RAL Tier1 Operations Report for 25th May 2016
Review of Issues during the week 18th to 25th May 2016.
|
- On Friday (20th) Atlas put batch queues offline. This can be seen in the HammerCloud results below. Problem traced to Atlas Castor instance which was fixed during the day.
- Also on Friday (20th) there was a problem seen on a disk servers in CMSDisk that was running out of xroot connections. This limit was increased although that change was not picked up by all servers in the service class. There was a further problem with one disk server over the weekend which was resolved on Monday by forcing the recent change to be picked up by all (CMSDisk) disk servers.
- Atlas put our batch queues offline yesterday (24th May). This was chased to an error in the new code that creates the 'machine job features'. This was disabled this morning.
- This morning (25th) we have a problem with the Tape Library control software that is crashing. This means tape access is affected and a call is open with Oracle.
Resolved Disk Server Issues
|
- GDSS635 (AtlasTape - D0T1) which had failed in the early hours of the 20th April was returned to service on Monday afternoon (23rd May). All its disk have been replaced and it has re-run the acceptance tests for a week. However, the next day, Tuesday (24th), the server was found not to be working correctly. It was taken out of production for a couple of hours while a software problem was identified and fixed.
- GDSS619 (GenTape - D0T1) failed on 26th April. As for GDSS635 (above) all its disk have been replaced and it has re-run the acceptance tests for a week. It was also returned to service on Monday (23rd May).
- GDSS664 (AtlasTape - D0T1) failed on Monday morning, 16th May. A verification of the RAID array found a faulty disk which has been replaced. It was returned to service last Wednesday (18th May).
- GDSS727 (CMSDIsk reported 'FSProbe' problems on Thursday (19th May) and was taken out or service. Following the replacement of two disk drives it was returned to service on Sunday evening (22nd May).
Current operational status and issues
|
- There is a problem seen by LHCb of a low but persistent rate of failure when copying the results of batch jobs to Castor. There is also a further problem that sometimes occurs when these (failed) writes are attempted to storage at other sites. A recent modification has improved, but not completed fixed this.
- The intermittent, low-level, load-related packet loss seen over external connections is still being tracked. Likewise we have been working to understand some remaining low level of packet loss seen within a part of our Tier1 network.
Ongoing Disk Server Issues
|
- The first two disk servers listed below have failed more than once and are being put through a week long acceptance test.
- GDSS635 (AtlasTape - D0T1) failed in the early hours of the 20th April.
- GDSS619 (GenTape - D0T1) failed on 26th April only a few hours after being put back into service after a previous failure.
- GDSS664 (AtlasTape - D0T1) failed on Monday morning, 16th May. A verification of the RAID array found a faulty disk which has been replaced. Expected back in production later today.
Notable Changes made since the last meeting.
|
- The migration of Atlas data from "C" to "D" tapes is continuing.
- Write access to GenScratch has been stopped.
- One (of two) new batches of Worker Nodes has bee put into production on Monday/Tuesday of this week.
Service
|
Scheduled?
|
Outage/At Risk
|
Start
|
End
|
Duration
|
Reason
|
lcgwms06.gridpp.rl.ac.uk,
|
SCHEDULED
|
OUTAGE
|
01/06/2016 11:00
|
30/06/2016 11:00
|
29 days,
|
Server lcgwms06.gridpp.rl.ac.uk Decommissioning
|
Advanced warning for other interventions
|
The following items are being discussed and are still to be formally scheduled and announced.
|
- The Castor 2.1.15 update is pending. Testing has shown a database related performance issue which is being followed up. We await successful resolution of that problem and completion of testing before scheduling. Following advice from the developers we will not upgrade the SRMs before the Castor 2.1.15 upgrade.
- Decommissioning of "GEN Scratch" storage in Castor. (Formally announced by EGI broadcast). Write access to this area has now been stopped in preparation for completely stopping access on the 20th June.
- Decommissioning of lcgwms06. This will leave two WMS systems remaining in service.
- The HAProxy Load Balancer will be added in front of the Production FTS3 Service.
Listing by category:
- Databases:
- Switch LFC/3D to new Database Infrastructure.
- Castor:
- Update SRMs to new version (includes updating to SL6).
- Update to Castor version 2.1.15.
- Migration of data from T10KC to T10KD tapes (Affects Atlas & LHCb data).
- Networking:
- Replace the UKLight Router. Then upgrade the 'bypass' link to the RAL border routers to 2*40Gbit.
- Fabric
- Firmware updates on remaining EMC disk arrays (Castor, LFC)
- Grid Services
- Once the use of the Load Balancer (HAProxy) has been proven for the test FTS service it will be extended to other services.
Entries in GOC DB starting since the last report.
|
Open GGUS Tickets (Snapshot during morning of meeting)
|
GGUS ID |
Level |
Urgency |
State |
Creation |
Last Update |
VO |
Subject
|
121687
|
Green
|
Less Urgent
|
On Hold
|
2016-05-20
|
2016-05-23
|
|
packet loss problems seen on RAL-LCG perfsonar
|
120954
|
Green
|
Less Urgent
|
In Progress
|
2016-04-21
|
2016-05-24
|
LHCb
|
SRM endpoint entry in GOCDB
|
120920
|
Green
|
Less Urgent
|
In Progress
|
2016-04-19
|
2016-05-06
|
SNO+
|
XRootD issues at RAL
|
120810
|
Yellow
|
Urgent
|
In Progress
|
2016-04-13
|
2016-05-24
|
Biomed
|
Decommissioning of SE srm-biomed.gridpp.rl.ac.uk - forbid write access for biomed users
|
120350
|
Green
|
Less Urgent
|
In Progress
|
2016-03-22
|
2016-05-06
|
LSST
|
Enable LSST at RAL
|
119841
|
Amber
|
Less Urgent
|
On Hold
|
2016-03-01
|
2016-04-26
|
LHCb
|
HTTP support for lcgcadm04.gridpp.rl.ac.uk
|
117683
|
Yellow
|
Less Urgent
|
On Hold
|
2015-11-18
|
2016-04-05
|
|
CASTOR at RAL not publishing GLUE 2
|
Key: Atlas HC = Atlas HammerCloud (Queue ANALY_RAL_SL6, Template 729); CMS HC = CMS HammerCloud
Day |
OPS |
Alice |
Atlas |
CMS |
LHCb |
Atlas HC |
CMS HC |
Comment
|
18/05/16 |
100 |
100 |
100 |
100 |
100 |
100 |
100 |
|
19/05/16 |
100 |
100 |
100 |
100 |
100 |
100 |
100 |
|
20/05/16 |
100 |
100 |
100 |
97 |
100 |
86 |
76 |
Problem with test on all CEs. Looks like a CMS problem.
|
21/05/16 |
100 |
100 |
28 |
98 |
100 |
99 |
90 |
Atlas: Central monitoring problem; CMS: Single SRM test failure ("Input/output error")
|
22/05/16 |
100 |
100 |
76 |
100 |
100 |
100 |
94 |
Atlas: Central monitoring problem
|
23/05/16 |
100 |
100 |
100 |
98 |
100 |
100 |
100 |
Single SRM Test failure: User timeout.
|
24/05/16 |
100 |
100 |
100 |
100 |
96 |
98 |
100 |
Single SRM Test failure: (No such file or directory)
|