1 basic question about ingress. I was able to setup ingress and working fine . My question is, when I specify the DNS hostname , the request hits the ingress and get the page. But if I specify the IP address instead , it hits the Loadbalancer and will give out the 404 error message, as expected. Wondering how this routing works !

Ia there a traefik dashboard in K3s? Not the Kubernetes dashboard.

hmm, using the k3s-airgap images from github works. but if i put additional images into the same folder, they appear in the registry but k3s is still trying to pull, even though the image is already available. anyone had that problem?

Hi, Is there a (simple) way to configure the local-path-provisionner shipped with k3s? I wish to declare multiple path for the

DEFAULT_PATH_FOR_NON_LISTED_NODES

. I noticed the local-path-provisionner use a

kube-system/local-path-config

ConfigMap. But as this ConfigMap provide

config.json

+ many other files, it is not simple to override the part I'm interested in. I suppose having two ConfigMap, one for

config.json

and one for other files can help. But there is possibly an other approach for that.

Good day all, tried installing k3s on my mac M1 and I have been getting this error "shell failed: instance "k3s-worker" does not exist" whenever I run this command multipass shell k3s-worker.

Hi folks, Perhaps someone will have an idea of what should be checked and what could cause that issue, let me describe. The K3S cluster version is

v1.26.2+k3s1 (ea094d1d)

. When the pod is running on the server nodes everything works fine, but when it gets created on the agent nodes after some time it stops working, logs from the controller that is running in that pod on the agent node:

NAMESPACE                       NAME                                         READY   STATUS             RESTARTS         AGE     IP           NODE                   NOMINATED NODE
  READINESS GATES
actions-runner-system           actions-runner-controller-798b5fdf46-7p9bn   1/2     CrashLoopBackOff   16 (3m12s ago)   84m     10.42.7.9    <http://yc3142.host.com|yc3142.host.com>   <none>

Defaulted container "manager" out of: manager, kube-rbac-proxy
2023-04-28T06:31:29Z	ERROR	Failed to get API Group-Resources	{"error": "Get \"<https://10.43.0.1:443/api?timeout=32s>\": dial tcp 10.43.0.1:443: i/o timeout"}
<http://sigs.k8s.io/controller-runtime/pkg/cluster.New|sigs.k8s.io/controller-runtime/pkg/cluster.New>
	<http://sigs.k8s.io/controller-runtime@v0.14.4/pkg/cluster/cluster.go:161|sigs.k8s.io/controller-runtime@v0.14.4/pkg/cluster/cluster.go:161>
<http://sigs.k8s.io/controller-runtime/pkg/manager.New|sigs.k8s.io/controller-runtime/pkg/manager.New>
	<http://sigs.k8s.io/controller-runtime@v0.14.4/pkg/manager/manager.go:359|sigs.k8s.io/controller-runtime@v0.14.4/pkg/manager/manager.go:359>
main.main
	./main.go:173
runtime.main
	runtime/proc.go:250
2023-04-28T06:31:29Z	ERROR	unable to start manager	{"error": "Get \"<https://10.43.0.1:443/api?timeout=32s>\": dial tcp 10.43.0.1:443: i/o timeout"}
main.main
	./main.go:188
runtime.main
	runtime/proc.go:250

Was trying to check connectivity from the node

<http://yc3142.host.com|yc3142.host.com>

directly:

nc -v 10.43.0.1 443
Ncat: Version 7.70 ( <https://nmap.org/ncat> )
Ncat: Connected to 10.43.0.1:443.

So, the node can access 10.43.0.1 on port 443, but can't get why the app stops working after some time on the agent nodes (again, on the master nodes works fine) Firewalld is disabled. In the internal firewall, I have allowed all traffic. Thanks in advance,

@refined-analyst-8898 has left the channel

Hello Team, I have created a K3s Cluster in which master nodes are in the cloud and worker nodes are Raspberry Pis in a remote location and both of them are in different networks. So when I am deploying my applications on the worker node pods can’t access the internet. However when I use

hostNetwork: true

pods can access the internet. I’m missing something ?? Please help me fix the issue. Thanks, Ashis

Hi, I try to follow https://docs.k3s.io/helm#customizing-packaged-components-with-helmchartconfig. I do

apiVersion: <http://helm.cattle.io/v1|helm.cattle.io/v1>
kind: HelmChartConfig
metadata:
  name: traefik
  namespace: kube-system
spec:
  valuesContent: |-
    persistence:
      enabled: true
      storageClass: local-path
    certResolvers:
      letsencrypt:
        dnsChallenge:
          provider: cloudflare
        storage: /data/acme.json

But helm install pod always crash. Logs says it cannot parse value

Error: failed to parse /config/values-10_HelmChartConfig.yaml: error converting YAML to JSON: yaml: line 7: mapping values are not allowed in this context

Do Check This Diagram and Correct Me.

I am new to k3s. But from what I understand, certain components are not managed within k3s process. These include:

Deployment/local-path-provisioner

,

Deployment/coredns

,

Job/helm-install-traefik-crd

,

Job/helm-install-traefik

,

Deployment/metrics-server

,

Deployment/traefik

.

Using k3s with 2 PC (1 server and 1 worker) and deploy Nginx apps How expose the Nginx app in the worker2 node (POD: 10.42.2.3) to the Internet?

Using WSL2 with k3d

Hi guys, after resizing a VM and joining it back to the cluster, the node is in state NotReady and I have an authentication error in the logs - Unable to authenticate the request due to an error" err="[invalid bearer token, Token has been invalidated].

I checked the tokens etc. and it looks good. Any help will be appreciated

Hi I'm trying to use in a mac m1 k3s using lima, also tried k3d using rancher but in both cases I have an issue with ingress, is generated but I cannot access the IP in this case the ingress is in ip

192.168.5.15

but when I ping or curl I get timeout, does anyone know if I'm missing something ? thanks in advance 🙂

There are a lot of High CPU use issues reported in GitHub. I sympathize with this latest post https://canthonyscott.com/k3s-cpu-usage-bug-remains/. Most of my tests are on arm64 (RockPi 4). I have tested versions from v1.24.x until v1.27.1. The use case is very simple: just start

k3s.service

and

k3s-server

takes >50% of CPU and a good chunk of RAM too.

k3s check-config

passes. Checked the k3s logs but it looks normal to me. What else could be done to address the High CPU use on arm64?

Hi community, During the K3S installation I have specified

resolv.conf

file:

INSTALL_K3S_EXEC="--resolv-conf=/etc/rancher/k3s/resolv.conf.

I changed

resolv.conf

content, how to apply those changes to the K3S cluster? Is it enough to reboot the service? Is there any need to add resolv.conf on all agent nodes or enough to add only on masters? Br,

For a 5 node cluster where HA is paramount, I think the best option is 3 embedded etcd servers and 2 agents instead of maybe running 5 servers with no agents. This allows it so apps aren't primarily run on control plane. any opinions in regards to this number of nodes? I don't think a cluster with only control plane nodes is normal.

@handsome-autumn-77266 you can taint the nodes?

Hello, I just upgraded our k3s from 1.23.6+k3s1 to 1.24.13+k3s1 in one of the node. After upgrade I see the pods are in CrashLoopBack state throwing below error:

Error: failed to create containerd task: failed to create shim task: OCI runtime create failed: runc create failed: unable to start container process: error during container init: error mounting "cgroup" to rootfs at "/sys/fs/cgroup": stat /run/k3s/containerd/io.containerd.runtime.v2.task/k8s.io/91a340ede5612d6b6d84046f23babf7263e418c8040f44b6bded8802820134e8/rootfs/sys/fs/kubepods-besteffort-pod2e3902cc_81b9_4dba_8157_cb7bccf8eb27.slice/cri-containerd-longhorn-ui.scope: no such file or directory: unknown

Any idea what may be going on?

Have a k3s cluster that was created from Rancher UI several months ago. Recently one of the etcd / control plane node shows (in Cluster Management screen):

Configuring bootstrap node(s) custom-1ca8010bce3f: waiting for probes: kube-controller-manager, kube-scheduler

kubectl get nodes

shows everything is normal. K3s version: v1.23.17+k3s1 I need some help understanding where the issue is?

Anyone know a solid learning path for k3s? I'm a big fan of udemy type stuff, but they're pretty empty in the k3s department.

Hello folks, There seems to be an issue with the k3s uninstallation on an airgapped environment(no internet) with the k3s-uninstall.sh script. The uninstallation hangs for too long at the

yum remove -y k3s-selinux

step. Since there is no internet this step times out after a long time(5+mins I guess). Shldn't we handle this case specifically for the airgap uninstallation? Tried searching, but cldn't find any issue in reference to this. Would appreciate any help on this.

My homelab is a k3s cluster running on a handful of raspberry pis, using an external mysql database for k3s datastore. Most of these pis have just 1Gb of memory, and with just the basics installed they are already using 40%-50% of available memory, with the server using over 80%. I'm always on the lookout for ways to conserve memory, and recently someone mentioned cilium being "better" than the other CNI implementations. Does anyone have experience with the resource usage of various CNI implementations compared to flannel in a k3s cluster?

HI there! I have a question of there are any instructions on how to run Rancher using k3d on a Mac? I seem to hit a wall trying todo so. I am using Colima but cannot get Rancher going. Any hints or tips?

Hi all My DNS is not working from pods running on worker nodes, getting ";; connection timed out; no servers could be reached" error. I have tested by ssh to pods running on worker and tried "nslookup google.com 10.43.0.10" command where 10.43.0.10 is ip of my kube-dns service. "nslookup google.com 10.43.0.10" command is not working from worker node directly either, where as only ""nslookup google.com" is working fine on worker node "nslookup google.com 10.43.0.10" command is working on pods running on Master node as well as from master node too. Please check the attached screenshot. Everything was working and tested on the same cluster earlier but started failing suddenly recently, and now it is not working on pods on worker nodes as well as from worker nodes Please help!

Hello. I am having trouble getting k3s + multus to pass traffic. i have modified the multis-daemonset deployment and the pods come up with IP addresses on the secondary network. traffic seems to be hitting the bridge that is created, but traffic doesn’t seem to get to the other pod. is there a system setting i am missing. i feel like it must be something obvious, but not sure what

Hi guys. One quicky, does the install script support Fedora CoreOS 38? It complains that it can’t download repodata/repomd.xml. 🤔

Hey all, been looking for details on customising initial installs of k3s to make it easier to spin up multiple clusters and merge kubeconfigs. Basically I'm looking to specify the default user account from default to bar. Would appreciate any assistance