Leader election is a key service for many dependable distributed systems.
Several consensus protocols, e.g., Paxos, rely on a leader election service.
We present a leader election algorithm for crash-recovery and omission environments.
We assume that eventually a majority of processes are up and do not omit messages.