👋 Has anyone who runs Flink on Kubernetes seen occasional seemingly synchronized waves of restarts of different Flink Clusters running on the same K8s cluster? I'm investigating an issue and it seems like the root case may be leader elections being triggered due to ConfigMaps used for HA being unavailable. This causes the job to restart. Wondering if anyone has seen something similar, and if they had any advice 🙏

Haven’t seen this on AWS, could be GCP specific?

Thanks for the response! Seems possible:

The default rate limit for the Kubernetes Engine API is 3000 requests per min, enforced at intervals of every 100 seconds.

(source) Planning to investigate further on that