Chris,
Looks you have reached the limit of an unsigned 32 bit integer (4TB). The solution I implemented for this was reporting memory metrics, per host, in MB. That pushes the aggregate 4TB limit away quite a bit. The long term solution would be implemented 64bit metrics - I started working on that awhile ago, but it kept getting pushed down on my priority list. If you like I can send you a patch to Ganglia's code which brings down the granularity of memory metrics to MB. Keep in mind that all your previously collected memory data will be lost, unless you scale it down in the RRDs (I never bothered doing that, as I encountered this issue early in the test phase). Cheers, Alex Chris Slaughter wrote: > I am using ganglia on several systems, including 2 large clusters. > > One of the clusters has 4TB of memory, and the second has 256GB. > Individually the memory graphs are correct, showing 4TB and 256GB. > The summary graph, however, shows no statistics, and shows a > total value of 300G. Have I reached some limit in rrdtool or > ganglia? Thanks > > Chris Slaughter > > > ------------------------------------------------------------------------- > Take Surveys. Earn Cash. Influence the Future of IT > Join SourceForge.net's Techsay panel and you'll get the chance to share your > opinions on IT & business topics through brief surveys -- and earn cash > http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV > _______________________________________________ > Ganglia-general mailing list > [email protected] > https://lists.sourceforge.net/lists/listinfo/ganglia-general >
