[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: ITP: xtradius -- Free radius server implementation.

On Tue, 9 Oct 2001, Russell Coker wrote:

> On Mon, 8 Oct 2001 16:27, Steve Langasek wrote:
> > On Mon, 8 Oct 2001, Francesco P. Lovergine wrote:
> > > You should conflict all current RADIUS servers, which are

> > > radius-livingston
> > > yardradius
> > > radius-cistron

> > > This is not strictly needed if you provide the same binaries 'radiusd'
> > > Anyway notify me when done in order to check yardradius package
> > > installation.

> > If this is to be done by conflicting with all radius servers (rather than
> > with a virtual 'radius-server' package?), might I suggest also adding
> > freeradius to this list?  There is no .deb for freeradius in the archive
> > yet, but given the density of Debian developers on that project I expect
> > one to appear once the code stabilizes.  It would be nice if it could be
> > integrated cleanly into Debian systems once that happens.

> Why is it so desirable to have all the RADIUS servers conflict with each
> other?

> Obviously if two packages contain a binary with the unimaginitive name of
> "radiusd" then they can't both be installed.  However if there aren't any
> binaries or configuration files conflicting then I think that installing two
> RADIUS servers at once should be possible.

Ah, well, I have not yet seen a radius server that /didn't/ call its daemon
'radiusd'.  Certainly freeradius follows the pattern set by radius-cistron (of
which it's a rewrite) and radius-livingston (on which Cistron is ultimately
based).  They also use the exact same config files, in the exact same
directory (/etc/raddb/, of course -- how intuitive), and they all listen on
the same ports (ok, choose from a set of two).

I can see the argument in favor of having non-conflicting packages, but that
would require a lot of work and restructuring -- both of the packages and of
the users' expectations.

Steve Langasek
postmodern programmer

Reply to: