https://rancher.com/ logo
Join Slack
Powered by
This message was deleted.
# rancher-desktop
a
This message was deleted.
f
RD uses the alpine-lima ISO corresponding to the host arch (aarch64 on M1) on macOS and Linux.
i
Hmm...K...Is it possible to change the image that is used?
f
Probably not; why would you want to?
i
I think there may be something going on with k3s that might be causing me headache's (TCP Timeouts, Handshake Failures, Etc). I originally thought it was related to the Cisco VPN on my work laptop, so I setup a POC using K3os on another external vm. When I connected, I suddenly started getting the same errors far more frequently (as in every couple of seconds if I simply did 
watch kubectl get pod
). After that failed, I setup a new VM using ubuntu server using the 'lightweight' install and went with their self installed version of microk8s. When I connected to this instead, the errors have gone away on both my personal and work laptops. The only issue I'm having now is that running the afore mentioned command is extremely slow when I'm connected to the vpn on my work laptop. Once I disconnect, it seems to be running fine. I was thinking if possible, I'd like to try using this same style 'setup' via rancher-desktop to see if this takes care of the rest of my issues.
Keep in mind this is only speculation at this time, so I'm kind of just throwing things at the wall to see what sticks.
f
When you say "external VM" and "new VM", are these VMs running on the same machine, and are they managed by lima/qemu, or something else?
I think any networking issues are due to qemu and not the specific distribution running on top of it.
If you have any particular networking issues on Rancher Desktop, and they are not happening when you are running on e.g. an Ubuntu VM, but also on Lima, then please file a bug report with the details!
Replacing alpine-lima would be a lot of work; it is custom-built for Rancher Desktop, and also different from other Lima configurations in that it runs from a read-only ISO image, with an attached data volume.
i
I mean that I am pointing to a completely different computer on my network that has a VM running on it.
That is to say I'm completely bypassing RD. I got the same errors with my first VM which was using k3os instead of ubuntu server (also on a different computer on my network). This is what led me to believe it might be related to k3s somehow. I got to the point that I ssh'd into the vm for k3os and ran 
watch kubectl get pod
directly on the machine and continued to have the every couple seconds I would get the TCP Timeouts, even though I was directly on the machine at that point (so shouldn't of been any TCP connections across the network). Using ubuntu w/microk8s is working like a charm so long as I'm not trying to do host mounts and the vpn is disabled.
f
Oh, that's a lot of additional changes, so hard to pin down the different behaviour to a particular change
i
Yes, I agree. That's why I was hoping to try and use RD with lima, but use a different base image to see if the problem was replicated if I simply used a different OS for the base image.
(and or service - that is to use microk8s instead of k3s)
f
We have different templates for this in Lima, so you could experiment with lima directly instead of using RD.
i
Do you have any documentation on how I might go about doing this?
f
Not really, but start with the examples in https://github.com/lima-vm/lima/tree/master/examples
i
Will Do, Thank You
f
You can see both a 
k3s
and a 
k8s
template, the later using 
kubeadm
When you have Lima installed, 
limactl start --list-templates
shows you the installed templates, and you can create a VM with e.g. 
limactl start <template://k8s>
72 Views
Open in Slack
PreviousNext
Powered by