Handle graceful termination of pod in Kubernetes in multi-container scenario

Question

There is a situation where there are 2 containers within the pod and one takes requests from 2nd while it is in termination.

Let's say Container 1 depends on Container 2 and both the containers receive SIGTERM simultaneously and Container 2 stops quickly and all the inflight requests in Container 1 being dependent on Container 2 starts failing.

Any solution for this?

Answer 1

In this case, the containers are better fitted to running on two different pods (IMHO)

And adding a retry mechanism using code/servicemesh to make the application more resilient

For example https://istio.io/latest/docs/concepts/traffic-management/#retries