From: glast-sccs-planning-l@slac.stanford.edu [mailto:glast-sccs-planning-l@slac.stanford.edu] On Behalf Of Adesanya, Adeyemi
Sent: Friday, November 22, 2013 12:42 PM
To: glast-sccs-planning-l
Subject: Fermi 11/21 meeting notes

glastlnx--->fermilnx

===============

Making progress with migration of services to fermilnx VMs. Check with users before shutting down any glastlnx machines. Instruct Tech Coords to wait a few days before unracking systems once they are shut down.

fermilnx01,2,4 should be configured as LSF clients.

Cisco will release switch firmware update in December to address our KVM link aggregation issue. Updating the switch will not require an outage.

 

Databases

=========

Brian estimates the migration work will require ~6 weeks of his time. Total wallclock time of around 3 months?

First step will be to shutdown glast-oracle04, migrate some dbs to the new server then point the test pipeline to the new databases. Bring glast-oracle04 back up and it will sync up with glast-oracle03. No production outage required.

Expect funding will be approved for glast-astro-db1 replacement ($17.2K).

 

Batch

=====

Suspect corrupt files are related to wain032. Wait to see if the problem persists afters the wains are patched on 12/4.

hequ cluster (1500 RHEL6 cores) will be unavailable week of 1/6/2014.

RHEL5 End-Of-Life is 3/31/2017. RHEL5 yilis will be retired once new bullets arrive. RHEL5 fells are also near end-of-life and are operating with a Do-Not-Resuscitate policy. Discuss OS support in more detail during the Town Hall for Unix Users on 12/12.

 

AOB

====

Rob mentioned that Paul Hertz will be visiting SLAC on 12/11. Set up a tour of building 50 Datacenter ~9-9:30am?

 

------

Yemi

  • No labels