Hy I'm currently using ganglia to monitor a cluster. The cluster is composed by physical nodes, each running a Xen hypervisor (v3.2) . A gmond is running on each dom0 and domU in a unicast mode. They send datas to an additional gmond running on a service node. On this node, a gmetad and the web frontend are running.
Data gathering works fine in the dom0s. However, I have issues with the domUs: All the gmond continuously send datas but the cpu_user metric is up-to-date : When I launch a program that consumes all the cputime, the cpu_user metric is updated to 100%. When I stop the program, the value does not return to 0 (using 'top' on the domU show a correct value) but gmond keep sending a cpu_metric equals to 100% (TN fields are refreshed) Anyone know a similar situations with or without virtualization ? ------------------------------------------------------------------------------ This SF.net email is sponsored by: SourcForge Community SourceForge wants to tell your story. http://p.sf.net/sfu/sf-spreadtheword _______________________________________________ Ganglia-general mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/ganglia-general
