Clarify how to apply chef-repo updates during failover
Noted in the 2018-07-24 failover
The chef-repo
contents and then running chef-client
on a set of machines.
In staging, we don't merge the MRs, but make the changes available to the run through a peculiar command I didn't quite catch.
In production, we expect to merge the MRs.
We should update the failover document so the details of what to do are explicit, for both staging and production.
/cc @alejandro