This message was deleted.

I can't see anything in the documentation about this. If I can force it down VLAN 4000, then I can reconfigure UFW to prevent internet access to any important ports

You can set

node-ip

in the config

oh great thanks. I missed that one. Will that all configure the etcd instance to use the same IP?

Yep it should! You can also set

advertise-address

to be double certain for those too

Great. thanks Max

Sorry if I overexplain… 1.

curl -sfL <https://get.k3s.io>

--> retrieves the contents from that URL 2.

| …

pipes the output, meaning it takes the output from the curl command and uses it as input for the next stuff. All the next stuff together is starting k3s: 3.

K3S_TOKEN=MY_KEY

is setting the environment variable, which the install script reads 4.

sh -s -

Running the install script, which will run a series of shell functions 5.

server --cluster-init --docker

Arguments to pass to the

k3s

process that the install script is aware of what to do with them. You’ll see these actually get added to the systemd service file at

/etc/systemd/system/k3s.service

SO the best way, if you want to restart, would probably be to uninstall and reinstall using the same process: 1.

sudo k3s-uninstall.sh

2.

curl -sfL <https://get.k3s.io> | K3S_TOKEN=MY_KEY sh -s - server --cluster-init --docker --node-ip=192.168.100.1

You can also just rerun this directly without uninstalling, but results may be unexpected because kubernetes doesn’t behave well when a node ip is changed

Thanks. Always better to over explain 🙂 I don't want to re-install, I have setup quite a bit of config and quite a lot of containers. Based on the above, should I just run the following then? K3S_TOKEN=MY_TOKEN k3s server --docker -i 192.168.100.1

Take a look at https://github.com/k3s-io/k3s/discussions/7326 and https://github.com/k3s-io/k3s/issues/6524#issuecomment-1321130017.. I think the second one will probably be the most helpful in this case. Changing node ips is weird, so you might have to jump through a few hoops in this case

ok, will do. thanks. I've already stopped the services, updated the firewalls, and now adding the -i param, so I am hoping it works. But I'll double check those

Yep and I would add, take a snapshot first in case you wanna say screw it and just keep your current stuff!

ok. good idea. i'll do that. so i presume the killall process doesn't delete that?

Correct! It just stops the running service and pods and stuff

and is that only required on the primary

killall is required on all, take the snapshot on each node as it will save it locally (unless you want to specify an s3 or minio storage endpoint): https://docs.k3s.io/cli/etcd-snapshot

ok got it

Feel free to open an issue at https://github.com/k3s-io/docs or on the main repo

I'll do that tomorrow. Thanks

It’s just

k3s.service

.

k3s-agent.service

is for agent (or worker) -only nodes.

no, there is no file called that because it was jus setup on the CLI. Yes, just found it in the service. Will try editing the IP in there:

You can create the config.yaml also and just put it there! It makes managing configuration much easier when it’s in its own file, and we’ve been recommending it

ah ok.. so if i put a new one there now, it will add whatever settings in

upon restart yep!

i'll install the etcd cli and take a look.. service is upset. I'll learn more that way

yeah this is getting you into the weeds of configuring, updating, and restoring a kubernetes cluster, which isn’t really ever easy, even though k3s tries its best to simplify it as much as possible 😓

service won't start now.. so i'll have to. unfortunately made the UFW changes prior to taking a snapshot, so I can't get in to take one. from systemctl status it is upset it can't get to the other nodes so throws an error 500

ah yeah ufw changes would have to happen as the last step

you live and learn