needs to perform maintenance on servers for various reasons, sometimes even urgently:
- operating system updates
- applying security fixes
- changes to server room or virtualization infrastructure
We try to do maintenance only when necessary and keep downtime to a minimum.
Each server has an owner
which is the primary contact for isginf
for maintenance and other issues. If the owner
is unavailable we fall back to the ITC
of the group.
For each server we negotiate a policy for downtime and maintenance. The two general policies are outlined below.
This is the preferred way. isginf
will just perform the necessary maintenance and power-cycle the system when we need to.
An automated mail notification to the owner and additional addresses on start-up can be requested by the server owner.
Coordination with Owner Required
For systems where there is no other way we will coordinate the maintenance tasks with the server owner or ITC
. Reasons why this may be the only option include:
- The server provides an critical service and unannounced downtime is unacceptable.
- Updates are likely to break something and require staging and testing.
- The server runs long-running jobs that cannot be interrupted.