Munin plugin for monitoring docker containers

Plugin system consists of

1. docker-spectator-daemon.py periodic collecting data from docker stats
2. docler-spectator.py for collecting data, retrieve average data and list of containers with collected data
3. docker_containers.py munin plugin

Requirements

1. Python 2.6+
2. argparse (pip install argparse)

How to use munin plugin

Download

cd /path/to/project/location # eg. `/root/scripts` or some another
# clone repo
git clone https://github.com/dvapelnik/docker-spectator

Start daemon

cd docker-spectator
# start daemon
python docker-spectator-daemon.py --start
# wait some time for data collecting

Make plugin symlinks

cd /etc/munin/plugin-conf.d
cp /path/to/project/location/docker-spectator/templates/plugin-conf.d/docker_containers_ .
cd ../plugins
# for monitoring single container
# replace [container_name] with container's name for monitoring
# replace [field_name] with field name for monitoring one of `cpu`, `mem` or `net`
ls -s /path/to/project/location/docker-spectator/docker_containers.py docker_containers_[container_name]_[field_name]
# for batch monitoring all containers
ls -s /path/to/project/location/docker-spectator/docker_containers.py docker_containers_[field_name]

Warning! Batch graph monitoring available only for cpu and mem fields not for network

Restart munin-node

service munin-node restart

Make init.d file

cd /path/to/project/location/docker-spectator
cp templates/init.d/docker-spectatord /etc/init.d/
chmod +x /etc/init.d/docker-spectatord

Attention! You should change daemon directory (DAEMON_PATH) in /etc/init.d/docker-spectatord to /path/to/project/location/docker-spectator

After this you can use init.d script for start/stop daemon which collects data from docker containers like this:

service docker-spectatord restart

or

/etc/init.d/docker-spectatord restart

How to use docker-spectator.py only

docker-spectator-daemon.py is the daemon which exec docker-spectator.py every 10 seconds This value is changeable. Look at config/config.py

Look at --help firstly:

python docker-spectator.py --help

usage: docker-spectator.py [-h] [-l {DEBUG,INFO,WARNING,ERROR,CRITICAL}] [-cl]
                           [-a {5,10,15}] [-f {cpu,mem,net}]
                           [-cid CONTAINER_ID] [-cn CONTAINER_NAME] [-cids]

optional arguments:
  -h, --help            show this help message and exit

Common:
  -l {DEBUG,INFO,WARNING,ERROR,CRITICAL}, --log-level {DEBUG,INFO,WARNING,ERROR,CRITICAL}
                        Define log level

Collecting data:
  -cl, --collect        Should collect current data?

Retrieving average data:
  -a {5,10,15}, --average {5,10,15}
                        Specify time period in minutes for retrieve average
                        data
  -f {cpu,mem,net}, --field {cpu,mem,net}
                        Specify data field
  -cid CONTAINER_ID, --container-id CONTAINER_ID
                        Specify docker container id. Container id is preferred
                        if specified both
  -cn CONTAINER_NAME, --container-name CONTAINER_NAME
                        Specify docker container name. Container id is
                        preferred if specified both

Container:
  -cids, --container-ids
                        Retrieve container ids

You can use docker-spectator.py for retrieve list of containers which data already collected

python docker-spectator.py --container-ids

fa0a12244b9f fancy_bebe
6e1a5f381010 little_penguin

You can get collect data from active containers

python docker-spectator.py --collect

You can retrieve average collected data for container by 5, 10, 15 minutes and specify container by name or id Retrieve cpu average value by 5 minutes container specified by id:

python docker-spectator.py --average 5 --field=cpu --container-id=fa0a12244b9f

0.17

Retrieve cpu average value by 10 minutes container specified by name:

python docker-spectator.py --average 10 --field=mem --container-name=little_penguin

55344679

Or network. First part of result is downloaded traffic and second is uploaded. Values separated with single space

python docker-spectator.py --average 10 --field=net --container-name=little_penguin

738 55644 

ToDo

1. In this moment network graph is not stable