Thinking about High Availability using vmware features.
Want to concentrate in improving Puppet configuration.
UNIBE:
added 320 cores to cluster (now 456 job slots, very efficiently used by ATLAS jobs from NDGF over ARC)
started run ATLAS analysis, folded in the ANALY_ARC queue from the NDGF cloud. Smooth integration
no time to make progress on DPM storage, hope to have it online within the next 7-10 days
now also support "Life" VO within the SMSCG infrastructure. Will support more SMSCG VOs and aplications in the near future
UNIGE:
one recent crash of a file server running Solaris; we have a few such crashes per year; probability below 1/machine/year; all our file servers (eight Sun X4500+ aka "Thumpers", 4 do NFS export, 4 are in DPM) are concerned
data tranfers from NDGF-T1 to Genava at 30 MB/s; we would like 60 MB/s (~1/2 of the line speed to SWITCH backbone); investigation in progress, involves SWITCH and Ljubliana, iperf measuremeents between IJS and Uni GE show wild oscillations of the rate; will be followed up