07_cgroups.md

Observe CGroups in action

Create VM

Create VM using docker-machine with 2 vCPUs.

Test CGroups

Notes:

• Use top command¹ to observe cpu loads,
• Use stress image for this exercise.

Test cases:

Assigning cpus to containers

• Fully load single CPU core of your choice
  • docker run -it --rm progrium/stress --cpu 1
  • Verify that only one core is under load
• Run container again but load both cpus
  • docker run -it --rm progrium/stress --cpu 2
• Modify CPUs that container can access directly in /sys/fs/cgroup/

Sharing cpu resources between contianers

• Run 2 containers, 1 with cpushare 1024, second with 2048
  • observe how much resources each container consumes
  • docker run -c 1024 -d progrium/stress --cpu 2
  • docker run -c 2048 -d progrium/stress --cpu 2
• Ensure that container won't consume more CPU resources than specified no matter if its the only container on the host
  • See cpushare and cpuquota
• Experiment with multiple containers

Limit memory to container

* Run container (from stress image) with memory limited to `128M`
* Verify that container can't consume more memory than allocated

---

Footnotes

1. set TERM=xterm to enable top in boot2docker image ↩