Techniques are disclosed relating to synchronizing state information across nodes in a cluster. In one embodiment, a first cluster node receives a request to update state information for a logical volume of the cluster. The first cluster node acquires a lock from a distributed lock manager of the cluster, the lock controlling modification of the state information. In response to acquiring the lock, the first cluster node updates the state information across cluster nodes of the cluster. In one embodiment, first cluster node updates the state information in some, but not all, of the cluster nodes. In some embodiments, the first cluster node updates the state information across only cluster nodes associated with the logical volume.