On Sun, Feb 8, 2009 at 2:43 AM, Ben Rockwood <b...@cuddletech.com> wrote: > Jason King wrote: >> What I wanted to do initially was expose the data used by vmstat, >> mpstat, and fsstat -- I think the data being exposed is stable enough >> that we could do that. Possibly also include some ZFS arc data as >> well. > > All of these are simple kstats. For fsstat data you will likely want to > wrap it a bit to expose based on the filesystem name rather than fs_id.
Well yes, you wouldn't want either the underlying kstats (which need extra information to make sense of), nor the final numbers the *stat tools present, but some stable intermediate representation. (And one that doesn't lose precision via rounding.) > The primary reason I suggest a kstat passthrough is that I'm pushing to > get more ZFS data available via kstat... As am I. Perhaps we need a little bit of coordination there - I've been thinking in terms of filing RFEs, but if we could all get on the same page... > but if the MIB even was a > wrapper around select kstats, it could be implemented in such a way that > addition of additional data would be straight forward to anyone who had > a specific need. As far as snmp goes, the underlying kstats are merely an implementation detail. >> I have an initial stab at a MIB that I wrote up Friday I could post >> this week (for discussion) if you think that would be useful, but I >> didn't want to get too deep into implementation for the discussion. >> > > I would love to participate in this where I can. I've spent a lot of > time working with Net-SNMP but I admit that I've never really figured > out how to implement a MIB. If you've got a working model I'd be > extremely interested in filling in my learning gap. I haven't much working knowledge of snmp, but I do have a reasonable knowledge of kstats and the *stat tools, so I would also love to participate. (Especially if we could see ourselves getting to the point where we could expose all the data we needed without dedicated agents.) > I am curious though, if this became a project would it not be best a > project of the Observability Community? It would. But, the Observability community has no core contributors and thus can't sponsor projects. I've tried to fix that a couple of times without much success. I note that one of the projects listed under the Observability community is to update the network MIBs, which would be a complementary effort. Doesn't look very active, though. -- -Peter Tribble http://www.petertribble.co.uk/ - http://ptribble.blogspot.com/ _______________________________________________ perf-discuss mailing list perf-discuss@opensolaris.org
