As to other points:
There are significant discounts for .edu-land sites going to RHEL.
There IS a migration path from previous RHL to RHEL. I took the time to register to download their migration whitepaper (after all, I'm interested in RHEL professionally for PARI (www.pari.edu) anyway, might as well), and found out the scoop in their White Papers. Their registration process is annoying! But I got the papers. Pretty good stuff. They explain why there's no direct upgrade, and how to migrate.
Was I astonished when the Red Hat Linux Project was first announced? No, there were tremors three releases ago about this; anyone watching closely enough saw it coming. The merger with the fedora 3rd party repository likewise was telegraphed at low baud.
I continue to run Red Hat and now Fedora in production. It's stable, it's well supported by a large community (see fedora.redhat.com about joining the community). Much the same as the support structure behind OpenACS/dotLRN itself, just larger. Where do .LRN and OACS users go for 'errata?'
Now, as to upgrading. A server has to be taken down to upgrade regardless. I typically do a fresh install after a backup anyway. I then recompile my third party stuff (AOLserver, etc) and restore the data. I plan it and execute it. I always had problems using the 'upgrade' option (being an RPM packager I understand the reasons, too). So I don't even bother at this point.
Frankly, I am astonished at the Chicken Little tenor of this thread. You guys are OpenACS developers; creme de la creme of the developer pool. You guys should know better.