Tier1 Operations Report 2017-05-31

RAL Tier1 Operations Report for 31st May 2017

Review of Issues during the week 24th to 31st May 2017.

A problem of ARP poisoning on the Tier1 network is affecting some monitoring used by Echo. The cause of this is now believed to be understood and the fix (a setting change in the OPN Router) has been applied.
There is a problem on the site firewall which is causing problems for some specific data flows. This was being investigated in connection with videoconferencing problems. It is expected that this is having an effect on our data that flows through the firewall (such as to/from worker nodes).
Atlas carried out a large deletion test of files in Echo last week. The bulk were deleted successfully and in a reasonable timescale. However, some (one to two thousand) files failed to delete. Most of these were cleaned up manually leaving a handful for debugging purposes. However, the system subsequently also deleted these remaining files without manual intervention.

Resolved Disk Server Issues

GDSS773 (LHCbDst - D1T0) crashed on Sunday (21st May). It was returned to service Thursday morning (25th) although no fault was found during the diagnostic testing.
GDSS804 (LHCbDst - D1T0) was taken out of production on Tuesday 23rd as it was showing memory errors. However, the memory tests then failed to find anything and it was returned to service the following day.

Current operational status and issues

We are still seeing a rate of failures of the CMS SAM tests against the SRM. These are affecting our (CMS) availabilities. CMS are also looking at file access performance and have turned off "lazy-download". The CMS SRM SAM test success rate has improved since the Castor 2.1.16 upgrade on the 25th May, although is still not 100%. It is still planned to re-visit this issue now Castor has been upgraded.
LHCb Castor performance has been OK since the 2.1.16 update although this was not under and high load. A load test (mimicking the stripping/merging campaign) was carried out last Wednesday (24th May). This was successful in that the specific performance/timeout problem seen before the Castor 2.1.16 upgrade was not seen. The main limitations the test hit within Castor was load on the older disk servers in the instance. Following the 2.1.16 upgrade and this (successful) test this item is being removed from the list of ongoing issues.

Ongoing Disk Server Issues

GDSS658 (AtlasScratchDisk - D1T0) crashed yesterday afternoon (30th May). It is still undergoing tests.

Limits on concurrent batch system jobs.

Notable Changes made since the last meeting.

CMS Castor instance updated to Castor version 2.1.16-13 and the CMS SRMs also upgraded to version 2.1.16.
For LHCb Castor the xroot manager was installed on the (LHCb) stager. This resolves a problem introduced by version 2.1.16 (and reported last week) that affected LHCb where a TURL returned by the SRM did not always work when used for xroot access owing to an incorrect hostname.
CEs being migrated to use the load balancers in front of the argus service.
Batch access has been enabled for LIGO and the MICE pilot role.

Declared in the GOC DB

Service	Scheduled?	Outage/At Risk	Start	End	Duration	Reason
Castor GEN instace (srm-alice, srm-biomed, srm-dteam, srm-ilc, srm-mice, srm-na62, srm-pheno, srm-snoplus, srm-t2k)	SCHEDULED	OUTAGE	31/05/2017 10:00	31/05/2017 15:00	5 hours	Upgrade of Castor GEN instance to version 2.1.16.

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:

Update Castor (including SRMs) to version 2.1.16. Central nameserver and LHCb and Atlas stagers done. Current plan: CMS stager and SRMs on Thursday 25th May. GEN to follow.
Update Castor SRMs - CMS & GEN still to do. These are being done at the same time as the Catsor 2.1.16 updates.

Listing by category:

Castor:
- Update SRMs to new version, including updating to SL6.
- Update Castor to version 2.1.16 (ongoing)
- Merge AtlasScratchDisk into larger Atlas disk pool.
Networking
- Increase OPN link to CERN from 2*10Gbit to 3*10Gbit links.
- Enable first services on production network with IPv6 now that the addressing scheme has been agreed. (Perfsonar already working over IPv6).
Services
- Put argus systems behind a load balancers to improve resilience.
- The production FTS needs updating. This will no longer support the soap interface. (The "test" FTS , used by Atlas, has already been upgraded.)

Entries in GOC DB starting since the last report.

Service	Scheduled?	Outage/At Risk	Start	End	Duration	Reason
Castor GEN instace (srm-alice, srm-biomed, srm-dteam, srm-ilc, srm-mice, srm-na62, srm-pheno, srm-snoplus, srm-t2k)	SCHEDULED	OUTAGE	31/05/2017 10:00	31/05/2017 15:00	5 hours	Upgrade of Castor GEN instance to version 2.1.16.
srm-cms-disk.gridpp.rl.ac.uk	SCHEDULED	OUTAGE	25/05/2017 10:00	25/05/2017 13:03	3 hours and 3 minutes	Upgrade of CMS Castor instance to version 2.1.16.

Open GGUS Tickets (Snapshot during morning of meeting)

Availability Report

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


Day	OPS	Alice	Atlas	CMS	LHCb	Atlas Echo	Atlas HC	Atlas HC Echo	CMS HC	Comment
24/05/17	100	100	100	97	100	100	99	99	100	SRM Test failures (user timeout)
25/05/17	100	100	100	87	100	100	100	100	98	SRM Test failures (user timeout)
26/05/17	100	100	98	98	100	100	100	100	100	SRM Test failures: Atlas: One error ‘User belonging to VO not authorized to access space token ATLASDATADISK; CMS: user timeout.
27/05/17	100	100	100	98	100	100	100	100	100	SRM Test failures (user timeout)
28/05/17	100	100	100	97	100	100	100	100	100	SRM Test failures (user timeout)
29/05/17	100	100	100	98	100	100	98	100	100	SRM Test failures (user timeout)
30/05/17	100	100	100	94	100	100	92	100	100	SRM Test failures (user timeout)

Notes from Meeting.