MinutesOSGUC110329

Glide-in WMS (Marco)

• This is the VO front-end: contains four parts: user-pool collector (a Condor collector); glide-in VO front-end - python service thats connects to the factory, triggering the submission of glide-ins. user-pool schedd, where the "user" submits his own jobs. This has been packaged together into an rpm, glidewmsvofrontend (by Burt and Derek, w/ Brian). Intention is that VDT takes over this. Glide-in (aka pilots) are submitted by a "factory" - you can install your own, or use the OSG production Glidein factory running by UCSD. Marco is using the UCSD factory. A VO manager uses Burt's rpm. All users of the VO connect using the schedd. There is a submit machine setup at Nebraska. Users are given a Unix account. This is odd - but its a submit host for a small community.
• Its a user interface machine.
• Is it okay to create accounts for users?
• How to manage individual ssh accounts?
• Do we need to make quaranteened machine? cilogon, shibboleth, ...
• Marco - develop a testing plan for CG components.
• A functional campus router; submit to ITB.

Suchandra

• setup "certify" - runs on itb-mon, checks host and service certs. will create new certs for that machine. Fermilab and UCSD is using it. cert-based Condor auth used by worker. John Hover is supporting it.
• Tanya convo re: storage. Testing of dCache - at least to install. xrootd - there are rpms and pacman version (which is old), still undergoing transition w/ rpms. Use pacman install for now. Timeline? Eval. of CERN rpm this week. What are they changing by default. FRM issue.
• Panda infrastructure and links into ITB. Switch to autopyfactory.
• Broken images in ganglia links.
• Completely off tier3. Clean up tier3.
• April 13 is first day of downtime for ITB cluster.
• Storage resources up this week. Will restart vm's for xrootd.
• Fix up Panda queues. Something broken w/ ATLAS pilots.
• Remove old queue.
• Get Engage and HCC running on ITB, and pilot factory.

Nate

• edge6 is done.

CVS