I totally agree. It is not doing that, however. I believe that Rancher starts installing the version you just said as it creates the docker group, but then somehow craps out. After that, all I get from the rancher console after it creates the node is “Waiting for SSH to come online” which it tries 60 times and then starts the whole process over again. This should be a simple operation, but it is not and thus why I did a workaround that I shouldn’t have to do. This does not work in 2.4.2 with CentOS 7.8 at least. How do I get this fixed. I have attached some workarounds, but should need to do this.

Nope as there is very little about creating a VM template in your documentation. That would be really helpful to see for each OS. Here is how I essentially “sysprep” the systems.

Yeah, I get that to a degree, but the OS template also plays a role in getting the cluster provisioned, so I think there should be more guidelines in your docs. My customer only uses RHEL/CentOS so I have to make it work using those OS’s. Your docs could just provide a primer that says, at a minimum, do this. As well, for CentOS do that. Basically what I do to create the template is this. The template is made by selecting the “Cloud Compute” option during install, updating the OS, installing cloud-init and nfs-utils, then “syspreping” it.

To get around the docker install failure, I have to do this in cloud-config, which I shouldn’t have to do.