I have to give props to the folks at Red Hat for their latest update packs for Red Hat Enterprise Linux AS/ES/WS 3 and 4. In the past there was always at least one nasty surprise in the updates. One update pack nuked our named.conf files. One added the audit subsystem, and turned it on, so that ten days later all, and I mean all, of our machines had full /var/log filesystems (and you couldn’t log in). The last updates added a new ksh package that uninstalled the pdksh that shipped with the OS and replaced it with a really buggy ksh. The old pdksh wouldn’t reinstall unless you removed ksh and installed the pdksh RPM manually. Every time you ran up2date it’d replace it with the crappy ksh again. Who thought that up?
This time it was straightforward, and just worked. We just finished a long-running deployment of the patches, anticipating the worst despite a month of careful testing and limited deployment, and had no problems at all.
Red Hat, I don’t know what you did with your QA processes but they’re working better than before. Thank you!