Documenting here in case any other users run into the issue. Working on a fix..
When redeploying a topology with saved configs XRd can have multiple IPv6 addresses on it's mgmt interface.
Saved configs as in: committed config within XR CLI.
On redeploy the IPv6 address from the old deployment are present in the saved configs, they are not removed.
The new management IPv6 address is simply added, unlike the IPv4 management address which overrides the old one in the config.
Have to find out the root cause, might be that the startup config loads the new mgmt ipv6 and the existing config has the old ipv6 and the startup config gets merged into the saved config.
This means that XRd nodes have both their old and new management IPv6 address, meaning possible overlaps which could mean:
A different mapping between hostname and node itself (the wrong IPv6 address is replying, SSH prefers IPv6).
This could render some nodes unreachable since you can get different replies from duplicate addresses.
Documenting here in case any other users run into the issue. Working on a fix..