This message was deleted.

that’s the state dir, which contains root filesystems for running containers. Space used there is directly related to what’s running on your node. Sounds like maybe you have some very large images?

@creamy-pencil-82913 Thank you for your return and interest. In my cluster, only the following applications are running. The following applications I added later in the Rke2 cluster: Rancher UI Cert-manager, Rancher-Monitoring, Longhorn, I restored a lot of etcd in this test cluster, could this be related to this?

what is the actual problem you’re troubleshooting here? These all look like normal mounts. I see you’re only using 61% of disk so that wouldn’t be anywhere near the 85% gc threshold that would trigger image pruning.

@creamy-pencil-82913 Actually there is no problem, I'm just trying to understand it better by consulting with you. Thank you for your return. Could it be harmful for the rke2 cluster to use the kubelet configs that I mentioned in the production environment? Do you recommend? When I researched on the internet, I learned that I could reduce disk pressure.

you don’t have any disk pressure though

@creamy-pencil-82913 thenk you for your interest. Should I use these settings only when there is disk pressure? If I use it before, is it possible to run kubelet image prune when the disk pressure reaches 85%? When we run our own containers in the production environment, the disk pressure here will constantly increase. Can I prevent the disk from filling up and becoming inoperable with these settings?

that is the whole point of the threshold. It prunes images when it gets over the high threshold, until it is below the low threshold. This is covered in the Kubelet docs.

@creamy-pencil-82913 Thank you for your help. It is written in the kubernetes documentation that these settings will not harm the containers and images related to the pods. It cleans old and unused comtainers and images, doesn't it?

I would defer to the Kubernetes docs

Thank you so much.