ZOS freezes with dual IPv6 connections

#1

On the TLRE farm we’ve been able to get IPv6 working on the 2 10Gb interfaces (that were meant to be bonded in the farm V1 setup). After rebooting all the nodes in the TLRE farm, slightly less than half of the nodes went offline and needed a reboot to get online again, but for 2 nodes that go online during boot, only to freeze and timeout/go offline.

These 2 nodes are:

  • CrRm1pzXsqhe3ExLUyC2aS9gSsr51ojdcdv2RQ7vhMMQ
  • HyZBxd6kJfpSv96Qv73oBJwWvf51pDiMXqhChmaffPyg

This is a screenshot of a frozen node

image
image800×599 84.7 KB

This is a screenshot of a working node

image
image800×580 105 KB

Is there someone who can take a deeper look?
I’ve turned these nodes off to conserve power, in case some one wants me to boot them to see what happens, best to contact me on telegram @chrisvdg

#2

We’re also working on getting IPv6 on the Bancadati farm, I will also post on this thread if something odd happens then.

#3

I’ve checked the logs of those 2 node at the date you created this post. Here is what I found: https://gist.github.com/zaibon/f2bf578fd7c7db79d51e75986240bdc1

I’m not sure yet if this is just a configuration issue or if this shows a bug in networkd boot logic, will open an issue to track it.

#4

So it turns out the router of those farm seems to not configure the default route for the IPv6 network interface.