Roger Johansson
Roger Johansson
Currently, we have OpenTracing support: https://github.com/asynkron/protoactor-go/tree/dev/actor/middleware/opentracing This should be updated to support OpenTelemetry.
Make it possible to give some messages priority for the Remote layer. E.g. Gossip messages. This can probably be solved with a priority queue, or an approach similar to system...
It would be interesting to evaluate how a dedicated thread dispatcher would affect e.g. Remote layer under heavy load. if that makes any difference or if the system still behaves...
WiP - V2
This has not been verified, but there might be cases where the member list incorrectly handles joined or leaving members. Specifically for the SeedNode cluster Provider. As there is no...
Currently, we don't have any shared flag if the system is shutting down. the cluster has its own flag. which is not reachable by e.g. remote. So while the cluster...
Gossip Actor can get into bad state. I've seen the gossip loop go into a failed state where it never recovers. when this happens, other nodes cannot talk to this...
Some mind notes on how large clusters behave with various settings. ## HeartbeatExpiration HeartbeatExpiration works poorly with large clusters. if one of the many nodes happens not to reach some...
when running the "AspNetGrains" example. it's common to see that one of the nodes complains about "Consensus not reached, Initiating rebalance" when the apps start up. This is odd, as...
In this screenshot we see how a cluster member leaves the cluster. proto remote correctly closes the endpoint. but soon after, tries to reestablish it (which it shouldn't)