[rescue] Sun wants money for patch bundles?
velociraptor at gmail.com
Fri Apr 13 17:43:57 CDT 2007
Speaking of patches, I also noticed that they've changed the way that
smpatch works in Solaris 10, at least I think they have.
When I was experimenting with it in our dev Sol 9 environment about 1
year ago, you could force kernel patches without going to single user
mode. Now, however that is not an option. And since smpatch requires
network connectivity, you have to use patchadd in single-user mode to
apply the remaining patches.
This wouldn't be so bad if smpatch outputted the exceptions in a
manner that could be easily fed to patchadd, but it doesn't.
Neither give really useful information for why patches fail to apply
either (not usual for Sun, but one always hopes). I'd expect better
errors given that the whole point of smpatch and the "SMC" is to
filter out the patches the box is *not* supposed to need. At this
point I am reserving judgement on SMC/smpatch until I do more
investigation, but all-in-all I can't say it's made a good first
impression on Sol 10.
On the other hand, I've read some high-level overviews of SMF, and
have written a couple of manifests (partially cribbed from other
sources) for services I've added to the box. From what I got of it
previously, I thought it was just another unnecessary "meta"-ization
of something that should be straightforward--e.g. /etc/init.d scripts.
Now I understand it's real purpose is to have the OS to be able to
"keep watch" on it's own services to make sure everything stays
running. Obviously this is not going to be 100% foolproof, but it's
certainly an interesting enhancement that people needing HA but not
having the $$ for a full-blown cluster solution could put to use.
More information about the rescue