CNO autodetects the MTU (pkg/network/mtu.go) but it only looks at IPv4 so it fails reliably on single-stack IPv6: 2020/08/31 23:42:13 Failed MTU probe, failling back to 1500: got no routes Also, "falling" is misspelled.
Can we assume "or force" that, in dual stack, IPv4 and IPv6 will have ALWAYS the same MTU (and that the default route will use the same interface)? Managing different MTUs per family seems a bad idea and a is very BAD networking practice. We can take the minimum common MTU in this case
This bug is pretty much just a matter changing netlink.FAMILY_V4 to netlink.FAMILY_ALL in GetDefaultMTU. I'm not sure why I filed a bug rather than filing the PR actually.
https://github.com/openshift/cluster-network-operator/pull/777
@arik, could you help us your team verify this since you have single/dual stack clusters being deployed? Please re-assign accordingly. Thanks
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory (OpenShift Container Platform 4.6 GA Images), and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHBA-2020:4196