I have a test 1.29 k8s cluster with 1.29 cri-o runtime. Because this is a test cluster, there is no any workload on it.
I will upgrade the k8s cluster to 1.30, then I need to upgrade cri-o to 1.30 first.
Before I run the cri-o upgrade procedure, I have stopped cri-o service multiple times, the pods still running and not re-created. which is the expected behavior. Then https://github.com/cri-o/cri-o/issues/2479 should be not my case.
However, I notice that the existing pods (flannel and kube-proxy) will be recreated after the cri-o upgrade to new version and restart cri-o service.
I'm not sure the pod recreation is the expected behavior when cri-o upgrade, or I have some not correct configuration to make the pod recreated.
In addition to upgrade cri-o from 1.29 to 1.30, I also have tried downgrade cri-o from 1.30 to 1.29. Once the major version of cri-o has changed, the pod will be re-created.
Following is the commands I used to upgrade the cri-o.
What happened?
I have a test 1.29 k8s cluster with 1.29 cri-o runtime. Because this is a test cluster, there is no any workload on it. I will upgrade the k8s cluster to 1.30, then I need to upgrade cri-o to 1.30 first.
Before I run the cri-o upgrade procedure, I have stopped cri-o service multiple times, the pods still running and not re-created. which is the expected behavior. Then https://github.com/cri-o/cri-o/issues/2479 should be not my case.
However, I notice that the existing pods (flannel and kube-proxy) will be recreated after the cri-o upgrade to new version and restart cri-o service.
I'm not sure the pod recreation is the expected behavior when cri-o upgrade, or I have some not correct configuration to make the pod recreated.
In addition to upgrade cri-o from 1.29 to 1.30, I also have tried downgrade cri-o from 1.30 to 1.29. Once the major version of cri-o has changed, the pod will be re-created.
Following is the commands I used to upgrade the cri-o.
Step for upgrade cri-o
What did you expect to happen?
When I upgrade crio, the existing pod should not re-created
How can we reproduce it (as minimally and precisely as possible)?
Anything else we need to know?
No response
CRI-O and Kubernetes version
OS version
Additional environment details (AWS, VirtualBox, physical, etc.)