RAL Tier1 Operations Report for 14th October 2015
Review of Issues during the week 7th to 14th October 2015.
|
- There was a problem running batch jobs that make use of glexec over the weekend. This was traced to a missing component on the new worker node configuration. This caused a very large loss of availability for CMS. How this problem did not become apparent until the final batch of worker nodes was upgraded has not yet been understood.
- We have been working to understand the remaining low level of packet loss seen within a part of our Tier1 network. Some housekeeping tasks have been carried out as part of this.
- LHCb have reported an ongoing low but persistent rate of failure when copying the results of batch jobs to other sites. They have also reported a current problem that sometimes occurs when writing these files to our Castor storage.
Resolved Disk Server Issues
|
- GDSS636 and GDSS637 (AtlasTape - D0T1) were both taken out of service for a time on Saturday (3rd Oct). Tests were failing against these machines but subsequent analysis showed they were just suffering extremely high load so they were returned to service.
Current operational status and issues
|
- The intermittent, low-level, load-related packet loss seen over external connections is still being tracked.
- Long-standing CMS issues. The two items that remain are CMS Xroot (AAA) redirection and file open times. Work is ongoing into the Xroot redirection with a new server having been added in recent weeks. File open times using Xroot remain slow but this is a less significant problem.
Ongoing Disk Server Issues
|
- GDSS657 (LHCb_RAW - D0T1) failed on Saturday (3rd October). After being checked it was returned to service the following day in read-only mode. However, investigations to fix a problem on this server (failed battery on the RAID card) during yesterday's Castor outage uncovered a further problem and the server has been kept out of service. There are no files on the server awaiting migration - so it has no impact on access to LHCb files. Four files were found to be corrupt (partially written) from the time when the server first failed and these have been reported to LHCb.
Notable Changes made since the last meeting.
|
- The final step in the update of the Castor Oracle databases to version 11.2.0.4 took place yesterday (Tuesday 13th Oct). This was the switch-over of the production and standby versions of the "pluto" database to return these to their correct configuration.
- Five additional disk servers were added to atlasTape in order to ease the high load seen on this instance. (Tuesday 13th).
- Disk server GDSS758 was deployed into into lhcbDst (D1T0). This is a server that had been out of service for some time.
- The update of the final batch of worker nodes to the new configuration has been done.
Advanced warning for other interventions
|
The following items are being discussed and are still to be formally scheduled and announced.
|
- Upgrade of remaining Castor disk servers (those in tape-backed service classes) to SL6. This will be transparent to users.
- Complete the upgrade of the Oracle databases behind Castor to version 11.2.0.4. The first steps of this multi-step upgrade has been carried out. Further dates are declared in the GOC DB with an 'at risk' on Castor tomorrow and then a shorter outage of Castor next Tuesday (13th October).
- Some detailed internal network re-configurations to enable the removal of the old 'core' switch from our network. This includes changing the way the UKLIGHT router connects into the Tier1 network.
Listing by category:
- Databases:
- Switch LFC/3D to new Database Infrastructure.
- Update to Oracle 11.2.0.4. This will affect all services that use Oracle databases: Castor, Atlas Frontier (LFC done)
- Castor:
- Update SRMs to new version (includes updating to SL6).
- Update the Oracle databases behind Castor to version 11.2.0.4. Will require some downtimes (See above)
- Update disk servers to SL6 (ongoing)
- Update to Castor version 2.1.15.
- Networking:
- Make routing changes to allow the removal of the UKLight Router.
- Fabric
- Firmware updates on remaining EMC disk arrays (Castor, LFC)
Entries in GOC DB starting since the last report.
|
Service
|
Scheduled?
|
Outage/At Risk
|
Start
|
End
|
Duration
|
Reason
|
All Castor (All SRMs)
|
UNSCHEDULED
|
OUTAGE
|
13/10/2015 08:00
|
13/10/2015 11:08
|
3 hours and 8 minutes
|
Outage of All Castor instances during upgrade of Oracle back end database.
|
All Castor (All SRMs)
|
SCHEDULED
|
WARNING
|
08/10/2015 08:30
|
08/10/2015 15:49
|
7 hours and 19 minutes
|
Warning (At Risk) on All Castor instances during upgrade of back end Oracle database.
|
Whole site
|
SCHEDULED
|
WARNING
|
07/10/2015 10:00
|
07/10/2015 11:30
|
1 hour and 30 minutes
|
Warning on site during quarterly UPS/Generator load test.
|
Open GGUS Tickets (Snapshot during morning of meeting)
|
GGUS ID |
Level |
Urgency |
State |
Creation |
Last Update |
VO |
Subject
|
116866
|
Green
|
Less Urgent
|
In Progress
|
2015-10-12
|
2015-10-13
|
SNO+
|
snoplus support at RAL-LCG2 (pilot role)
|
116864
|
Green
|
Urgent
|
In Progress
|
2015-10-12
|
2015-10-12
|
CMS
|
T1_UK_RAL AAA opening and reading test failing again...
|
Key: Atlas HC = Atlas HammerCloud (Queue ANALY_RAL_SL6, Template 508); CMS HC = CMS HammerCloud
Day |
OPS |
Alice |
Atlas |
CMS |
LHCb |
Atlas HC |
CMS HC |
Comment
|
07/10/15 |
100 |
100 |
100 |
100 |
100 |
100 |
100 |
|
08/10/15 |
100 |
88 |
100 |
90 |
76 |
96 |
100 |
Lots of CE test failures caused by "No compatible resource found in BDII" (BDII at CERN)
|
09/10/15 |
100 |
100 |
100 |
100 |
100 |
100 |
92 |
|
10/10/15 |
100 |
100 |
100 |
100 |
100 |
100 |
100 |
|
11/10/15 |
100 |
100 |
100 |
100 |
100 |
100 |
100 |
|
12/10/15 |
100 |
100 |
100 |
100 |
100 |
97 |
n/a |
|
13/10/15 |
96.4 |
96 |
89 |
91 |
91 |
96 |
100 |
Planned outage of all of Tier1 Castor for final step of Oracle 11.2.0.4 upgrade.
|