Thank you for the replies. My main takeaway is do the major LTS upgrades "rapidly-sequentially": for example, make a one week "pit stop" at 24.04 while upgrading from 22.04 to 26.04 .
Practically speaking, re-learning Linux system details takes a few weeks for me, even with copious notes from "last time". My profession is hardware engineering, the details are physics and chemistry (and lately, orbital mechanics). Relearning Linux OS details for maintenance is distracting. A lot of those details are encoded in home-made simulation code connected to math libraries - I must transition that work too. The VM idea sounds be a good way to encapsulate some of that custom work - along with notes about how I did it and the software I used to do it and where I found the software. The downside of VM is that container systems themselves go in and out of fashion; before I do that, I hope to learn which virtualization is most likely to endure ... for decades. Being 68 years old, "decades" seems optimistic ... but I just signed a birthday card for my 105 year old father-in-law, who led building repair teams as recently as ten years ago. Some of my own distributed team is in their 20's, so I probably won't be able to answer questions ("what the HELL was Keith thinking???") 50 years from now. Simulation plus VM - hopefully the VM overhead added to a kludgy numerical simulation spread out across multiple processor cores won't slow it down by more than a factor of two. I ponder migrating some of those nested C loops graphics coprocessors, but that seems incompatible with abstract virtualization. One of the advantages of Ubuntu/Debian over CentOS/RH is that there are fewer copyright/ownership hassles for the proliferation of third party repositories for source archives. And there are enough renegade geeks maintaining oddball variants that I can probably find archives compatible with the in-the-future-unfashionable variants that I've chosen. So -- thanks again! Keith P.S. "Clock of the long now" ... how do you format a maintenance manual so that it is readable 10,000 years from now? -- Keith Lofstrom kei...@keithl.com