If you are facing a problem like
- Adding terminationGracePeriodSecond to your Kubernetes Deployment to handle graceful shutdown, but your Kubernetes Pod cannot connect Database, Redis, Queue to do some clean up
- Your Kubernetes Pod is terminated immediately (before terminationGracePeriodSecond period)
The root cause maybe Istio sidecar is terminated before the Kubernetes Pod
Solution
For example, you have terminationGracePeriodSeconds: 315
You need to add terminationDrainDuration to your Kubernetes Deployment also, e.g.,
annotations:
proxy.istio.io/config: |
terminationDrainDuration: 315s
Reference
- How can I prevent Istio sidecar from shutting down before my service has finished gracefully terminating? Link
ไม่มีความคิดเห็น:
แสดงความคิดเห็น