Description of problem: Pods are stuck in a container-creating state due to kuryr-cni not being able to assign IP to the Pod. Version-Release number of selected component (if applicable): How reproducible: Happens quite frequently on a specific namespace where more than 300 pods are running. Steps to Reproduce: 1. Install OpenShift 4.8 on OpenStack with Kuryr. 2. Create a project and deploy 300+ pods within the namespace. 3. Try to rollout deployment/restart the app pods. Actual results: kuryr-cni pod is restarting and some pods within the namespace are stuck in ContainerCreating. Expected results: Pods should run just fine. Additional info: Crio Logs: https://drive.google.com/file/d/1olEb5B2N6MO-9syIaFnrMuaktHYfrA61/view?usp=sharing
I'm marking this as a duplicate as there's a fix in 4.8.26 that greatly improves such situations. Please reopen if this proves to include different issues. *** This bug has been marked as a duplicate of bug 2028307 ***