What to do when a server dies
In General
- Try to connect from another machine - you could have been automatically firewalled.
- Notify yoyo@.
- Try to fix it.
Specific servers
Codd
- Contact Union ITS to find out what's happened.
- Mail should be queued on pat and backus automatically for at least a week.
- IRC is at risk, with backup on Backus.
Backus
- (First, how are you reading this?)
- Visit the DCS cupboard.
- In the event of extended downtime, Backus will have to be made the primary DNS server, so that changes can be made to the configuration. Codd won't do, because it's not ns2.
- IRC is at risk, with backup on Codd.
- When it comes back, restart IRC services.
- In the event of extended downtime, restore the SVN repository and possibly Trac from the latest backups, on another server. Be sure to change the DNS to move {svn,trac}.uwcs.co.uk.
Pat
- Check pat's admin console.
- Redirect outgoing mail from molotov to be sent via the Warwick mail relays (or 'som?? - untested).
Consider setting up a BackupMailRelay for incoming mail from off-campus.
- Consider whether DNS services are at risk.
Insomnia
- Contact ITS to find out what's happened.
Consider setting up a BackupMailRelay for incoming mail from off-campus.