Limit Memory (RAM) and CPU in Docker Compose

Set the values in Docker Compose

version: '3.7'
services:
  foo:
    image: alpine:3
    deploy:
      resources:
        limits:
          cpus: '0.1'
          memory: 100M

Find the right values

There are many tools to choose from. Here’s what we use:

docker stats

To get current values, simply run

docker stats

This is useful to get current memory and see spikes in CPU usage.

Prometheus Metrics

CPUs [1] correlates with the Prometheus metric container_cpu_user_seconds_total as exposed by cAdvisor, e.g.

sum by (name) (rate(container_cpu_user_seconds_total{image!=""}[$__rate_interval]))

Note the usage of the variable rate_interval (new Grafana 7.2).

Memory can be queried like this:

sum by (name) (container_memory_usage_bytes{image!=""})

Bemerkung

If you want a fully functional Prometheus and Grafana stack, be sure to check out hukudo/prom/

---

[1]

The number of CPUs the container may use. See also https://docs.docker.com/config/containers/resource_constraints/#configure-the-default-cfs-scheduler

Ubuntu Kernel Cleanup   Bind mounts vs. Volumes in Docker Compose