never touch a running system
expressionops cultureIT support
The conservative operator's creed: if it works, leave it alone.
Born from hard experience with fragile systems, and half-obsolete in the patch-or-be-breached era: not touching a running system also means not patching it. The modern compromise is making change safe enough to do routinely rather than avoiding it.
Never touch a running system is the instinct that anything currently working should be left alone, and it is genuinely correct about a narrow class of situation and catastrophic as a general policy.
The narrow correctness is real: an unnecessary change to a production system that has no fault is pure risk with no benefit, and a great many outages are caused by maintenance nobody needed to perform. Change is the leading cause of incidents in most environments, which is exactly why change management exists.
The catastrophic version is the system nobody has touched for six years, running an operating system with no remaining support, with an undocumented configuration, on hardware whose replacement is no longer manufactured. It is not stable, it is fragile in a way that has not been tested, and the accumulated distance between it and everything around it grows continuously while nobody looks. That is bit rot, and the day it finally fails, the people who understood it have left and the recovery is an archaeology project rather than a restore. The defensible reading is do not change without reason, and never being able to change safely is itself the reason.
Also known as: if it ain't broke, don't fix it