But why? If you have a docker application stack and salt configuration management, wouldn’t it be easier and better practice to spin up a new node to replace it? I’m curious what is critical about this version of docker that special effort was made to keep it on that version.
It might be physical hardware where an in place upgrade is easier than racking up new servers?
I’ve got physical hardware where I regularly write 0s to the first meg of the disk, reboot, and watch a pxe based autoinstall replace the OS and run a salt highstate.
It’s easier, once the infrastructure is set up, to just replace systems with new ones, rather than updating.