James Carlson wrote: > Bart Smaalders writes: >> I think we really need to move towards a model where upgrade & >> patching really only adds or removes files from the system. Any >> other sort of manipulations in the context of the alternate OS >> seem fraught with hazard and difficult to test issues. > > I mostly agree, but as long as application programs (such as SMF) have > private data stores that may need 'tweaking' during the upgrade > process, I don't think that's always possible. > > The alternative is building things that run _after_ the reboot and are > able to complete the upgrade operation later. These are also fraught > with hazards. >
But this is pretty testable, and has a much more constrained test matrix. Today, we need to worry about: (2 older releases) * (2 global vs local zone) * (2 local vs alternate root) * ( 2 archtectures) (ignoring patch vs upgrade) I'd much rather worry about just 2 older releases. - Bart -- Bart Smaalders Solaris Kernel Performance barts at cyber.eng.sun.com http://blogs.sun.com/barts
