#!/usr/bin/env python3 """ =head1 NAME docker_ - Docker wildcard-plugin to monitor a L host. This wildcard plugin provides at the moment only the suffixes C, C, C, C, C and C. =head1 INSTALLATION - Copy this plugin in your munin plugins directory - Install Python3 "docker" package =over 2 ln -s /usr/share/munin/plugins/docker_ /etc/munin/plugins/docker_containers ln -s /usr/share/munin/plugins/docker_ /etc/munin/plugins/docker_cpu ln -s /usr/share/munin/plugins/docker_ /etc/munin/plugins/docker_images ln -s /usr/share/munin/plugins/docker_ /etc/munin/plugins/docker_memory ln -s /usr/share/munin/plugins/docker_ /etc/munin/plugins/docker_status ln -s /usr/share/munin/plugins/docker_ /etc/munin/plugins/docker_volumes =back After the installation you need to restart your munin-node: =over 2 systemctl restart munin-node =back =head1 CONFIGURATION This plugin need to run as root, you need to create a file named docker placed in the directory /etc/munin/plugin-conf.d/ with the following config (you can also use Docker environment variables here as described in https://docs.docker.com/compose/reference/envvars/): You can use the EXCLUDE_CONTAINER_NAME environment variable to specify a regular expression which if matched will exclude the matching containers from the memory and cpu graphs. For example env.EXCLUDE_CONTAINER_NAME runner Would exclude all containers with the word "runner" in the name. =over 2 [docker_*] user root env.DOCKER_HOST unix://var/run/docker.sock env.EXCLUDE_CONTAINER_NAME regexp =back """ import os import sys import docker import re from multiprocessing import Process, Queue def print_containers_status(client): running = 0 paused = 0 created = 0 restarting = 0 removing = 0 exited = 0 dead = 0 for container in client.containers.list(): if container.status == 'running': running += 1 elif container.status == 'paused': paused += 1 elif container.status == 'created': created += 1 elif container.status == 'restarting': restarting += 1 elif container.status == 'removing': removing += 1 elif container.status == 'exited': exited += 1 elif container.status == 'dead': dead += 1 print('running.value', running) print('paused.value', paused) print('created.value', created) print('restarting.value', restarting) print('removing.value', removing) print('exited.value', exited) print('dead.value', dead) def get_container_stats(container, q): q.put(container.stats(stream=False)) def parallel_container_stats(client): proc_list = [] stats = {} exclude = os.getenv('EXCLUDE_CONTAINER_NAME') for container in client.containers.list(): if exclude and re.search(exclude, container.name): break q = Queue() p = Process(target=get_container_stats, args=(container, q)) proc_list.append({'proc': p, 'queue': q, 'container': container}) p.start() for proc in proc_list: proc['proc'].join() stats[proc['container']] = proc['queue'].get() return stats.items() def print_containers_cpu(client): for container, stats in parallel_container_stats(client): cpu_count = len(stats["cpu_stats"]["cpu_usage"]["percpu_usage"]) cpu_percent = 0.0 cpu_delta = float(stats["cpu_stats"]["cpu_usage"]["total_usage"]) \ - float(stats["precpu_stats"]["cpu_usage"]["total_usage"]) system_delta = float(stats["cpu_stats"]["system_cpu_usage"]) \ - float(stats["precpu_stats"]["system_cpu_usage"]) if system_delta > 0.0: cpu_percent = cpu_delta / system_delta * 100.0 * cpu_count print(container.name + '.value', cpu_percent) def print_containers_memory(client): for container, stats in parallel_container_stats(client): print(container.name + '.value', stats['memory_stats']['stats']['total_rss']) def main(): try: mode = sys.argv[1] except IndexError: mode = "" wildcard = sys.argv[0].split("docker_")[1].split("_")[0] if mode == "suggest": print("containers") print("cpu") print("images") print("memory") print("status") print("volumes") client = docker.from_env() if wildcard == "status": if mode == "config": print("graph_title Docker status") print("graph_vlabel containers") print("graph_category virtualization") print("running.label RUNNING") print("paused.label PAUSED") print("created.label CREATED") print("restarting.label RESTARTING") print("removing.label REMOVING") print("exited.label EXITED") print("dead.label DEAD") else: print_containers_status(client) elif wildcard == "containers": if mode == "config": print("graph_title Docker containers") print("graph_vlabel containers") print("graph_category virtualization") print("containers_quantity.label Containers") else: print('containers_quantity.value', len(client.containers.list())) elif wildcard == "images": if mode == "config": print("graph_title Docker images") print("graph_vlabel images") print("graph_category virtualization") print("images_quantity.label Images") else: print('images_quantity.value', len(client.images.list())) elif wildcard == "volumes": if mode == "config": print("graph_title Docker volumes") print("graph_vlabel volumes") print("graph_category virtualization") print("volumes_quantity.label Volumes") else: print('volumes_quantity.value', len(client.volumes.list())) elif wildcard == "cpu": if mode == "config": graphlimit = str(os.cpu_count() * 100) print("graph_title Docker containers CPU usage") print("graph_args --base 1000 -r --lower-limit 0 --upper-limit " + graphlimit) print("graph_scale no") print("graph_period second") print("graph_vlabel CPU usage (%)") print("graph_category virtualization") print("graph_info This graph shows docker container CPU usage.") for container in client.containers.list(): print("{}.label {}".format(container.name, container.name)) else: print_containers_cpu(client) elif wildcard == "memory": if mode == "config": print("graph_title Docker containers memory usage") print("graph_args --base 1024 -l 0") print("graph_vlabel Bytes") print("graph_category virtualization") print("graph_info This graph shows docker container memory usage.") for container in client.containers.list(): print("{}.label {}".format(container.name, container.name)) else: print_containers_memory(client) if __name__ == '__main__': main()