Re: [glue-wg] [Glue-schema] The mapping of Glue Schema v1.3 to old classat format

Hi, I have updated the mapping document to take into consideration Stephen's comment. Everything else in the document is the same. http://home.fnal.gov/~garzogli/Glue2OldClassadMapping/Glue-v1.3/glue-v1.3-ol... I would like to have this document uploaded to the official glue schema mapping page. Please, give me your feedback. Thanks Gabriele Burke, S (Stephen) wrote:
glue-schema-bounces@george.lbl.gov
[mailto:glue-schema-bounces@george.lbl.gov] On Behalf Of Gabriele Garzoglio said: I have updated the old mapping document to include Glue Schema v1.3.
"In some General Information Provider implementations (GIP), the ContactString Info attribute of the ComputingElement entity is not necessarily a GRAM URL and may have the form <GRAM-URL>-<VO-Name>. This form cannot be used directly as a contact string by most Grid clients. In LCG, the Resource Broker processes this value so that it can be used by the standard Grid clients. In the mapping to old ClassAd, instead, the value is published in the form <GRAM-URL>, irrespectively of the GIP implementation, by comparing the value of ContactString with the values in the GlueCEAccessControlBaseRule (i.e., the VO names). This is important in order to use the Condor Matchmaking Service out-of-the-box for Resource Selection."
Unless I'm missing the point, this is wrong. The LCG format is <GRAM-URL>-<QUEUE-Name> (at least for LCG CEs, glite CEs don't use GRAM at all). It happens to be the case that many sites have queues named for VOs, but it isn't a general rule.
Stephen
_______________________________________________ Glue-schema mailing list Glue-schema@hicb.org http://dsd.lbl.gov/mailman/listinfo/glue-schema
participants (1)
-
Gabriele Garzoglio