On Fri, Oct 10, 2008 at 11:16 AM, Buccaneer for Hire. <[EMAIL PROTECTED]> wrote:
> We have a new cluster-1250 compute nodes, 10,000 cores in 3 rows. Each row > is a composed of two subnets, Gb management and 10Gb for production. > > We would like to have each gmond multicast in its own subnet have a gmetad on > each row on a ganglia node. The only exception is management nodes which > will unicast to a different port on Row 3 ganglia node. Once this is in > place, we will collate all rows into once higher level ganglia server. At > this point, I should be able to see a total row, and then 4 rows of > items-rows 1,2,3 and MGT. > > We would then like to federate that information to our global ganglia server > exporting the entire cluster as one unit. So these management nodes -- are you expecting them to be multicasting within their own row and also unicasting with the Row 3 ganglia node? Cheers, Bernard P.S. With that many hosts, you are bound to run into the I/O issue with gmetad/RRD updates. You may be interested in the recent work that is being done with rrdcached[1]. [1] http://ganglia.wiki.sourceforge.net/rrdcached_integration ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/ _______________________________________________ Ganglia-general mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/ganglia-general
