Rules for live systems.
WHAT TO ESTABLISH BEFORE TOUCHING ANYTHING
Which machine, and what depends on it.
WHAT TO DO FIRST
Confirm there is a recent backup.
WHAT TO PREFER
A change tested elsewhere first.
WHAT TO AVOID ENTIRELY
Experimenting on a production machine.
WHAT TO DO WHEN THAT IS UNAVOIDABLE
Announce it, schedule it, and prepare a way back.
WHAT TO WRITE DOWN BEFORE STARTING
The commands you intend to run.
WHY
Improvising under pressure is where mistakes happen.
WHAT TO DO AT EACH STEP
Verify the result before continuing.
WHAT TO NEVER DO
Chain destructive commands together Run something you have not read Work while exhausted or rushed
WHY THE LAST ONE MATTERS
Most serious incidents involve someone tired and in a hurry.
WHAT TO DO IF SOMETHING GOES WRONG
Stop immediately.
WHY STOPPING FIRST
Continuing usually compounds it.
WHAT TO DO NEXT
Establish what state things are actually in, before acting.
WHAT TO COMMUNICATE
That there is a problem, promptly.
WHY
Concealing it costs more than the fault.
WHAT TO RECORD AFTERWARDS
What happened and what prevents it recurring.
WHAT TO ACCEPT
That everyone breaks something eventually, and the difference is preparation.
