search

dmwm / PHEDEX

CMS data-placement suite
8 stars 18 forks source link

Inconsistent data #1080

Open vlimant opened 7 years ago

vlimant commented 7 years ago

https://cmsweb.cern.ch/phedex/datasvc/xml/prod/requestlist?dataset=/MinBias_TuneCUETP8M1_13TeV-pythia8/RunIISummer15GS-MCRUN2_71_V1-v2/GEN-SIM

shows

however, XT1_US_FNAL_Old_Disk does not pass the valid node callback

400 Bad Request

The 'NODE' parameter ("XT1_US_FNAL_Old_Disk") to (eval) did not pass the 'node' callback\n\n

https://cmsweb.cern.ch/phedex/datasvc/xml/prod/subscriptions?block=/MinBias_TuneCUETP8M1_13TeV-pythia8/RunIISummer15GS-MCRUN2_71_V1-v2/GEN-SIM%23*&node=XT1_US_FNAL_Old_Disk

nataliaratnikova commented 7 years ago

The X pre-pended to the node name means that this node has been deprecated, but we still want to keep the history of requests for such nodes in PhEDEx, see https://twiki.cern.ch/twiki/bin/view/CMS/PhedexOperationsAdmin#Removing_a_node . I will check with the request submitter (Dave M) how he managed to create a subscription to a deprecated node (probably some scripts to handle the left-overs from disk-tape separation). Thanks for reporting this.

DAMason commented 7 years ago

If I'm looking at the same thing, the subscription was made March 4 in 2016 -- which would have been before the endpoint was decommissioned and renamed to XT1....

nataliaratnikova commented 7 years ago

Thanks for clarifying, Dave. Then this is an understood and expected behavior.

Jean-Roch, is this a show-stopper for any of your work? Is it worth the effort to try to exclude the requests to deprecated X-nodes from the query results ? I do not know if anyone wants them in, I think they are there because of the node ID being preserved after the node name change. Or perhaps you can exclude the subscription queries for the X-nodes in your code, or handle an exception for the 'node' parameter validation errors?