Il 18/03/2011 10.11, Andreas Haupt ha scritto:
> Hi Stephen,
>
> we are currently also observing a strange BDII error.
>
> On Mon, 2011-02-14 at 17:16 +0000, Stephen Burke wrote:
>> LHC Computer Grid - Rollout [mailto:[log in to unmask]]
>>> On Behalf Of Daniela Bauer said:
>>> I've just reyaimed one of the machines (topbdii02) behind
>>> topbdii.grid. The CE is still not there.
>>> If I look in top-urls.conf-glue2-2011-02-14.log I see
>>>
>>> --> RU-Protvino-IHEP:
>>> No such object (32)
>> That may be some kind of clue, if the site bdii doesn't know about the glue 2 schema at all it must be pretty old since that was rolled into the resource bdii in 2009, but that doesn't directly explain why you're missing some (but not all) of the glue 1 objects ...
> This discussion led to ticket:
> https://gus.fzk.de/dmsu/dmsu_ticket.php?ticket=67372
>
> But in the case the site-BDII seemed to be old. That's not the case
> here:
>
> Top-BDII:
>
> [root@lcg-bdii1 ~]# rpm -q glite-BDII_top
> glite-BDII_top-3.2.10-3.sl5
>
> Site-BDII:
>
> [root@lcg-giis1 ~]# rpm -q glite-BDII_site
> glite-BDII_site-3.2.10-1.sl5
>
> The GlueCEUniqeID entry of our lcg-CE disappears in the Top-BDII just
> for the cms VO! All others are ok ...
>
> Top-BDII:
>
> [oreade38] ~ % for vo in atlas lhcb cms; do
> ldapsearch -x -LLL -H ldap://lcg-bdii.ifh.de:2170 -b 'Mds-Vo-name=DESY-ZN,Mds-Vo-name=local,o=grid' "(GlueCEUniqueID=lcg-ce0.ifh.de:2119/jobmanager-lcgpbs-$vo)" GlueCEName
> done
> dn: GlueCEUniqueID=lcg-ce0.ifh.de:2119/jobmanager-lcgpbs-atlas,Mds-Vo-name=desy-zn,Mds-Vo-name=local,o=grid
> GlueCEName: atlas
>
> dn: GlueCEUniqueID=lcg-ce0.ifh.de:2119/jobmanager-lcgpbs-lhcb,Mds-Vo-name=desy-zn,Mds-Vo-name=local,o=grid
> GlueCEName: lhcb
>
> Site-BDII:
>
> [oreade38] ~ % for vo in atlas lhcb cms; do
> ldapsearch -x -LLL -H ldap://lcg-giis.ifh.de:2170 -b 'Mds-Vo-name=DESY-ZN,o=grid' "(GlueCEUniqueID=lcg-ce0.ifh.de:2119/jobmanager-lcgpbs-$vo)" GlueCEName
> done
> dn: GlueCEUniqueID=lcg-ce0.ifh.de:2119/jobmanager-lcgpbs-atlas,Mds-Vo-name=DESY-ZN,o=grid
> GlueCEName: atlas
>
> dn: GlueCEUniqueID=lcg-ce0.ifh.de:2119/jobmanager-lcgpbs-lhcb,Mds-Vo-name=DESY-ZN,o=grid
> GlueCEName: lhcb
>
> dn: GlueCEUniqueID=lcg-ce0.ifh.de:2119/jobmanager-lcgpbs-cms,Mds-Vo-name=DESY-ZN,o=grid
> GlueCEName: cms
>
> Any idea what goes wrong here?
>
> Cheers,
> Andreas
Hi Andreas,
when only a particular queue disappear from a top-level BDII, the
"guilty" is always the FCR: in your case, your site should be failing
some CMS test, so the CMS queue has been excluded by the production BDII
cheers,
Alessandro
|