Jesse Becker wrote: > On Tue, May 12, 2009 at 16:17, Jeff Haferman <[email protected]> wrote: > >> I *could* update ganglia, but I'm hoping someone could help me out... >> I had this configuration working earlier, but then we moved the cluster >> to a different sub-net and I'm having some problems: >> >> 1) I remember when I initially setup ganglia on this system, I was not >> able to get things going, and the problem was that I had to use >> 224.0.0.1 as the multicast address in all my gmond.conf files >> (this is a special multicast address that means "all IP hosts"... >> I remember trouble-shooting and finding a post somewhere that said this >> had to be done on Apple systems... sure enough, if I use any other >> multicast address, I cannot "see" any of the machines in the cluster). > > Honestly, this sounds like either a bug or mis-/obscure- configuration > issue in OSX. >
I can't find much regarding ganglia on OSX, but this is what got me going previously... >> 2) The problem now is that when I bring up the ganglia webpage, I see >> all the compute nodes just fine, but the head node appears with its IP >> address, and not its name. I can't for the life of me figure out >> where this is coming from. > > IIRC, a reverse DNS lookup is done on the IP address. Does the > reverse lookup for the head node work as expected? This seems to be where things are hosed... we definitely need to review our network setup. I like working from the command line, and I probably messed something up that can be easily fixed if I configure from the Mac Desktop. > >> 3) We have another Mac cluster running ganglia on the same sub-net, and >> one of its nodes is showing up (by IP address only) on the ganglia page >> of the cluster that I am maintaining. I suspect this is because the >> admin for that cluster is also using the 224.0.0.1 multicast address and >> somehow my cluster is picking up its info... but I'm not sure. > > Yep, you're sending to *all hosts*, and your network switches are > dutifully relaying the traffic. > Though, Cluster A has 64 nodes, and Cluster B has 4 nodes, and Ganglia on Cluster A is only picking up one node from Cluster B, and vice-versa. I'll get this one figured out, though it would be nice to hear from other Mac users. Ganglia on our Linux systems work really well. We also have an AIX 5.3 system that we are using Ganglia with (IIRC, there is an individual that put together a special AIX ganglia build, and we simply downloaded and it just works). OSX is the problem child.... ------------------------------------------------------------------------------ The NEW KODAK i700 Series Scanners deliver under ANY circumstances! Your production scanning environment may not be a perfect world - but thanks to Kodak, there's a perfect scanner to get the job done! With the NEW KODAK i700 Series Scanner you'll get full speed at 300 dpi even with all image processing features enabled. http://p.sf.net/sfu/kodak-com _______________________________________________ Ganglia-general mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/ganglia-general
