Heya all,
Just a quickie due to the Bank Holiday, we're up to 29 tickets this week.
SUSSEX
https://ggus.eu/?mode=ticket_info&ticket_id=107814 (22/8)
Ops failures on the Sussex Cream. Matt sent a e-mail to TB-SUPPORT about
this so if anyone could chime in that would be appreciated - atlas are
running fine, but Ops tests (and possibly other jobs coming in via WMS)
are hitting a spot of bother. In my experience delegation errors like
this often pass in time, but the errors have been going for over 4 days.
Any help appreciated.
https://ggus.eu/?mode=ticket_info&ticket_id=107801 (21/8)
Perhaps this problem is also affecting Sno+? In Progress (22/8)
RALPP
https://ggus.eu/?mode=ticket_info&ticket_id=107844 (24/8)
Just a heads up that this atlas "no free space" ticket has been reopened
with (possibly unrelated) srm errors. Reopened tickets often sneak past
our sentries. Reopened (24/8)
SNOPLUS SOFTWARE DIR to CVMFS (21/8)
LIVERPOOL: https://ggus.eu/?mode=ticket_info&ticket_id=107796
SHEFFIELD: https://ggus.eu/?mode=ticket_info&ticket_id=107798
QMUL: https://ggus.eu/?mode=ticket_info&ticket_id=107799
Sno+ has asked sites to have their VO SW DIR environmental variable
point to their cvmfs directory. All three sites are on it, something for
anyone rolling out Sno+ support.
BRISTOL
https://ggus.eu/?mode=ticket_info&ticket_id=106325 (18/6)
Winnie has spotted that the CMS pilots losing contact to the submission
host problem is only (at least recently) affecting their ARC CE. Whilst
the CE flavour isn't the only difference between the clusters this is
strongly suggesting that the problem isn't with the site firewall. On
Hold (19/8)
UCL
https://ggus.eu/?mode=ticket_info&ticket_id=107711 (15/8)
UCL received an Apel-Pub Ops ticket nearly a fortnight ago which has yet
to be even acknowledged. I suspect Ben is on holiday, can someone
(looking at the Londoners) poke through other channels? Assigned (15/8)
TIER 1
https://ggus.eu/?mode=ticket_info&ticket_id=107815 (22/8)
DirectJobSubmit Ops failures at the Tier 1. Catalin asks if the Ops jobs
can be tuned and have their registration timeouts increased - as it
appears only the test jobs are suffering failures of this kind. Waiting
for reply (22/8)
Cheers!
Matt
|