Performing node maintenance

There are scenarios where you may want to perform a node maintenance activity, such as the following:

When applying security patches;
When performing an operating system upgrade;
When changing any network configuration;
When performing any other activity that your organization mandates.

While performing node maintenance operations, it is possible that you may accidentally break the cluster. To avoid any adverse situation, follow the guidance provided here.

Note:

UiPath® does not provide guidance on how to perform node maintenance activities. You must contact your IT team for this.
The following guidelines only provide instructions on the steps you must take before and after the node maintenance operation, to ensure the cluster is healthy.
It is good practice to perform node maintenance activities on one node at a time.

Pre-node maintenance

To ensure that the cluster is healthy while you are performing node maintenance activity, you must drain the workloads running on that node to other nodes. To drain the node, run the following command:
```
systemctl stop node-drainsystemctl stop node-drain
```
Stop the Kubernetes process running on the node. Run either of the following commands:
- Server node:
```
systemctl stop rke2-serversystemctl stop rke2-server
```
- Agent node:
```
systemctl stop rke2-agentsystemctl stop rke2-agent
```
If your maintenance activity includes upgrading the RPM packages on the machine, you must skip upgrading the rke2 package to avoid any compatibility issues.
- It is recommended to add the rke2 package to the exclusion list of the RPM upgrade. To modify the /etc/yum.conf file, add rke2 in exclusion. For details, see these instructions.
- Alternatively, you can temporarily exclude rke2 during yum upgrade using the following command:
```
yum upgrade --exclude "rke2-*"yum upgrade --exclude "rke2-*"
```
  Important: If not excluded, rke2- packages might get upgraded to the latest version, causing issues in the Automation Suite cluster. rke2-* package upgrade will be handled via the Automation Suite upgrade.
Proceed with your node maintenance activity. Once the upgrade is complete, continue with the post-node maintenance activity.

Post-node maintenance

Reboot the node either by running sudo reboot or by using any other safe reboot mechanism you may prefer.

The RKE2 service should start automatically. To check the status of the RKE2 service, run the following commands:

Server node:

systemctl is-active rke2-server.service || systemctl start rke2-server.servicesystemctl is-active rke2-server.service || systemctl start rke2-server.service

Agent node:

systemctl is-active rke2-agent.service || systemctl start rke2-agent.servicesystemctl is-active rke2-agent.service || systemctl start rke2-agent.service

If the RKE2 service did not start automatically, start it manually by running the following commands:
- Server node:
```
systemctl start rke2-serversystemctl start rke2-server
```
- Agent node:
```
systemctl start rke2-agentsystemctl start rke2-agent
```
Once the RKE2 service is started, you must uncordon the node to ensure Kubernetes can now schedule workloads on this node. To uncordon, run the following command:
```
systemctl start node-uncordonsystemctl start node-uncordon
```