OK I think I got it.
I took a look on an old RB, which is using an old BDII (lcg 2.3 or
something like that).
Same problem.
Sooooo...
lxn1184 is installed with lcg 2.6 I think because it's publishing
information using GlueSchema v. 1.2
All "GlueCESEBindCEAccesspoint" attributes I could find in the "old"
BDII were published for nodes with :
GlueSchemaVersionMajor: 1
GlueSchemaVersionMinor: 1
So I think all CEs that are publishing GlueSchema v1.2 information will
cause the SE accesspoint to disappear in BDIIs that were not upgraded
yet to LCG 2.6
Note : this is a good way to check who's upgraded or not ;)
Conclusion : use $LCG_GFAL_INFOSYS = lcg-bdii.cern.ch:2170 (or
equivalent) until the BDIIs are upgraded.
Cheers,
Fred
Frederic Schaer a écrit :
> Hi,
>
> We are facing the same problem with a French site (CGG) : the lxn1183
> accesspoint is not found.
>
> + lcg-rep -v --vo dteam -d lxn1183.cern.ch
> lfn:sft-lcg-rm-cr-eio99.pp.weizmann.ac.il.0508221822
> CE Accesspoint not found
> lcg_rep: Invalid argument
> Using grid catalog type: edg
> + result=1
> + set +x
>
> I checked in the BDII used by CGG (rb1.egee.fr.cgg.com:2170), and it
> seems some of the information published by lxn1184 (which is -correct
> me if I'm wrong- the node which publishes the accesspoint for lxn1183)
> is missing... but I don't know why.
>
> For example, following is missing in the BDII :
>
> # lxn1183.cern.ch, lxn1184.cern.ch:2119/jobmanager-lcgpbs-dteam,
> cernlcg2, lo
> cal, grid
> dn:
> GlueCESEBindSEUniqueID=lxn1183.cern.ch,GlueCESEBindGroupCEUniqueID=lxn1184
>
> .cern.ch:2119/jobmanager-lcgpbs-dteam,mds-vo-name=cernlcg2,mds-vo-name=local,
>
> o=grid
> objectClass: GlueGeneralTop
> objectClass: GlueCESEBind
> objectClass: GlueSchemaVersion
> GlueCESEBindSEUniqueID: lxn1183.cern.ch
> GlueCESEBindCEAccesspoint: /storage
> GlueCESEBindCEUniqueID: lxn1184.cern.ch:2119/jobmanager-lcgpbs-dteam
> GlueCESEBindMountInfo: none
> GlueCESEBindWeight: 0
> GlueSchemaVersionMajor: 1
> GlueSchemaVersionMinor: 2
>
> While following is present :
>
> # lxn1184.cern.ch:2119/jobmanager-lcgpbs-dteam, cernlcg2, local, grid
> dn:
> GlueCESEBindGroupCEUniqueID=lxn1184.cern.ch:2119/jobmanager-lcgpbs-dteam,m
>
> ds-vo-name=cernlcg2,mds-vo-name=local,o=grid
> objectClass: GlueGeneralTop
> objectClass: GlueCESEBindGroup
> objectClass: GlueSchemaVersion
> GlueCESEBindGroupCEUniqueID: lxn1184.cern.ch:2119/jobmanager-lcgpbs-dteam
> GlueCESEBindGroupSEUniqueID: lxn1183.cern.ch
> GlueSchemaVersionMajor: 1
> GlueSchemaVersionMinor: 2
>
>
> I'm wondering if this could be some kind of GlueSchema version problem
> in the BDIIs ?
> Any idea ?
>
> Cheers,
> Fred
>
> Dan Schrager a écrit :
>
>> Hi,
>>
>> The so called "Replica Management test" fails for WEIZMANN-LCG2 when
>> performed against lxn1183.cern.ch while it's OK with castorgrid.cern.ch.
>> Is it a local cause or is it something out there ?!
>>
>> Regards,
>> Dan
>>
>>
>
|