Ceph

     0> 2012-08-20 16:01:00.845740 7f85eae05700 -1 mon/Paxos.cc: In function 'void Paxos::handle_accept(MMonPaxos*)' thread 7f85eae05700 time 2012-08-20 16:01:00.843937
mon/Paxos.cc: 468: FAILED assert(state == STATE_UPDATING)

 ceph version 0.48.1argonaut (commit:a7ad701b9bd479f20429f19e6fea7373ca6bba7c)
 1: (Paxos::handle_accept(MMonPaxos*)+0x8ce) [0x4911de]
 2: (Paxos::dispatch(PaxosServiceMessage*)+0x24b) [0x49429b]
 3: (Monitor::_ms_dispatch(Message*)+0x10b0) [0x47cf30]
 4: (Monitor::ms_dispatch(Message*)+0x32) [0x48a912]
 5: (SimpleMessenger::DispatchQueue::entry()+0x903) [0x5bf483]
 6: (SimpleMessenger::dispatch_entry()+0x24) [0x5c0224]
 7: (SimpleMessenger::DispatchThread::entry()+0xd) [0x58f2cd]
 8: (()+0x7e9a) [0x7f85f01e4e9a]
 9: (clone()+0x6d) [0x7f85ef0884bd]

the problem is that we get a paxos accept on another paxos machine that raced with the monmap contraction, but did not ignore it because win_standalone_election() didn't bump the election epoch.