Open chenchun opened 4 years ago
I think it's better to call cni cmd_check for pods and if the result is unexpected, we should cmd_delete and recall cmd_add for them. But first we should upgrade to cni spec-v0.4.0. Depend on https://github.com/tkestack/galaxy/issues/48
Network manager and restarting network can delete veth and container routes. We should resync of veth and container routes in a loop?