This should probably have some tests

[j-baker]

In the past I've spent considerable time debugging and fixing base images where the base image has at some point in its lifecycle broken for one reason or another. One recent example might have been... the Java vendor version string switched from something like 11.0.3 to 11.0.5.3 and some code in an older version of Hadoop broke and Hadoop then wouldn't start up without itself being upgraded. When issues like this occur, the bug is taken from time of introduction to the first time when someone is blocked by it, which may be indefinitely later and when they're under some time pressure.
Where this base image is 'doing things' - installing kubectl, installing random versions of python, etc, basically stuff beyond using the linux package manager to install packages, I suspect it's worth a quick bash script which does something with each of the required packages - just to make sure that the requirements of the software haven't changed. With this, we can have some peace of mind that functionality keeps working over time.