I've not explored why it would have gotten 2 different IP addresses, but it could also be the fact that during commissioning, we try to bring up other interfaces to see if we can discover networks. I wonder if this has regressed and it is trying to dhcp scan the interface used for PXE, which would cause it from potentially obtaining a new IP?
Or rather, there was a bug were cloud-initramfs-tools (i think it was there) that would copy the network config from the initrd to the ephemeral environment, causing the machine not to re-dhcp and that caused a few regressions, such as not renewing the IP lease because network configuration was "statically" configured?
I've not explored why it would have gotten 2 different IP addresses, but it could also be the fact that during commissioning, we try to bring up other interfaces to see if we can discover networks. I wonder if this has regressed and it is trying to dhcp scan the interface used for PXE, which would cause it from potentially obtaining a new IP?
Or rather, there was a bug were cloud-initramfs -tools (i think it was there) that would copy the network config from the initrd to the ephemeral environment, causing the machine not to re-dhcp and that caused a few regressions, such as not renewing the IP lease because network configuration was "statically" configured?