https://rancher.com/ logo
Join Slack
Powered by
This message was deleted.
# k3s
a
This message was deleted.
v
I guess I can set the name, but then it tells me that the first host is unable to join the cluster.... that k3s is initializing? 😕
Example:
Copy code
Oct 23 12:56:20 dal3-qz2-sr3-rk248-s47 k3s[795102]: time="2024-10-23T12:56:20Z" level=info msg="Failed to test data store connection: this server is a not a member of the etcd cluster. Found [mzoned3c=<https://10.22.67.65:2380>], expect: dal3-qz2-sr3-rk248-s47-42c12652=<https://203.0.113.254:2380>"
config.yaml:
Copy code
etcd-arg:
  - 'name=mzoned3c'
  - 'initial-cluster=mzoned3c=<https://10.22.67.65:2380>'
  - 'listen-peer-urls=<https://0.0.0.0:2380>'
  - 'initial-advertise-peer-urls=<https://10.22.67.65:2380>'
  - 'advertise-client-urls=<https://10.22.67.65:2380>'
node-external-ip: 10.22.67.65
advertise-address: 10.22.67.65
data-dir: /home/k3s
bind-address: 0.0.0.0
cluster-init: true
selinux: true
node-name: "dal3-qz2-sr3-rk248-s47"
disable-default-registry-endpoint: true
c
Don’t do that. Why are you trying to muck about with the peer URLs? Those not intended to be overridden; the etcd cluster membership is managed by k3s itself and it will get very confused if you try to change IDs or addresses.
Why are you overriding so many things to that IP? I think you just want to set --node-ip=10.22.67.65 --node-external-ip=203.0.113.254
I don’t understand why you’re using a dummy interface in a HA cluster either, normally that’s only necessary if you’ve got a single airgapped node that has no network connection and need a dummy interface for Kubernetes to work properly.
v
Yep came to the realization that I only needed to override these 2:
Copy code
- 'listen-peer-urls=<https://0.0.0.0:2380>'
  - 'listen-client-urls=<https://0.0.0.0:2379>'
In order to get to listen on something other than the 203 address. Will try with just 
--node-ip
and 
--node-external-ip
Complete noob here man - just trying to figure stuff out. Not a kube guy really.
c
in general I think you’re doing too much. See if it works out of the box, if not then configure just what you need. Unless you have some really weird network setup or are trying to do something that just isn’t going to work at all, none of these things should need to be poked at.
Just start the first server with --cluster-init then join 2 more servers. Thats all you need to do to get 3-node cluster with etcd.
t
To be fair, we were trying a few different things and we have a very funky network. Thanks for helping out exactly what we needed something crazy simple.
11 Views
Open in Slack
PreviousNext
Powered by