diff --git a/consensus/raft/raft.go b/consensus/raft/raft.go
index d34c9f47..79db55c9 100644
--- a/consensus/raft/raft.go
+++ b/consensus/raft/raft.go
@@ -38,11 +38,6 @@ var RaftMaxSnapshots = 5
 // This is used to reduce disk I/O for the recently committed entries.
 var RaftLogCacheSize = 512
 
-// Are we compiled on a 64-bit architecture?
-// https://groups.google.com/forum/#!topic/golang-nuts/vAckmhUMAdQ
-// This is used below because raft Observers panic on 32-bit.
-const sixtyfour = uint64(^uint(0)) == ^uint64(0)
-
 // How long we wait for updates during shutdown before snapshotting
 var waitForUpdatesShutdownTimeout = 5 * time.Second
 var waitForUpdatesInterval = 100 * time.Millisecond
@@ -260,25 +255,9 @@ func (rw *raftWrapper) WaitForLeader(ctx context.Context) (string, error) {
 	ctx, span := trace.StartSpan(ctx, "consensus/raft/WaitForLeader")
 	defer span.End()
 
-	obsCh := make(chan hraft.Observation, 1)
-	if sixtyfour { // 32-bit systems don't support observers
-		observer := hraft.NewObserver(obsCh, false, nil)
-		rw.raft.RegisterObserver(observer)
-		defer rw.raft.DeregisterObserver(observer)
-	}
 	ticker := time.NewTicker(time.Second / 2)
 	for {
 		select {
-		case obs := <-obsCh:
-			_ = obs
-			// See https://github.com/hashicorp/raft/issues/254
-			// switch obs.Data.(type) {
-			// case hraft.LeaderObservation:
-			// 	lObs := obs.Data.(hraft.LeaderObservation)
-			// 	logger.Infof("Raft Leader elected: %s",
-			// 		lObs.Leader)
-			// 	return string(lObs.Leader), nil
-			// }
 		case <-ticker.C:
 			if l := rw.raft.Leader(); l != "" {
 				logger.Debug("waitForleaderTimer")