Closed StevenCTimm closed 1 year ago
REQ000000437120 has been opened to make this happen.
Fermilab has made gpce03/gpce04 now support DUNE. Now we have to add DUNE support at the OSG factory so it will show up in the vofeed.xml
Ken will open a ticket to factory ops but we also have to block the gpfrontend01/02 to not send glideins there.
Ken has filed the ticket to the factory ticket 65480 https://support.opensciencegrid.org/public/tickets/c85bd303d15dd5791a9f5be5e84a2c87268d2dcb91d1975b5fbd8e7cd3a72279
Factory entry is now in, now we have to modify the frontends at Fermilab so they don't request to it. deny the site FNAL_GPGrid in the two frontend groups dune and dune_cern Maybe merge dune and dune_cern while we are at it.
Frontends are now modified. All we have to do is make a new vofeed.xml with FNAL_GPGrid in it.
ok now testing is ok on gpce03/04 Keep the ticket open, see if CMS Tier 1 will also let us run.
Not sure I understand. What are we watching for from CMS?
There are two clusters at Fermilab of roughly equal size, one called "FermiGrid" which is what we have been working on , which is behind gpce03/gpce04, and the other one is the CMS Tier 1. There are rarely many cycles available on the latter but most other Fermilab experiments do have opportunistic access to it. At the moment DUNE does not, but we would like to get it.
I'll send a few test jobs to the FNAL T1 and force them to go there so that we have constant pressure until they run.
CMS Tier 1 (host cmsosgce-opp1) now supports DUNE jobs. But now the next step is to get it to take the DUNE ETF tokens as well (which now exist, as of today).
This was completed some time ago .
Right now gpce03.fnal.gov and gpce04.fnal.gov don't actually support the DUNE VO because all DUNE submission at Fermilab is done locally through jobsub and doesn't glide in. We can request that they do and they probably will.