that configures the condor batch system. You
will need to create the script /opt/lcg/libexec/lcg-info-dynamic-ce
yourself. This is a wrapper script that calls the dynamic plugin for
condor with all the arguments etc. There is a new condor dynamic plugin
that contains a number of bug fixes in the release.
3. R-GMA
Can you check what version of R-GMA you have installed, you should have
edg-rgma-api-java-4.0.2-1
Service Status is now published via
/etc/rc.d/init.d/edg-rgma-servicetool
not the script that your trying to start. The old rgam rpms should have
been removed when you upgraded.
Hernath Szabolcs wrote:
> Dear List,
>
> here are a few notes on the problems experienced while/after upgrading to
> 2_4_0 from 2_3_1 (OS is/was SLC) with yaim:
>
> 1. GIP
>
> /opt/lcg/var/gip/lcg-info-generic.conf (on the CE) has been configured to
> use the dynamic script /opt/lcg/libexec/lcg-info-dynamic-ce, but there is
> no such file on our system (which package owns this file?). In earlier
> versions, where lcg-info-dynamic-condor was configured, the invocation
> was
> incorrect (the correct form is 'lcg-info-dynamic-condor <condor_bin_path>
> <config_file>'). I do not know whether this has been corrected, as yaim
> did not configure it this time. The lcg-info-dynamic-condor script itself
> was somewhat buggy in 2_3_1 - again, I do not know if it has been
> corrected, as our own corrected version did not get overwritten.
>
> 2. JM
>
> As already pointed out by Gergely Debreczeni, the jobmanager invocation
> (/opt/globus/etc/grid-services/jobmanager-lcgcondor) lacks the critical
> condor-arch (and the not-so critical condor-os) options.
>
> 3. R-GMA
>
> Well, apart from the fact that it still does not work :-),
> /etc/init.d/edg-rgma-service-status complains about the missing files
> /opt/edg/etc/rgma/rgma-defaults and
> /opt/edg/etc/rgma/rgma-tools-defaults.
> How should these be created?
>
> /opt/edg/sbin/test/edg-rgma-check and edg-rgma-run-examples only
> complained about the ServiceStatus applet, but I can't check them right
> now, as there seems to be a problem with the registry...
>
> 4. Monitoring
>
> As already announced earlier, our infosys contact string has been changed
> to
>
> ldap://grid109.kfki.hu:2170/mds-vo-name=BUDAPEST,o=grid
>
> The gstat monitoring, however, still tries to query
>
> ldap://grid109.kfki.hu:2135/mds-vo-name=budapestlcg2,o=grid
>
> thus we have been getting an "ERROR" state lately (Min, could you please
> change that?).
>
> The lcg testzone reports still show yesterday's results for our site,
> whereas other sites seem to have been running the tests today as well.
> Any
> reason for that? The new set of CA rpms still seem to spoil the test
> (the UK eScience md5sum thing) and so mark a critical test as failed.
>
> Regards,
>
> Szabolcs Hernath
>
|