|
Hi Roberto
> >> Ok, so we need a #define for this or people on a alpha will not be
> >> happy.
> >>
> >
> > to include <sys/param.h> and appropriate arithmetics using "HZ"
> > would be the most portable implementation I guess.
>
> Fair enough. Would asm/param.h be enough?
>
Ack.
>
> Maybe we could sit together once and talk about it some day?
>
Yes. When I remember correct you live in Zurich Chreis 5 (at the end of
the lvs presentation there where discussiona about ADSL providers
and you mentioned to live in Zurich 5)
I live in Zurich Wipkingen so it's several minutes by bike. I suggest to
meet in a bar and discuss this while having a beer.
>
> Noone seems to have time but I think to module needs to be maintained
> somehow.
>
Ack.
> > I had some weired thougths in my mind about a "one-stop-SNMP-shop"
> > for lvs and keepalived (first of all the vrrpd2) which provide
> > together a good HA / LB setup.
>
> Do you think you can merge those in an appropriate way?
Sorry the "one-stop-shop" was misleading. I don't want it to be merged.
From my point of view they should be independent modules and MIBs
I just meant that to have an snmp implementation for both would be more
valuable than just the one for lvs.
>
> > The interesting part ist that vrrpd2 which is defined in RFC 2338,
> > has also defined an MIB in RFC 2787 (
> > ftp://ftp.isi.edu/in-notes/rfc2787.txt ). I think that could be a
> > very interesting combination.
>
> But then, if I understand you correctly, you cannot extend that rfc2787
> MIB. It's defined in a fixed way. Or what exactly is your intention?
I know you can't change an RFC (not even spelling mistakes).
I just meant, that while designing the definitiv MIB for the lvs part
taking care of that the two MIBs would be consistent in structure and
naming
conventions etc.
cu
romeo
|
