On Thu, May 22, 2008 at 5:07 PM, Nuno Covas <[EMAIL PROTECTED]> wrote: > On Wed, May 21, 2008 at 6:33 PM, Joao Ferreira > <[EMAIL PROTECTED]> wrote: >> >> On Wed, 2008-05-21 at 19:20 +0200, Andrew Beekhof wrote: >>> On Wed, May 21, 2008 at 11:54 AM, Joao Ferreira >>> <[EMAIL PROTECTED]> wrote: >>> > >>> >> >>> >> what was the motivation for this feature? >>> > >>> > IP aliasing limitations. Our kernel does not yet support IP alisaing. >>> > It's on the roadmap but we could not wait for it. >>> >>> dare one ask what kind of kernel doesn't support IPs? >> >> well.... my dab... it's not really the kernel.... the kernell supports >> it just fine... the problem is that our own software would behave >> incorrectly in other un-related areas if we were to use IP-aliasing... >> >>> >>> > So basically we had to use and "IP-ignorant" heartbeat.... our heartbeat >>> > is totally unaware of IP addresses.... always uses MAC addresses. When >>> > we need to perform IP take over we use a diferent approach. >>> >>> did you implement it as a separate comms plugin or some other method? >>> because if its a new comms plugin we'd probably accept it into the >>> tree if you're interested > > I wrote it as a HBcomm plugin, but didn't create a new one. Instead, I > modified the ucast plugin so I could patch Heartbeat to fit our needs. > Anyway, I guess it wouldn't be hard to make a specific MAC-based ucast > plugin for that kind of use, although Heartbeat would have to depend > on libpcap from now on. Would you be interested in taking a look at my > code, Andrew?
I doubt I'd understand it to be honest - I never did take the time to understand the comms layer completely :-) But if you post the code, it might be useful to someone and/or you might get some feedback. _______________________________________________ Linux-HA mailing list [email protected] http://lists.linux-ha.org/mailman/listinfo/linux-ha See also: http://linux-ha.org/ReportingProblems
