Runtimecfg broken for ovirt about baremetal-runtimecfg HOT 6 CLOSED

@Gal-Zaidman restarting the pod does not re-generate the mdns configuration, could you delete the configuration and use crictl to delete the pod? If you do, kubelet should create the pod again and that will trigger the initcontainer again.

from baremetal-runtimecfg.

This is happening even on baremetal IPI, all installs started failing for us yesterday.

In mdns-publisher, I see this:

[core@master-0 ~]$ sudo crictl logs 3ec9fcdd466c2
time="2020-01-16T14:54:21Z" level=info msg="Publishing with settings" collision_avoidance=hostname ip=192.168.111.0
time="2020-01-16T14:54:21Z" level=fatal msg="Failed to find interface for specified ip" ip=192.168.111.0

Happy to share access to the env if you'd like

from baremetal-runtimecfg.

@Gal-Zaidman restarting the pod does not re-generate the mdns configuration, could you delete the configuration and use crictl to delete the pod? If you do, kubelet should create the pod again and that will trigger the initcontainer again.

That is what we did, we stopped kubelet, removed the mdns pod, removed the file and then started the kubelet. We made sure the init container runs again

from baremetal-runtimecfg.

This is because net.ParseCIDR zeroes out the part of the address past the netmask. See https://play.golang.org/p/ZZcvN-JpS6A

I think we need to reconstruct the IPNet by keeping the first two return values from ParseCIDR and then packing them back together.

ip, n, _ := net.ParseCIDR("1.1.1.1/16")
n = &net.IPNet{IP: ip, Mask: n.Mask}

from baremetal-runtimecfg.

was this fixed by #42 ?

from baremetal-runtimecfg.

Technically it was fixed by #41, but yeah it should be working now.

from baremetal-runtimecfg.