Friday, 26th August 2016
Whenever we have planned (and sometimes unplanned) downtime, at work, I'm usually asked the question "While we've got the entire system down to do X, shall we do Y also?"
Typically X is planned, and we're doing major maintenance - There's one coming up when there's grid circuit maintenance, where we're hoping it'll be fine on UPS and emergency generator - with an at-risk period.
Occasionally, X is unplanned, like the time that the air conditioning failed, and everything shut down to save itself.
I always decline the option to do Y at the same time, because it violates the "Change One Thing" rule.
If I'm declaring a system outage to, say, upgrade the firmware on the core switch stack, I don't want to also take that opportunity to rewire a cabinet, or simultaneously upgrade VMware Hypervisors. I'll declare another outage for those individually.
The problem with breaking the Change One Thing rule is that if you change two things, and something doesn't work quite right afterwards, you can't be 100% certain which to roll back, and it'll typically take N times longer to fix (where N is the number of things you changed).
So I'm a bit of a stickler for this. I don't really enjoy giving up a weekend to work on a system when nobody else is using it but I'd rather do one task at a time, and get it right, and be able to make the most of the rest of the weekend; as opposed to having to pick through the permutations of the things that've changed, to try to restore service before 9AM on a Monday morning.
Fortunately, I've got my team quite well trained not to get distracted from the One Thing we're doing when Outage Time comes. I can almost guarantee though, that someone will ask "Can we do 'this other thing' at the same time?".
To which, my answer will always be: No.
That can wait for another day, and we shall do that, and only that, then.
Tom was once mistaken for the director general of Black Mesa.