Get Remote Actor fail

[hemaolong]

When I run the distribute example, the worker sometimes can not get the remote actor.
It seems that the registry return Ok but with Nothing(None):

received actor node (current_node=example-worker, actor_id=echo-actxor, actor_node=None)

        match self
            .inner
            .registry_ref
            .send(GetActorNode {
                actor_id: actor_id.clone(),
                sender: tx,
            })
            .await
        {
            Ok(_) => {
                // TODO: configurable timeouts (?)
                match tokio::time::timeout(tokio::time::Duration::from_secs(5), rx).await {
                    Ok(Ok(res)) => {
                        trace!(
                            "received actor node (current_node={}, actor_id={}, actor_node={:?})",
                            self.node_tag(),
                            &actor_id,
                            &res
                        ); 
                        res

[hemaolong]

I think this is a time-sequence problem. The "this.nodes" is not all ready will get the wrong node id.

#[async_trait]
impl Handler<GetActorNode> for RemoteRegistry {
    async fn handle(&mut self, message: GetActorNode, _: &mut ActorContext) {
        // let span = tracing::trace_span!(
        //     "RemoteRegistry::GetActorNode",
        //     actor_id = message.actor_id.as_str()
        // );
        //
        // let _enter = span.enter();

        let id = message.actor_id;
        let current_system = self.system.as_ref().unwrap().node_id();
        // Not all nodes registered.
        let assigned_registry_node = self.nodes.get_by_key(&id).map(|n| n.id);

        let assigned_registry_node = assigned_registry_node.map_or_else(
            || {
                trace!("no nodes configured, assigning locally");
                current_system
            },
            |n| n,
        );

        trace!("{:?}", &self.nodes.get_all());