Planning for the downtime starting monday 2007-10-29

We have a major downtime this week for dcache upgrade, rls migration, etc, etc.

 Schedule

• 08:30 dcache pools are set rdonly
• 09:00 we take down central dcache instance for upgrades
• 09:00 rls@uppsala shuts down and a database dump is created
• 09-10 while dcache is down, all pools stop and begin the upgrade
• 11:00 testing of the central services at örestaden (dcache+rls)
• 12:00 rls and central dcache services online
• 12-16 pools coming online, by 16 enough should be online to accept writes
• 16-08 spare time for delays, rest of pools coming online
• 09:00 whole-system verification that everything works as intended

Things to do on monday

• set pools rdonly
• shut down dcache
• wait for network/power maintenance to get done 
• upgrade dcache on the core nodes
• grab a new dCacheSetup from the .template and reapply all changes
• rls migration
• testing rls
• testing dcache

DQ2 stuff to do (David)

• Stop DQ2 site services on lxb7277
• After migration:
  • Change panda-lrc-server.cfg
  • Run RLS server testcases
  • Run panda.lrc.client testcases
• Change ToA file
  • NDGFLRC = 'lrc:100://rls1.ndgf.org:8001||rls1.ndgf.org:8001'
  • NDGFLRCHTTP = 'http://rls1.ndgf.org:8001/lrc/ws_'
• Once RLS is running tell Miguel to change dq2-info.cfg on voboxes to remove rls endpoint
• Start DQ2 site services