networkop / meshnet-cni

a (K8s) CNI plugin to create arbitrary virtual network topologies
BSD 3-Clause "New" or "Revised" License
116 stars 28 forks source link

Meshnet + init wait container getting stuck after pod deletion + recreation #48

Open alexmasi opened 2 years ago

alexmasi commented 2 years ago

We have a KNE (https://github.com/openconfig/kne) user seeing errors with init-wait container hanging waiting for interfaces after pod deletion + recreation using the command kubectl get pod -n <namespace> <container> -o yaml | kubectl replace --force -f -.

@robshakir to add more details on specific log messages

networkop commented 2 years ago

Hey @alexmasi , do you mean this init-wait container https://github.com/networkop/k8s-topo/blob/master/init-wait/entrypoint.sh ?

alexmasi commented 2 years ago

yes

Cerebus commented 1 year ago

Can this be closed as resolved by #74 ?

alexmasi commented 1 year ago

Probably, I haven't been testing scenarios with repeated pod deletion and recreation. I'm assuming the linked change + https://github.com/networkop/meshnet-cni/pull/73 + the upcoming grpc wire reconciliation work should fix the issues. Happy to close and re-open if I see the issue again.