This message was deleted.
# k3s
a
This message was deleted.
c
found that it’s attempting to check cluster domains first, resulting in timeouts.
Yes, that is how DNS search order works. It tries the cluster entries first, walking up the various prefixes, and only if it doesn’t find any there does it continue to external DNS
Why are the lookups for cluster DNS addresses timing out?
is coredns working properly? You said it does this even if there is a coredns replica on the agent node?
Are you sure that you’re not seeing inter-node traffic dropped? If you’re sure that the CNI ports are open, you might try disabling checksum offload on the flannel adapter, that is a common problem with buggy ethernet drivers on VMs.