On Fri, Feb 13, 2009 at 03:07:29PM -0600, Jason King wrote:
> On Fri, Feb 13, 2009 at 2:48 PM, David Collier-Brown <[email protected]> wrote:
> >  I'll see you on the list (;-))
> >
> >  Back when I was full-time (I'm a contractor
> > who does capacity planning), I was part of the
> > ABI team, who managed the stable/unstable
> > interfaces, so as not to suffer "dll hell"
> > like some other folks we all know.
> >
> >  To oversimplify, "private" interfaces which
> > change from release to release have to be
> > tracked by outside software vendors, and
> > they really *really* don't like doing that.
> > Imagine being Oracle and having to check that
> > all your uses of SUNWprivate interfaces
> > still worked whenever you got a new OS release.
> >
> >  There are a few companies who do so, and
> > Oracle is (or at least was) one of them, but
> > everyone else wants stuff to remain unchanged.
> >
> >  In the case of a mib, which can't check a
> > library to see if something is private or not,
> > we'll have to set ground-rules so that one
> > can introduce new interface names, drop old
> > ones and never ever change the meaning of a
> > datum without creating a new interface name
> > for it.
> >  Not a small task!
> 
> I understand that.   The problem is, it seems Sun has taken the stand
> that _all_ kstats are essentially 'private'.  Which is fine, but then
> there seems to be no way to make _specific_ kstats have a more stable
> classification, though even that may not be necessary -- I'd rather
> have good useful metrics be what's 'stable' (and what you see in the
> MIB) and how they're obtained be an implementation detail.  If they
> just happen to match up exactly with an existing kstat, well that
> makes things much easier, but if that wasn't always the case, I'd
> rather change the module to still produce the expected result.
> 
> The problem is, currently the line is '_any_ kstat cannot be used by
> anyone (that isn't Sun) for any reason', so with that position, we
> kind of go in circles.  I.e. we can't say
> 'unix:foo:0:really_useful_stat is extremely good to know, and making
> what it's measuring available via the perfmib would be good', because
> _all_ kstats are effectively off limits.  That's why the *stat
> commands were chosen as a starting point -- while some of the things
> aren't useful, some are, so at least with those that are we can reply
> in defence 'we're only doing what XXXXstat is doing, and that command
> is unlikely to go away any time soon'

I think a number of us would love kstats to have the same embedded stability
semantics that DTrace has for its probes and arguments - that way, we can
specify which kstats are stable (and as engineers, keep them stable) and
which are not.  If I could work on anything I wanted, this would be high on
the list. :)

Brendan

-- 
Brendan Gregg, Sun Microsystems Fishworks.    http://blogs.sun.com/brendan
_______________________________________________
perf-discuss mailing list
[email protected]

Reply via email to