Per bug 798517, mradm01 was shut down yesterday. However, from what I can tell these nameservers are still slaves to mradm01, meaning that they're not getting updates now (e.g., to mozilla.org), and that they'll stop answering next week since their data is stale.
I turned mradm01 back on for now, since we'll need to do some changes to the nameservers in scl1 as well and shouldn't be done on a Friday.
I've updated the configuration on ns1b.build.mtv1 to no longer depend on mradm01. After this host has been tested I'll migrate the keepalive vip from ns1a and apply the same configuration changes there.
From all the tests I can see, it's giving the right results. Turn it on! By the way, admin1a/b.infra.scl1 are also slaves.
I've applied the same changes to ns1a.build.mtv1 and temporarily disabled keepalived to prevent the vip from moving back. Once validated I'll re-enable keepalived. I'll look into admin1[ab].infra.scl1 now
Great, thanks! I've re-enabled keepalived and confirmed the vip properly migrated back to ns1a and is answering queries. I've filed 792586 to track admin1[ab].infra.scl1