Skip to content

What do we do when the leap apt repository fails?

We need to figure out what we should do when http://deb.leap.se is not available. This happened the other day, when apache was down on that machine and I was trying to deploy to a node.

With other apt sources, we have the fall-back method, but with the leap sources we have packages that are only in that repository that will make it so the rest of the deployment wont work. Although I think maybe it is only haproxy right now, and later we will not be depending on haproxy being in our repository.... but depending on when providers deploy and when things are available, it is not unlikely that we will have packages in the leap repository that are required for deployment.

It seems somewhat bad to have the reliability of our infrastructure be tied to deployments functioning properly. If our goal is to make a deployment that fails on a non-responsive apt source be more resilient to those failures, or at least fail in a less bloody way, we need to decide what to do when our apt repository isn't available.

Since it is a hard dependency right now, I feel like the best thing we can do right now is to try and fail more gracefully in the apt wrapper that elijah is working on.

(from redmine: created on 2013-04-30, closed on 2013-04-30, relates #1091)