Is a reboot a change?
Today I got asked that old perennial: Is a reboot a change?
This is an endless source of debate and will never be resolved. The answer is whatever your organisation decides it is.
Personally I think it is a change. i don't understand why this is debated. But is is.
Rebooting a server is a conscious act to change the state of a system, with predictable potential impacts.
It should be planned, with backout recovery plan.
It should be tested
It should be risk assessed
It should be agreed and approved
It should be scheduled so as not to impact the business or other changes
I hope you are familiar with standard change.
If the risk is low, then a server reboot should be agreed to be a standard change, pre-approved and pre-documented.
So that at a very minimum we have a change record.
So that when the service suddenly stops we can rapidly identify a primary cause.
To me this is obvious, but there is not industry consensus.
In a DevOps world I might think differently. There, servers are cattle not pets, the restart has been rehearsed in near identical environments, the tool chain creates an audit trail, and recovery is quick and seamless.
In the Real IT in which most of us operate, none of that is true. Every restart of a server has a level of uncertainty and a potential for significant knock-on effects. It's a change.
[From comments:
Again, a reboot can be a standard change: create the record , get any business approvals if required, put required risk mitigation in place, then do it.
if it is a "pet" - a fragile snowflake server - then I'd want a fail-over image of it standing by.
If it is " cattle" then no biggie. If it doesnt come up cleanly then slaughter it because you will load balance over to others in the "herd".
There is no simplistic answer to "is a reboot a change" because it depends so much on the nature of the server and the organisation in which it lives.
Made in New Zealand 
