-
Notifications
You must be signed in to change notification settings - Fork 3
Trending Memory Per Process
Dave Carroll edited this page Jul 10, 2016
·
4 revisions
I like to trend everything I can about systems. Trending the VmRss size per process can be valuable in watching for spikes or abnormal behavior after deploys or other changes in one's environment. My Ganglia procmem module does just this. It scans the proc tree for a list of interesting processes by name, sums the VMRss sizes by name (in the case of apps like Apache Web Server where there will be numerous pids) and graphs them.
Below we see my Gmetad/Gweb system showing the various components I trend over time. (looks like my Apache usage has really increased over time!)
(more later on this topic)
Home
Improving Gweb and Gmond Performance
Ganglia DynamicGraph
- Overview
- DynamicGraph: Gmond Example
- Dynamic Metric Sweeper
- DynamicGraph - Add A Config File
- Managing Metric Graphs
GangliaRest API
- GangliaRest Part I - Web
- GangliaRest Part II - Redis
- GangliaRest Part III - Indexer
- GangliaRest Part IV - Options
- GangliaRest Part V - Fun Example
Metric Modules