OL7.9 high docker memory/cpu usage after Rancher / Kubernetes upgrade
Our Rancher environment runs on Oracle Linux 7.9. We recently updated the environment to Rancher v2.6.8 (RKE v1.3.14), and all k8s clusters in the environment to k8s v1.24.4. There has since been an episode of cluster instability, and we suspect it was related to dockerd's very high CPU consumption. In the Rancher repository there is the following recommendation:
https://github.com/rancher/rancher/issues/38816#issuecomment-1259798220
Here is the relevant part of the comment above:
Add the following to /etc/modprobe.d/disable-overlay-redirect-dir.conf
:
options overlay metacopy=off redirect_dir=off
Then, you can stop docker, reload the module, and restart docker to check your new status: