Fix possible problems with ClusterClient Discovery #7270
Conversation
| @@ -49,6 +50,7 @@ private sealed record ResolveResult(Contact Contact, IActorRef? Subject); | |||
| private readonly string _targetActorSystemName; | |||
| private readonly string _receptionistName; | |||
| private readonly string _transportProtocol; | |||
| private readonly int _numberOfContacts; | |||
There was a problem hiding this comment.
This is just the number of initial contacts we need, right?
There was a problem hiding this comment.
it limits the number of contacts we use when we instantiate the ClusterClient actor. Too many of these will set the initial contacts to a higher than optimum number. The thing is that ClusterClient initial contacts never gets trimmed when they can't be contacted, if we use the whole discovery resolve result as the ClusterClient initial contact list, there will be a big burst of network connection attempt when the ClusterClient actor starts up and whenever its internal contact list could not be contacted anymore.
| /// </summary> | ||
| /// <param name="array"></param> | ||
| /// <typeparam name="T"></typeparam> | ||
| private static void Shuffle<T>(T[] array) |
There was a problem hiding this comment.
Do we have another implementation of this lying around somewhere? I could have sworn I'd written one before
There was a problem hiding this comment.
There is one in the Akka.Util.Internal.ArrayExtensions, but its internal
There was a problem hiding this comment.
I would add a link to that file and just use that (or you can add a friend assembly to Akka.Discovery; link is probably cleaner.)
There was a problem hiding this comment.
That's just a nitpick though
There was a problem hiding this comment.
Done, turns out the core Akka project is already a friend of Akka.Cluster.Tools
Changes