Does anyone have running clusters on docker with cgroups v2 (systemd driver)?
I am wondering which is the best approach to fix the docker configuration until it is supported by rancher and the agent is passing the correct parameters for v2
I have described the issue here:
here .
So basically how do we configure the daemon?
1. Leave it as it is by default and modify individual containers - cons - until rancher is passing the correct parameters it would be very unrealiable
2. Change the
default-cgroupns-mode
to
host
- cons - all containers will be running with
CgroupnsMode: host
3. Change the docker driver to cgroupfs v1 ?