Running an Akka cluster on k8s and it is using a downing strategy (let’s say Autodowning), so in the case where a node goes unreachable the container which went unreachable exits. The problem is that this node went unreachable because of a network issue/ issue with the platform provided by k8s and as such the entire pod should be restarted and scheduled onto a new healthy k8s node. Because scheduling can take some time we only want to reschedule the container onto a new pod on a new node if there are unreachability is the cause of the failure. Is there any way to propagate failure messages to the parent in k8s like use an exit code to make the decision of when to restart the container and when to delete the pod.
Related topics
Topic | Replies | Views | Activity | |
---|---|---|---|---|
Self-termination for application running in Kubernetes | 1 | 830 | October 17, 2018 | |
Downing an unreachable node via akka management doesn't solve the unreachability | 3 | 46 | January 21, 2025 | |
Cluster gets down automatically with "Shutting down myself" message | 3 | 2099 | August 27, 2018 | |
Akka cluster in kubernetes gets into inconsistent state | 6 | 580 | February 14, 2023 | |
Downed member is not removed from cluster and becomes reachable again | 1 | 1123 | July 13, 2018 |