Brian Utterback writes:
> There are replacements for ntpdate, ntpq and ntptrace which have the
> same name, but
> they are pretty much backwards compatible. This is something I will be
> looking at
> more closely to see if there are any caveats.
I'd simply ship them under different names or under a different
directory. This is just a beta test, right? Why try to cause
trouble?
> There is a manifest and method to go with ntpd. The method is called
> xntp and the
> manifest is ntp.xml. So, the methods could happily co-exist, but the
> manifest has
> a problem. I suppose we could call the manifest ntp4.xml and the method ntp.
That should be fine.
> Unfortunately, SMF currently is happy accepting "ntp" when a FMRI is
> expected.
It's "happy" because that's the abbreviated name of the FMRI:
svc:/network/ntp:default
^^^
If you have one called ntp and the other called ntp4, you won't have a
problem.
Another option would be to create this:
svc:/network/ntp:ntp4
... analogous to svc:/network/physical:default and
svc:/network/physical:nwam, which are alternative implementations of
the network start-up mechanism.
> Since running ntpd and xntpd are mutually exclusive, it would certainly
> be nice
> to have them both in a single manifest, but since we will be delivering
> ntpd into
> the SFW consolidation and xntpd is in ON, this is problematic.
I don't see how having them in the same manifest helps. In fact, I
think it hurts a *LOT* in that you will be causing yourself trouble to
have the same file delivered by two different packages.
Don't confuse manifests (an internal implementation detail) with the
administrative features -- the FMRIs.
> On the other hand, we might very well be able to replace all the current
> bits with
> the new ones. The ntpd and xntpd daemons are very nearly compatible and
> accept
> very nearly the same configuration options. Or I should say that ntpd is
> backwards
> compatible with xntpd, except for the keywords that we added at Sun.
> Again, I
> will be looking at the compatibility issues more closely.
>
> But there still are the man pages for ntpdate, ntpq and ntptrace which
> are already installed.
I'm still confused about why you would do this, when installing it
alongside the old one seems so much simpler for something that's just
a beta test.
--
James Carlson, Solaris Networking <james.d.carlson at sun.com>
Sun Microsystems / 35 Network Drive 71.232W Vox +1 781 442 2084
MS UBUR02-212 / Burlington MA 01803-2757 42.496N Fax +1 781 442 1677
