Big production migration finished last night, closing out 2 months of very fiddly well-tested Ops work, to move a critical business service off an ephemeral IP onto a new resilient load balanced setup with no single point of failure.

Thanks to #haproxy, #freebsd, and #ansible, nobody noticed except my 2 colleagues and me, as we baby-sat the production changes, and didn’t lose a single customer request in the process.