On Sun, Feb 10, 2008 at 11:55:34AM -0500, Jesse Becker wrote: > On Feb 10, 2008 2:23 AM, Carlo Marcelo Arenas Belon > <[EMAIL PROTECTED]> wrote: > > On Fri, Feb 08, 2008 at 06:24:05PM -0800, Ben Rockwood wrote: > > > I then found that I could use multicast in a non-multicast > > > environment on Solaris and just let it all become broadcast. So now I > > > have gmond's on each subnet working together via broadcast, however I > > > need to "bridge" the gmond's together. > > > > you are abusing the application and the network to send a UDP packet to a > > broadcast IP, it will be better to fix the source of your environmental > > limitations instead of hacking around them. > > Actually, there is some precedent for this (ab)use of > unicast->broadcast traffic, and it can be really useful on occasion, > and if use carefully.
agree that is useful but is expensive (broadcast propagation is usually done by CPU in the switches) and when is really needed (because the application protocol relies on it as done by CIFS name resolution) should be configured appropriately : http://cisco.com/univercd/cc/td/doc/cisintwk/ics/cs006.htm ganglia's protocol (by gmond) doesn't use broadcast by design. > However, in this case it is probably best to just have gmond poll one > host in each subnet, instead of trying to send traffic between the > subnets. I am not sure what the restriction for this environment was (as it wasn't explained) but if UDP traffic is blocked between VLANs and UDP broadcast is allowed inside the VLAN then it is broken (as being more insecure than if the opposite were true). I agree that enabling multicast routing between VLANs (and increasing the TTL of the UDP metric packages) is a bad idea, but see no reason why UDP unicast between VLANs would be a problem. Carlo ------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2008. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ _______________________________________________ Ganglia-general mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/ganglia-general
