Fixes issue when worker not running, burrow not getting new state

MTime in zookeeper will only get updating when being committed or new message pushed, however, burrow has a short-interval check for current and previous timestamp, this will prevent the next checking interval to complete if partition MTime not changed. This fix is basically to tell burrow checks anyway if two adjacent timestamp is the same.

offsets_store.go

@@ -274,7 +274,7 @@ func (storage *OffsetStorage) addConsumerOffset(offset *PartitionOffset) {
 	} else {
 		// Prevent old offset commits, and new commits that are too fast (less than the min-distance config)
 		previousTimestamp := storage.offsets[offset.Cluster].consumer[offset.Group][offset.Topic][offset.Partition].Prev().Value.(*ConsumerOffset).Timestamp
-		if offset.Timestamp-previousTimestamp < (storage.app.Config.Lagcheck.MinDistance * 1000) {
+		if offset.Timestamp-previousTimestamp < (storage.app.Config.Lagcheck.MinDistance * 1000) && offset.Timestamp-previousTimestamp != 0{
 			storage.offsets[offset.Cluster].consumerLock.Unlock()
 			return
 		}