Platform monitoring

While Deis itself doesn’t have a built-in monitoring system, Deis components and deployed applications alike run entirely within Docker containers. This means that monitoring tools and services which support Docker containers should work with Deis. A few tools and monitoring services which support Docker integrations are detailed below.



Google’s Container Advisor (cadvisor) runs inside a Docker container and shows memory and CPU usage for all containers running on the host. To run cAdvisor:

sudo docker run \
--volume=/:/rootfs:ro \
--volume=/var/run:/var/run:rw \
--volume=/sys:/sys:ro \
--volume=/var/lib/docker/:/var/lib/docker:ro \
--publish=8080:8080 \
--detach=true \
--name=cadvisor \

To run cAdvisor on all hosts in the cluster, you can submit and start a fleet service:

Description=Google Container Advisor

ExecStartPre=/bin/sh -c "docker history google/cadvisor:latest >/dev/null 2>&1 || docker pull google/cadvisor:latest"
ExecStartPre=/bin/sh -c "docker inspect cadvisor >/dev/null 2>&1 && docker rm -f cadvisor || true"
ExecStart=/usr/bin/docker run --volume=/:/rootfs:ro --volume=/var/run:/var/run:rw --volume=/sys:/sys:ro --volume=/var/lib/docker/:/var/lib/docker:ro --publish=8080:8080 --name=cadvisor google/cadvisor:latest
ExecStopPost=-/usr/bin/docker rm -f cadvisor



Save the file as cadvisor.service. Load and start the service with fleetctl load cadvisor.service && fleetctl start cadvisor.service.

The web interface will be accessible at port 8080 on each host.

In addition to starting a cAdvisor instance on each CoreOS host, there’s also a project called heapster from the Google Cloud Platform team, which seems to be a cluster-aware cAdvisor.

Monitoring services

These are a few monitoring services which are known to provide Docker integrations. Additions to this reference guide are much appreciated!


The Datadog cloud monitoring service provides a monitor agent which runs on the host and provides metrics for all Docker containers (which is functionally similar to cAdvisor’s implementation). See this blog post for details. The Datadog agent for Docker can be run on a single host as follows:

docker run -d --privileged --name dd-agent -h `hostname` -v /var/run/docker.sock:/var/run/docker.sock -v /proc/mounts:/host/proc/mounts:ro -v /sys/fs/cgroup/:/host/sys/fs/cgroup:ro -e API_KEY=YOUR_REAL_API_KEY datadog/docker-dd-agent

Be sure to substitute YOUR_REAL_API_KEY for your Datadog API key.

To run Datadog for the entire cluster, you can submit and start a fleet service (again, substitute YOUR_REAL_API_KEY):


ExecStartPre=/bin/sh -c "docker history datadog/docker-dd-agent:latest >/dev/null || docker pull datadog/docker-dd-agent:latest"
ExecStart=/usr/bin/docker run --privileged --name dd-agent -h %H -v /var/run/docker.sock:/var/run/docker.sock -v /proc/mounts:/host/proc/mounts:ro -v /sys/fs/cgroup/:/host/sys/fs/cgroup:ro -e API_KEY=YOUR_REAL_API_KEY datadog/docker-dd-agent



Save the file as datadog.service. Load and start the service with fleetctl load datadog.service && fleetctl start datadog.service.

Shortly thereafter, you should start to see metrics from your Deis cluster appear in your Datadog dashboard.

New Relic

The New Relic monitoring service’s agent will run on the CoreOS host and report metrics to New Relic.

Unlike Datadog, however, the agent running on the host doesn’t send metrics for individual containers unless those containers have been built with a Dockerfile that installs their own instance of the agent.

The Deis community’s own Johannes Würbach has developed a fleet service for New Relic in his newrelic-sysmond repository.

SPM Performance Monitoring

SPM for Docker provides Monitoring, Anomaly Detection and Alerting. Detailed metrics and events are collected for the host machine and running containers (e.g. cpu, memory, IO, network, limits). The SPM agent runs in a Docker container.

Create a new SPM App of type “Docker” and copy the Application Token

Set the Application Token (SPM_TOKEN) via etcd to make it available in the cluster.


SPM Docker Agent can be run using Docker as follows (assuming the SPM_TOKEN is set via etcdctl):

docker run -d --name spm-agent -e SPM_TOKEN=`etcdctl get SPM_TOKEN` -e HOSTNAME=$HOSTNAME -v /var/run/docker.sock:/var/run/docker.sock sematext/spm-agent-docker

To activate SPM Docker Agent for the entire cluster submit this unit file to fleet

Description=SPM Docker Agent

ExecStartPre=-/usr/bin/docker kill spm-agent
ExecStartPre=-/usr/bin/docker rm spm-agent
ExecStartPre=/usr/bin/docker pull sematext/spm-agent-docker:latest
ExecStart=/bin/sh -c 'set -ex; /usr/bin/docker run --name spm-agent -e SPM_TOKEN=$(etcdctl get SPM_TOKEN) -e HOSTNAME=$HOSTNAME -v /var/run/docker.sock:/var/run/docker.sock sematext/spm-agent-docker'
ExecStop=/usr/bin/docker stop spm-agent



Save the file as spm-agent.service.


Load and start the service with

fleetctl load spm-agent.service && fleetctl start spm-agent.service

After one minute, you should see metrics in SPM.

Documentation, source code and support information is available here: