● Outstanding tickets
GGUS #145614 Timeouts also caused by CERN FTS falling over. changed memory allocation for xrootd, and added timeout of 2 mins. Now transfers all green.
GGUS #145688 Problem for many sites in UK accessing CERN Squids directly, rather than using the RAL Stratum 1. Manchester tries to contact RAL squid on IPV6, falls back to IPV4. Now seeing connections with RAL.
Alessandra will contact Jose at RAL again to check if IPv6 is enabled for squid (we believe it is) and if RAL is using the OS Squid or the Frontier version (we believe the latter).
Vip has upgraded Squids at Oxford, but still using IPv4.
Details can be seen on the WLCG Squid monitor page: http://wlcg-squid-monitor.cern.ch/awstats/bin/awstats.pl?month=02&year=2020&output=allhosts&config=cvmfsbproxy.cern.ch&framename=index
● CPU
QMUL hammercloud: not as many jobs running HC across the board. At QMUL, the jobs are failing with SIGCONT.
● Glasgow Ceph storage
Fixed a few niggles, but as of a few days ago, Grid transfers had problems talking to Ceph pool.
xrootd cache endpoint for reads still works, but writes don't work. xrootd occasionally segfaults.
Using locally compiled version of xrootd (latest from 6 months ago), but will have to rebuild when upgrading to Nautilus.
● AOB
Alessandra: NETR
Dan: NETR
Elena:
problem with rucio (mentioned on Rucio-support list).
Is the recommendation to update Singularity to 3.5.3? Will ask Alessandra.
Gareth: Will need to have fully moved to new server room + Ceph by June/July, requiring move of all storage by then.
James: NETR
Matt: jobs running out of memory at Lancaster - same user as in January.
Sam: NETR
Stewart: NETR
Tim: NETR
Vip: NETR
There are minutes attached to this event.
Show them.