Johan Andrén

That sounds about right, can happen in a highly loaded system where subscribing is delayed for some reason. Saturated CPU or starved thread pool because of blocking could explain it,...

I think you are scratching at the core of the problem here, there is no reasonable unit type that you have an instance of in Java. I think we should...

And after: https://github.com/akka/akka/actions/runs/5873846885/job/15927685999 I wonder if this is related, in the logs: ``` error: Get "https://34.135.244.241/api/v1/namespaces/default/pods?labelSelector=host%3Dtest-node1&limit=500": dial tcp 34.135.244.241:443: connect: connection refused - error from a previous attempt: read tcp...

Also in the logs: ``` Run gcloud container clusters delete "akka-multi-node-$***GITHUB_RUN_ID***" --quiet ERROR: (gcloud.container.clusters.delete) Some requests did not succeed: - ResponseError: code=400, message=Operation operation-1692155995629-76c0bd5d-5885-4271-8442-17d9eea82c51 is currently repairing cluster akka-multi-node-5873846885. Please...

One more thing, also k8 related: ``` [error] Error from server: error dialing backend: No agent available ```

https://github.com/akka/akka/actions/runs/9057019166/job/24880380958#step:10:20741

https://github.com/akka/akka/actions/runs/5861971593/job/15893013023#step:5:17943

https://github.com/akka/akka/actions/runs/9931758360/job/27432083762#step:5:18164

https://github.com/akka/akka/actions/runs/9883249590/job/27297645642#step:5:18168

I think this problem caused by multiple tests cases spawn streams but not waiting for the stream to complete before moving on, so a previous test case stream is seen...