-
Notifications
You must be signed in to change notification settings - Fork 39
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
debian instance not accessible when multiple network interfaces present #4836
Comments
Stopping the instance, removing the non-primary interfaces and starting again lead a different state, but still no connectivity. Console output shows this from cloud init
I'm not able to log into this instance without network connectivity or I would do so and dig in more. |
I've been trying out how different guest OS and versions work with multiple nics and they all seem to behave differently when it comes to the default route settings. The observations so far are summarized as follows:
The question here is - should we leave the net assignment to user (i.e. for them to configure cloud-init to achieve what they want)? The answer seems to be "no" because we explicitly provide a way for user to designate the primary nic. @pfmooney, thoughts? |
I should also mention that I've inspected the opte entries on the sled in some of the cases when I couldn't SSH in because the default route was on the non-primary nic. As I had set the password of the default user, I accessed the VM via serial console and removed the incorrect default route from there, e.g.,
Once the bad route was taken out, I was able to SSH into the VM on the external IP (which is NAT-ed to the 172.30.0.x primary nic). All that is to say the opte side of things is working as expected so what remains unclear is the plumbing into the guest OS. I tried digging into the propolis zone settings and all I could tell was that the nics were passed but there was nothing that marked an interface as primary:
|
I ran into this when deleting the network interface on an existing VM and creating a new one in a different VPC. The console showed that an IP address had been assigned to the host, but the link was down.
After enabling the link and rebinding DHCP, the connection on the new VPC was established:
|
Some of the OPTE state does appear to be a little messed up here. I set up an instance with two NICs on dogfood, then swapped them and set the newly created one as primary. The original NIC is in the subnet
Note that
It's being assigned to
|
The OPTE-related symptoms should be fixed by #6890 -- from poking at it further the issue would only crop up on changes to EIP configuration (i.e., new FIP/Ephemeral added dynamically, or associated IGW tags for existing addresses). |
I created an instance with a Debian 12 cloud-init image with 3 network interfaces. When it came up, I was not able to access it via its external IP.
This is what it looks like from the CLI.
From the instance console output, it looks like the default route is being assigned the wrong gateway.
It also looks like some address assignments are missing.
The text was updated successfully, but these errors were encountered: