all-things-risingwave
How to resolve pod health issues when deploying RisingWave on GKE Autopilot?
I'm facing pod health issues when deploying RisingWave on GKE Autopilot due to constraints like no privileges and kube-dns. The issue seems related to pod liveness and readiness probes failing with connection refused errors. How can I resolve these pod health issues on GKE Autopilot?
In
Indy
Asked on May 17, 2023
- The pod health issues on GKE Autopilot when deploying RisingWave are related to kube-dns not creating valid DNS records for endpoints of Pods controlled by StatefulSet.
- One workaround is to replace kube-dns with CoreDNS, but this is not allowed in the GKE Autopilot context due to system components protection.
- The RisingWave operator has fixed this issue and will release a new version at the end of the month.
- To resolve the pod health issues, consider using GKE Standard with CoreDNS or wait for the updated RisingWave operator version.
May 25, 2023Edited by