Attendee: Douglas, Booker, Wei, Randy, Alf
1. Tier 2 Production and Operation
No jobs. New pilot can't contact BNL Panda server. Panda team is
working on it. Complaint to Kaushik De to get priority to fix this
problem. However, US ATLAS is running out of jobs.
ATLAS is switching to run jobs using WLCG static space token. All Tier
2s are ready. Jobs writing output to storage via SRM gateway is seen
as a potential performance bottleneck; US ATLAS is switching its
catalog system to LFC which is in use by International ATLAS. Still
have lots of issues because the LFC is designed to following the LCG
environment, not OSG environment.
2. dq2 end user tools.
David Miller reported trouble with the new dq2-put command. It
coincided with hardware issues we have on Sun/Mon. We are asking David
to try again. Also suggested users to use bbcp to transfer data from
SLAC to CERN.
3. Conditions DB
Rod Walker of TRIUMF is trying access CondDB via Frontier (which
translate SQL to http request, thus allow using Squid caching). Tried
at SLAC. The initial result is that we reduced the access time from
~2300 seconds (via directly SQL to CERN) to ~300 seconds with a cold
Squid cache (we don't understand why), and to ~38 seconds with a warm
Squid cache. According to Rod, the initial reaction from ALTAS DB
group is that this is interesting for Tier 2 to access CondDB at Tier
1 but will not solve the current problem of DB replication (oracle
stream) from T0 to T1s.
4. WT2 hardware and infrastructure.
All storage and batch nodes are installed and in service. Wain022 is
running Solaris 10 update 5.
5. Network monitoring hosts.
Agreed with Les that we would work with Yi-Ting Lee on this when he is
back.
6. Interactive machine for SLAC-ATLAS users and 1/2 Thumper Disk space
Requests sent in. Asking to prioritized these requests.
7. Xrootd storage and OSG
At the request of US ATLAS and Michael Ernst, OSG is now integrating
Xrootd and Berkeley SRM into OSG/VDT as a storage system suite,
similar to dCache. OSG/VDT will provide front line support to Tier 3
users. A question comes out about how to divide support scope for OSG/
VDT storage group and SLAC Xrootd team (and LBNL SRM team). It is not
clear how much resource we can provide to backup OSG/VDT. We will
setup a RT queue for OSG storage group to submit requests, and seen
how much work load we get.
Wei Yang | [log in to unmask] | 650-926-3338(O)
|