The moral of the story? For high availability you should have a testing environment that closely mirrors production and build redundancy into your production operations so that you may update/patch systems.
Also, if your business depends on a single customer, don’t fuck up. :)
I don’t know the whole story, but it seems if a router crashes whenever you run a certain command, you should stop running that command on all your other routers.
Pssh, devops means you can run a foot shooting command on all the routers at once. >.<