Kubernetes Cluster

1 Sealos node notready, kubelet reports node not found¶

Log in to the problematic node and check the kubelet logs.

journalctl -xeu kubelet

The logs reveal that apiserver cannot be connected; test network connectivity.

telnet apiserver.cluster.local 6443

Testing reveals network issues; disable the firewall on the master node to resolve the issue.

systemctl stop firewalld

2 Kylin ARM architecture Sealos deployment timeout¶

Check if other containerized products have been deployed.

rpm -qa | grep docker
rpm -qa | grep podman

Redundant software is found; uninstall the redundant software.

rpm -e <redundant_software_name>

Uninstall Sealos and reinstall it.

sealos reset

3 Physical machine k8s cluster deployment, calico component fails to start¶

When deploying a k8s cluster on a physical machine, the calico component fails to start with the following error.

kubectl -n calico-system get installations.operator.tigera.io 
kubectl -n calico-system edit installations.operator.tigera.io default

# Change to interface: eth.*|en.*|bound.* What bound.* is changed to depends on the physical machine's network card name, requiring a regular expression match
nodeAddressAutodetectionV4:
  interface: eth.*|en.*

4 Kylin ARM architecture sealos deployment reports segmentation fault¶

Install the new version of sealos and the k8s cluster.

# sealos
wget https://mirror.ghproxy.com/https://github.com/labring/sealos/releases/download/v4.3.7/sealos_4.3.7_linux_arm64.tar.gz \
   && tar zxvf sealos_4.3.7_linux_arm64.tar.gz sealos && chmod +x sealos && mv sealos /usr/bin
# k8s cluster   
sealos run pubrepo.guance.com/googleimages/kubernetes:v1.25.16-4.3.7 pubrepo.guance.com/googleimages/helm:v3.8.2 pubrepo.guance.com/googleimages/calico:v3.24.1  --single  --debug

5 Selaos installation reports port conflict, 'Port: 10249 occupied'¶

Check if there is a port conflict; if so, close the conflicting port.

lsof -i: <port_number>

If the above command has no output, it may indicate an abnormality with the lsof command.

whereis lsof
# If multiple lsof commands exist and only one works normally, you can rename or move the other lsof binary commands to another path.

6 sealos cluster deployment kubelet fails to start¶

OS is openEuler 22.03 version Problem description: During sealos deployment, errors keep occurring, preventing the cluster from starting up. It is judged to be a kubelet issue; check the kubelet logs. By checking the /var/log/message log, it is discovered that the /etc/resolv.conf file does not exist. Create the file and configure the correct DNS server, then reinstall sealos.