Difference between revisions of "Tier1 Operations Report 2017-09-20"

From GridPP Wiki
Jump to: navigation, search
()
()
 
Line 316: Line 316:
 
|}
 
|}
 
* The Dell '16 batch of storage for Echo (CEPH) has almost completed its testing with the Fabric Team. It will then be handed over to the Echo team for installation into the development CEPH cluster for further testing.
 
* The Dell '16 batch of storage for Echo (CEPH) has almost completed its testing with the Fabric Team. It will then be handed over to the Echo team for installation into the development CEPH cluster for further testing.
* There was a discussion about the testing of disk servers that have stood idle before being installed into Echo (Ceph). The servers currently giving problems were put into a Ceph test area and exercised. However, this was clearly insufficient to shake out enough of the disk problems.
+
* There was a discussion about the testing of disk servers that have stood idle before being installed into Echo (Ceph). The servers currently giving problems were put into a Ceph test area and exercised. However, this was clearly insufficient to shake out enough of the disk problems and the process will be re-visited.
 
* Discussions are underway with ALICE about using Echo.
 
* Discussions are underway with ALICE about using Echo.
 
* We will work with CMS to find the 'dark data' that is in Castor but not known to CMS. This would fix the problems of CMSDisk filling.
 
* We will work with CMS to find the 'dark data' that is in Castor but not known to CMS. This would fix the problems of CMSDisk filling.

Latest revision as of 15:50, 20 September 2017

RAL Tier1 Operations Report for 20th September 2017

Review of Issues during the fortnight 6th to 20th September 2017.
  • The better understanding of CEPH problems with Echo has provided stability. However, careful management of the disks in servers recently added into Echo has been needed. This has slowed the rate at which the new hardware is being brought into full use.
  • We have seen problems with the LHCb SRM SAM tests since the 8th September. However, LHCb are not seeing operational problems.
  • The CMSDisk area in Castor became full over the weekend 9/10 Sep. This caused SAM test failures.
  • There have been three cases where the network link to one of the batches of worker nodes (Dell '16) has dropped out and needed resetting. Switch firmware updates were applied yesterday (19th Sep) to try and fix this.
  • There was a problem with the LFC service. A new node was added to the alias but it was not configured to support dteam VO. This was corrected in response to a GGUS ticket.
Resolved Disk Server Issues
  • GDSS762 (AtlasDataDisk - D1T0) Failed on Wednesday evening 6th Sep. It was returned to service on the Friday (8th) although no cause for the failure was identified.
  • GDSS732 (LHCbDst - D1T0) failed on Sunday 10th Sep. It would not boot and the OS was re-installed. It was returned to service on Tuesday 12th Sep. Two files were lost.
  • GDSS769 (AtlasDataDisk - D1T0) Also failed on Sunday 10th Sep. It was returned to service the following day (11th) although no cause for the failure was identified.
  • GDSS776 (LHCbDst - D1T0) failed on Monday 11th Sep. It was returned to service the following day (12th). One faulty disk drive was replaced.
  • GDSS748 (AtlasDataDisk - D1T0) failed on Sunday 17th Sep. It was returned to service at the following afternoon (18th). It had reported read-only file systems and the OS needed to be reinstalled. Four files in flight at the time of the failure were declared lost to Atlas.
Current operational status and issues
  • The high rate of SRM SAM failures for CMS stopped in mid-August although why this improvement came about is not understood. However, we will no longer track this issue here.
  • There is a problem on the site firewall which is causing problems for some specific data flows. Discussions have been taking place with the vendor. This is having an effect on our data that flows through the firewall (such as to/from worker nodes).
Ongoing Disk Server Issues
  • None
Limits on concurrent batch system jobs.
  • CMS Multicore 550
Notable Changes made since the last meeting.
  • Re-deployed six disk servers from 2012 batches into lhcbRawRdst (D0-T1).
  • Enabled access to Castor for the SOLID experiment
Declared in the GOC DB
  • None
Advanced warning for other interventions
The following items are being discussed and are still to be formally scheduled and announced.

Pending - but not yet formally announced:

  • Upgrade the FTS3 service to a version that will no longer support the SOAP interface. The SOAP interface was disabled on Monday (17th July).
  • Re-distribute the data in Echo onto the 2015 capacity hardware. (Ongoing)

Listing by category:

  • Castor:
    • Move to generic Castor headnodes.
  • Echo:
    • Re-distribute the data in Echo onto the remaining 2015 capacity hardware.
  • Networking
    • Extend the number of services on the production network with IPv6 dual stack. (Done for Perfsonar, all squids and the CVMFS Stratum-1 servers).
  • Services
    • The production FTS will be updated now the requirement to support the deprecated SOAP interface has been removed.
Entries in GOC DB starting since the last report.
  • None
Open GGUS Tickets (Snapshot during morning of meeting)
GGUS ID Level Urgency State Creation Last Update VO Subject
130573 Green Urgent In Progress 2017-09-14 2017-09-18 CMS Possible missing files at T1_UK_RAL_Disk
130537 Green Less Urgent In Progress 2017-09-13 2017-09-13 OPS [Rod Dashboard] Issue detected : org.sam.SRM-GetSURLs-ops@srm-solid.gridpp.rl.ac.uk
130467 Green Urgent Waiting for Reply 2017-09-10 2017-09-10 CMS T1_UK_RAL has SAM3 CE and SRM critical > 2hours
130207 Green Urgent In Progress 2017-08-24 2017-09-06 MICE Timeouts when copying MICE reco data to CASTOR
130193 Green Urgent In Progress 2017-08-23 2017-09-04 CMS Staging from RAL tape systems
128991 Green Less Urgent Waiting for Reply 2017-06-16 2017-09-18 Solid solidexperiment.org CASTOR tape support
127597 Red Urgent On Hold 2017-04-07 2017-06-14 CMS Check networking and xrootd RAL-CERN performance
124876 Red Less Urgent On Hold 2016-11-07 2017-01-01 OPS [Rod Dashboard] Issue detected : hr.srce.GridFTP-Transfer-ops@gridftp.echo.stfc.ac.uk
117683 Red Less Urgent On Hold 2015-11-18 2017-07-06 CASTOR at RAL not publishing GLUE 2.
Availability Report
Day OPS Alice Atlas CMS LHCb Atlas Echo Comment
06/09/17 100 100 100 100 100 100
07/09/17 100 100 100 100 100 100
08/09/17 100 100 100 97 88 100 CMS: Disk full; LHCb: [SRM_INVALID_PATH] No such file or directory
09/09/17 100 100 100 42 88 100 CMS: Disk full; LHCb: [SRM_INVALID_PATH] No such file or directory
10/09/17 100 100 100 56 75 100 CMS: Disk full; LHCb: [SRM_INVALID_PATH] No such file or directory
11/09/17 100 100 98 100 83 100 Atlas: Single test failure: could not open connection to srm-atlas.gridpp.rl.ac.uk:8443; LHCb: [SRM_INVALID_PATH] No such file or directory
12/09/17 100 100 98 100 84 100 Atlas: Single test failure: error 500 Command failed. : an unknown error occurred; LHCb: [SRM_INVALID_PATH] No such file or directory
13/09/17 100 100 100 100 76 100 [SRM_INVALID_PATH] No such file or directory
14/09/17 100 100 100 100 91 100 [SRM_INVALID_PATH] No such file or directory
15/09/17 100 100 100 100 72 100 [SRM_INVALID_PATH] No such file or directory
16/09/17 100 100 100 100 83 100 [SRM_INVALID_PATH] No such file or directory
17/09/17 100 100 100 100 88 100 [SRM_INVALID_PATH] No such file or directory
18/09/17 100 100 100 100 67 100 [SRM_INVALID_PATH] No such file or directory
19/09/17 100 100 100 100 84 100 [SRM_INVALID_PATH] No such file or directory
Hammercloud Test Report

Key: Atlas HC = Atlas HammerCloud (Queue ANALY_RAL_SL6, Template 845); Atlas HC Echo = Atlas Echo (Template 841);CMS HC = CMS HammerCloud

Day Atlas HC Atlas HC Echo CMS HC Comment
06/09/17 100 100 99
07/09/17 100 100 100
08/09/17 99 100 100
09/09/17 100 100 100
10/09/17 99 100 100
11/09/17 100 100 100
12/09/17 100 100 100
13/09/17 96 96 100
14/09/17 88 100 99
15/09/17 100 100 99
16/09/17 100 100 100
17/09/17 98 100 100
18/09/17 85 89 100
19/09/17 97 100 100
Notes from Meeting.
  • The Dell '16 batch of storage for Echo (CEPH) has almost completed its testing with the Fabric Team. It will then be handed over to the Echo team for installation into the development CEPH cluster for further testing.
  • There was a discussion about the testing of disk servers that have stood idle before being installed into Echo (Ceph). The servers currently giving problems were put into a Ceph test area and exercised. However, this was clearly insufficient to shake out enough of the disk problems and the process will be re-visited.
  • Discussions are underway with ALICE about using Echo.
  • We will work with CMS to find the 'dark data' that is in Castor but not known to CMS. This would fix the problems of CMSDisk filling.