tag:blogger.com,1999:blog-1978652979840829013.post6350067892513349341..comments2024-02-13T07:05:41.069-08:00Comments on sFlow: Using Ganglia to monitor Java virtual machinesPeterhttp://www.blogger.com/profile/00856599914190257147noreply@blogger.comBlogger1125tag:blogger.com,1999:blog-1978652979840829013.post-33686684153553962112015-07-10T16:43:38.214-07:002015-07-10T16:43:38.214-07:00Hello,
Great tool, I just had a quick question. I ...Hello,<br />Great tool, I just had a quick question. I use jmx-sflow-agents to monitor JVM data in a hadoop cluster. Basically, jmx-sflow-agents send data to the host sflow on the node which sends data to a gmond server. I then monitor this with g-web, which works great. Since there is multiple JVMs per node, I have multiple_jvm_instances set to "yes". This is working, and when I go to view graphs, I can see JVM data in the form of xxxxx@yyyyy.zzzz, such as 11233@hadoop.test.<br /><br />I was wondering, is there any way to see aggregate JVM data like you can CPU or physical memory data? For example, if you go to view idle cpu graphs, there is an entry on the graph for each sFlow host. Can we do the same thing for JVM metrics for each JVM monitored? If not, how is jmx-sflow-agent useful for monitoring hadoop clusters other than just the namenode, resourcemanager, etc...? Thanks for the feedback! Great tool!JamesMhttps://www.blogger.com/profile/06903164850484916486noreply@blogger.com