Fix mmr-hammer

[adehnert]

mmr-hammer is apparently wrong, and should possibly be fixed or deprecated or something.

Discussion:

-c scripts -i r-m.d 2013-01-01 ~19:40:

Mitch says:

it was wrong from the get-go. The original thing it was derived from that we found on the web never even set up initial replication properly the way the documentation tells you to, and caused race conditions. I think there were other bugs in that thing, too, but I'd have to go find it and cry a bit.

[...]

The cleanruv thing is a somewhat separate issue that amounts to us having had no idea how to decommission an MMR peer, and using that tool to try to fix stuff by reinitializing replication simply kept reintroducing the problem that would cause the cluster to fall apart

days later each time. Other replication problems are fixable by reinitializing machines, but punting and reinstalling all the agreements is silly and not really related to that process.

mmr-hammer has previously come up on -i ldap on 2011-11-13 01:33 (and the next day) with some discussion of its issues, on -i wizard on 2012-07-21 21:25, and on -i o-f on 2012-08-01 00:21. Most of them do not have much in terms of details of its issues.