Difference between revisions of "Tier1 Operations Report 2018-01-24"

From GridPP Wiki
Jump to: navigation, search
()
 
(3 intermediate revisions by 2 users not shown)
Line 14: Line 14:
 
*Problems since last week with the Atlas Castor instance. There has been a high rate of deletion requests, and the back end stager database has been struggling to keep up with the total load (reads, writes and deletes).
 
*Problems since last week with the Atlas Castor instance. There has been a high rate of deletion requests, and the back end stager database has been struggling to keep up with the total load (reads, writes and deletes).
 
* A prolonged downtime required with CASTOR/ATLAS. The issue is primarily to do with  the increased load on the Atlas Stager DB.  Extensive investigations has revelled it to be linked with garbage collection.  The Castor/DB team have discovered that the diskcopy table is fragmented. The table has to be recreated. For this the stager daemon has to be shut down and, hence, the declared extended downtime.
 
* A prolonged downtime required with CASTOR/ATLAS. The issue is primarily to do with  the increased load on the Atlas Stager DB.  Extensive investigations has revelled it to be linked with garbage collection.  The Castor/DB team have discovered that the diskcopy table is fragmented. The table has to be recreated. For this the stager daemon has to be shut down and, hence, the declared extended downtime.
* One of the BMS (Building Management Systems) in the R89 machine room has been intermittently faulty causing some system restarts.  This system manages the pumps. In one case the chillers failed to restart. It was planned to replace a faulty card in the system this morning (24th Jan) - which should have taken around 30minutes. However, a test yesterday with one of the pumps running ‘stand-alone’ faioled. During the start-up procedure the remaining pumps ramped-up to balance the water flow-rate to compensate for the loss of one pump. This caused a power surge to the one pump4 and it tripped causing the damage to the fuse carrier. This isnow be looked at and a new date for the BMS board swap will be planned.
+
* One of the BMS (Building Management Systems) in the R89 machine room has been intermittently faulty causing some system restarts.  This system manages the pumps. In one case the chillers failed to restart. It was planned to replace a faulty card in the system this morning (24th Jan) - which should have taken around 30minutes. However, a test yesterday with one of the pumps running ‘stand-alone’ failed. During the start-up procedure the remaining pumps ramped-up to balance the water flow-rate to compensate for the loss of one pump. This caused a power surge to pump4 and it tripped causing damage to the fuse carrier. This is now being looked at and a new date for the BMS board swap will be planned.
 
<!-- ***********End Review of Issues during last week*********** ----->
 
<!-- ***********End Review of Issues during last week*********** ----->
 
<!-- *********************************************************** ----->
 
<!-- *********************************************************** ----->
Line 37: Line 37:
 
|}
 
|}
 
* gdss736 (lhcbDst - D1T0) – rebuilt and back in production (RO)
 
* gdss736 (lhcbDst - D1T0) – rebuilt and back in production (RO)
 +
* gdss776 (lhcbDst - D1T0) - Failed Wednesday afternoon (17th). Returned to service on Friday.)
 
<!-- ***************************************************** ----->
 
<!-- ***************************************************** ----->
  
Line 349: Line 350:
 
| style="background-color: #b7f1ce; border-bottom: 1px solid silver; text-align: center; font-size: 1em; font-weight: bold; margin-top: 0; margin-bottom: 0; padding-top: 0.1em; padding-bottom: 0.1em;" | Notes from Meeting.
 
| style="background-color: #b7f1ce; border-bottom: 1px solid silver; text-align: center; font-size: 1em; font-weight: bold; margin-top: 0; margin-bottom: 0; padding-top: 0.1em; padding-bottom: 0.1em;" | Notes from Meeting.
 
|}
 
|}
* None yet.
+
* There was a discussion around performance of the Tier1 when accessing off-site data from the worker nodes. This data path goes through the site firewall which is causing some restrictions.

Latest revision as of 14:26, 24 January 2018

RAL Tier1 Operations Report for 24th January 2018

Review of Issues during the week 18th January 2018 to 24th January 2018
  • Patching for Spectre and Meltdown was taking place last week. (E.g. an outage of Castor for a few hours last Wednesday, 17th Jan).
  • Problems since last week with the Atlas Castor instance. There has been a high rate of deletion requests, and the back end stager database has been struggling to keep up with the total load (reads, writes and deletes).
  • A prolonged downtime required with CASTOR/ATLAS. The issue is primarily to do with the increased load on the Atlas Stager DB. Extensive investigations has revelled it to be linked with garbage collection. The Castor/DB team have discovered that the diskcopy table is fragmented. The table has to be recreated. For this the stager daemon has to be shut down and, hence, the declared extended downtime.
  • One of the BMS (Building Management Systems) in the R89 machine room has been intermittently faulty causing some system restarts. This system manages the pumps. In one case the chillers failed to restart. It was planned to replace a faulty card in the system this morning (24th Jan) - which should have taken around 30minutes. However, a test yesterday with one of the pumps running ‘stand-alone’ failed. During the start-up procedure the remaining pumps ramped-up to balance the water flow-rate to compensate for the loss of one pump. This caused a power surge to pump4 and it tripped causing damage to the fuse carrier. This is now being looked at and a new date for the BMS board swap will be planned.
Current operational status and issues
  • Ongoing security patching.
Resolved Castor Disk Server Issues
  • gdss736 (lhcbDst - D1T0) – rebuilt and back in production (RO)
  • gdss776 (lhcbDst - D1T0) - Failed Wednesday afternoon (17th). Returned to service on Friday.)
Ongoing Castor Disk Server Issues
  • gdss717 (CMSTape - D0T1) – multiple drive failure
Limits on concurrent batch system jobs.
  • CMS Multicore 550
Notable Changes made since the last meeting.
  • Security patching done or underway.
  • The WMS service has been declared as not in production in the GOC DB.
  • Updating Echo CEPH to the "Luminous" version underway. The service will continue to operate during this intervention.
Entries in GOC DB starting since the last report.
Service Scheduled? Outage/At Risk Start End Duration Reason
srm-atlas.gridpp.rl.ac.uk, UNSCHEDULED OUTAGE 22/01/2018 22:00 23/01/2018 10:00 12 hours urgent fixes needed on Oracle DB backend - extension
srm-atlas.gridpp.rl.ac.uk UNSCHEDULED OUTAGE 22/01/2018 17:45 22/01/2018 22:00 4 hours and 15 minutes urgent fixes needed on Oracle DB backend
srm-alice.gridpp.rl.ac.uk, srm-atlas.gridpp.rl.ac.uk, srm-biomed.gridpp.rl.ac.uk, srm-cert.gridpp.rl.ac.uk, srm-cms-disk.gridpp.rl.ac.uk, srm-cms.gridpp.rl.ac.uk, srm-dteam.gridpp.rl.ac.uk, srm-ilc.gridpp.rl.ac.uk, srm-lhcb.gridpp.rl.ac.uk, srm-lhcb.gridpp.rl.ac.uk, srm-mice.gridpp.rl.ac.uk, srm-minos.gridpp.rl.ac.uk, srm-na62.gridpp.rl.ac.uk, srm-pheno.gridpp.rl.ac.uk, srm-preprod.gridpp.rl.ac.uk, srm-snoplus.gridpp.rl.ac.uk, srm-solid.gridpp.rl.ac.uk, srm-t2k.gridpp.rl.ac.uk, SCHEDULED OUTAGE 17/01/2018 10:00 17/01/2018 13:00 3 hours Outage of Castor Storage to apply Security patches.
lcglb01.gridpp.rl.ac.uk, lcglb02.gridpp.rl.ac.uk, lcgwms04.gridpp.rl.ac.uk, lcgwms05.gridpp.rl.ac.uk, SCHEDULED OUTAGE 12/01/2018 10:00 19/01/2018 12:00 7 days, 2 hours WMS Decommissioning RAL Tier1
Declared in the GOC DB
Service Scheduled? Outage/At Risk Start End Duration Reason
srm-atlas.gridpp.rl.ac.uk, UNSCHEDULED OUTAGE 23/01/2018 15:45 24/01/2018 14:00 22 hours and 15 minutes emergency downtime of Castor Atlas while rebuilding some database tables
Advanced warning for other interventions
The following items are being discussed and are still to be formally scheduled and announced.

Ongoing or Pending - but not yet formally announced:

  • Update to next CEPH version ("Luminous"). Ongoing.

Listing by category:

  • Castor:
    • Update systems to use SL7 and configured by Quattor/Aquilon. (Tape servers done)
    • Move to generic Castor headnodes.
  • Echo:
    • Update to next CEPH version ("Luminous"). Ongoing.
  • Networking
    • Extend the number of services on the production network with IPv6 dual stack. (Done for Perfsonar, FTS3, all squids and the CVMFS Stratum-1 servers).
  • Internal
    • DNS servers will be rolled out within the Tier1 network.
    • Infrastructure
  • Testing of power distribution boards in the R89 machine room is being scheduled for some time late July / Early August. The effect of this on our services is being discussed.
Open GGUS Tickets (Snapshot during morning of meeting)
Request id Affected vo Status Priority Date of creation Last update Type of problem Subject
117683 none on hold less urgent 18/11/2015 03/01/2018 Information System CASTOR at RAL not publishing GLUE 2
124876 ops on hold less urgent 07/11/2016 13/11/2017 Operations [Rod Dashboard] Issue detected : hr.srce.GridFTP-Transfer-ops@gridftp.echo.stfc.ac.uk
127597 cms on hold urgent 07/04/2017 05/10/2017 File Transfer Check networking and xrootd RAL-CERN performance
132589 lhcb in progress very urgent 21/12/2017 24/01/2018 Local Batch System Killed pilots at RAL
132712 other in progress less urgent 04/01/2018 23/01/2018 Other support for the hyperk VO (RAL-LCG2)
132802 cms in progress urgent 11/01/2018 24/01/2018 CMS_AAA WAN Access Low HC xrootd success rates at T1_UK_RAL
132830 cms reopened very urgent 12/01/2018 24/01/2018 CMS_AAA WAN Access Reading issues T1_UK_RAL
132844 atlas in progress urgent 14/01/2018 19/01/2018 Storage Systems UK RAL-LCG2 DATADISK transfer errors "DESTINATION OVERWRITE srm-ifce err:"
132935 atlas in progress less urgent 18/01/2018 22/01/2018 Storage Systems UK RAL-LCG2: deletion errors
Availability Report
Day OPS Alice Atlas CMS LHCb Atlas Echo Comment
17/01/18 87.15 100 100 100 100 100
18/01/18 100 100 100 100 98 100
19/01/18 100 100 100 100 100 100
20/01/18 100 92 100 100 100 100
21/01/18 100 0 100 100 100 100
22/01/18 100 49 65 100 89 100
23/01/18 100 94 0 100 94 100
Hammercloud Test Report

Key: Atlas HC = Atlas HammerCloud (Queue ANALY_RAL_UCORE, Template 841); CMS HC = CMS HammerCloud

Day Atlas HC CMS HC Comment
17/01/18 92 100
18/01/18 100 100
19/01/18 100 99
20/01/18 100 99
21/01/18 100 100
22/01/18 59 100
23/01/18 100 99
Notes from Meeting.
  • There was a discussion around performance of the Tier1 when accessing off-site data from the worker nodes. This data path goes through the site firewall which is causing some restrictions.