On Tue, Sep 28, 2004 at 10:44:30AM +0200, Robert Olsson wrote:
> Stephen Hemminger writes:
> > > Harald Welte <laforge@xxxxxxxxxxxx> wrote:
> > >
> > > > As stated before, I would like to change rt_stat and ct_stat in order
> > > > include a first 'template' line, too. This way it is easier to write a
> > > > generic foo_stat program, that could deal with any of those statistics
> > > > files, even with new ones... but this of course would break existing
> > > > rtstat binaries. I personally don't care, since it's a little-known
> > > > and little-used feature, which to my knowledge is in a lot of
> > > > distributions either non-existant [Debian] or incompatible [SuSE]. What
> > > > do you think?
> > >
> > > I agree. And while we're add it let's get a fixed rtstat into
> > > iproute2 and make sure that binary gets installed by default
> > > so maybe the dists will start shipping it properly.
> > I have the old one in the repository.
> I sent you the latest rtstat (Martin sent ctstat) during netfilter workshop.
> Remember? Having a common foo_stat sounds like a good idea.
I'm working on this right now. The plan is to make the parser code
shared between a commandline tool and some daemon that can be run on
production systems to gather long-term statistics. You can select the
fields you're interested in with keys like "rt_stat:entries". Maybe
I'll even add stuff like mysql/pgsql-logging like ulogd has, but that
future plans for now.
I've called it lnstat for now (as in linux networking statistics), but
I'm open to better suggestions on the name ;)
Please find the current kernel-part patch attached to this mail,
it adds teplate-headerlines and moves all statistics files to
/proc/net/stat. It even cleans up the neigh_stat code a bit.
Depends on my latest neighbour statistics patch that davem has already
- Harald Welte <laforge@xxxxxxxxxxxx> http://www.gnumonks.org/
Programming is like sex: One mistake and you have to support it your lifetime
Description: Text document
Description: Digital signature