RAL Tier1 weekly operations castor 27/05/2016
Contents
Minutes from the previous meeting
Operation news
Automated workflow for disk server deployment has been disabled New CASTOR functional testing using xrootd will be enabled on Monday 23/5/2016 CASTOR issues
Heavy workload on the Atlas scratch disk resulting in almost nothing being achieved
Full recovery from the tape robot and air condition problems. Chris checked status of migration queues last weekend and on Mon 16/5
Double putStart problem on CASTOR facilities
Some work to be done on the improvement of the new draining script's logic
gdss664 was brought back to production on 18/05/2016 at ca. 15:00 folowing a sucessfull rebuilding
gdss727 (production D1T0 CMS disk server) FSProbe Error Removed from Production and Overwatch Updated RT 172141
xrootd segmentation fault on atlas-xrd-proxy01. John Kelly investigated /var/log/messages and /var/log/xrootd/manager/atlas/xrootd.log.20160516 and found that the machine was busy shortly before the error. GP tried to debug the dumped core file but could not run xrootd as root.
Ongoing work on the upgrade to CASTOR 2.1.15 on preprod
GP and BD to chase the dteam VO for the GP membership request
GP and BD to perform stress testing of gdss596 to evaluate the new WAN parameters
GP to talk to Andrew Lahiff about a SL7 upgrade on the worker nodes using aquilon.
SRM DB duplicates removal script is under testing
BD AND RA will test the newly created tape families for ATLAS today Fri 20/5
Operations news
New tape pools created for LHCb and CMS
Draining on GDSS680 (atlas strip) with 30k files per partition on server worked
Operations problems
Tape robot - ran OK overnight (26-27/5). Currently using backup controller server. Oracle engineer due 11:00 27/5.
GDSS635 - atlas tape ... slight confusion, staged files on the server (not canbemigrs) when filesystem was rebuilt
40 files in atlas scratch had zero size in CASTOR namespace, BD declare lost to Atlas
Atlas seeing failing transfers because file size and checksum that rucio held were different.
SNO+ GGUS ticket has been outstanding for some time
Planned, Scheduled and Cancelled Interventions
CASTOR 2.1.15
Long-term projects
SL6 to SL7 upgrade on all CASTOR tape servers
Staffing
Bank Holiday on Monday
RA out Fri 3rd
Oncall CP from Tuesday
Actions
GP needs to review mailing lists he is on / can he access GGUS
GP to discuss with DB team to include file size in Nameserver dumps. The goal is to identify zero length files
GP to review outstanding SNO+ GGUS ticket
GP and BD to chase the dteam VO for the GP membership request
GP and BD to perform stress testing of gdss596 to evaluate the new WAN parameters
GP to talk to Andrew Lahiff about a SL7 upgrade on the worker nodes using aquilon.
SRM DB duplicates removal script is under testing
Completed actions
BD AND RA to test the newly created tape families for ATLAS today