Basic collectd-based server monitoring. Sends stats to Graphite.
With inspiration from andreasjansson/docker-collectd-write-graphite. But not using python to replace the settings.
- CPU used/free/idle/etc
- Free disk (via mounting hosts '/' into container, eg: -v /:/hostfs:ro)
- Disk performance
- Load average
- Memory used/free/etc
- Uptime
- Network interface
- Swap
HOSTNAME
- Will be sent to Graphite
- Required
GRAPHITE_HOST
- Graphite IP or hostname
- Required
GRAPHITE_PORT
- Graphite port
- Optional, defaults to 2003
GRAPHITE_PREFIX
- Graphite prefix
- Optional, defaults to collectd.
REPORT_BY_CPU
- Report per-CPU metrics if true, global sum of CPU metrics if false (details: collectd.conf man page)
- Optional, defaults to false.
INTERVAL
- Controls how often registered read functions are called and with that the resolution of the collected data. (details: collectd.conf man page)
- Optional, defaults to 10.
docker run \
--net=host --privileged \
-v /:/hostfs:ro \
-e HOST_NAME=myhost \
-e GRAPHITE_HOST=your.graphite.com \
--name collectd pboos/collectd-graphite