Hi Peter: On Thu, Jul 23, 2009 at 2:32 PM, Peter Doherty<[email protected]> wrote:
> I've got ganglia monitoring five cluster. At the top of the page is > an overview that shows the total nodes, and cores for the combined > clusters. And a graph with the load of all the clusters, combined. > > And ganglia has gone mad. I have about 200 nodes, but ganglia claims > it's 80,000, and climbing. Wow aren't your users happy? Unlimited capacity! :-) > It looks like each new update from the nodes is considered unique, > because it just keeps adding. > I didn't change anything, it was working fine for months, and then > this week it just went crazy. > I restarted all the gmetad processes. I removed the RRD files for the > top level grid, but that didn't help either. > > here's a picture. > http://imgur.com/1UyjN.png What version are you running and what OS? Any errors in Apache logs? Are you using ntp to keep time in sync across all nodes? Have you localized the issue to gmetad and not gmond? Have you tried running the daemons in debug mode? Hopefully something shows up with these tests. Cheers, Bernard ------------------------------------------------------------------------------ _______________________________________________ Ganglia-general mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/ganglia-general
