Docker Traffic Control

Docker Traffic Control allows to set a rate limit on the container network and can emulate network conditions like delay, packet loss, duplication, and corrupt for the Docker containers, all that basing only on labels. HTTP API allows to fetch and pause existing rules and to manually overwrite them, command-line interface is also available. Project is written entirely in Bash and is distributed as a Docker image.

Running

First run Docker Traffic Control daemon in Docker. The container needs NET_ADMIN capability and the host network mode to manage network interfaces on the host system, /var/run/docker.sock volume allows to observe Docker events and query container details.

docker run -d \
    --name docker-tc \
    --network host \
    --cap-add NET_ADMIN \
    --restart always \
    -v /var/run/docker.sock:/var/run/docker.sock \
    -v /var/docker-tc:/var/docker-tc \
    lukaszlach/docker-tc

You can also pass HTTP_BIND and HTTP_PORT environment variables, which default to 127.0.0.1:4080.

This repository contains docker-compose.yml file in root directory, you can use it instead of manually running docker run command. Newest version of image will be pulled automatically and the container will run in daemon mode.

git clone https://github.com/lukaszlach/docker-tc.git
cd docker-tc
docker-compose up -d

When using docker-compose.yml configuration is stored in the .env file, also provided in this repository.

Usage

After the daemon is up it scans all running containers and starts listening for container:start events triggered by Docker Engine. When a new container is up (also a new Swarm service) and contains com.docker-tc.enabled label set to 1, Docker Traffic Control starts applying network traffic rules according to the rest of the labels from com.docker-tc namespace it finds.

Docker Traffic Control recognizes the following labels:

• com.docker-tc.enabled - when set to 1 the container network rules will be set automatically, any other value or if the label is not specified - the container will be ignored
• com.docker-tc.limit - bandwidth or rate limit for the container, accepts a floating point number, followed by a unit, or a percentage value of the device's speed (e.g. 70.5%). Following units are recognized:
  • bit, kbit, mbit, gbit, tbit
  • bps, kbps, mbps, gbps, tbps
  • to specify in IEC units, replace the SI prefix (k-, m-, g-, t-) with IEC prefix (ki-, mi-, gi- and ti-) respectively
• com.docker-tc.delay - length of time packets will be delayed, accepts a floating point number followed by an optional unit:
  • s, sec, secs
  • ms, msec, msecs
  • us, usec, usecs or a bare number
• com.docker-tc.loss - percentage loss probability to the packets outgoing from the chosen network interface
• com.docker-tc.duplicate - percentage value of network packets to be duplicated before queueing
• com.docker-tc.corrupt - emulation of random noise introducing an error in a random position for a chosen percent of packets

Read the tc command manual to get detailed information about parameter types and possible values.

Run the busybox container on custom network to create virtual network interface, specify all possible labels and try to ping google.com domain.

docker network create test-net
docker run -it \
	--net test-net \
	--label "com.docker-tc.enabled=1" \
	--label "com.docker-tc.limit=1mbps" \
	--label "com.docker-tc.delay=100ms" \
	--label "com.docker-tc.loss=50%" \
	--label "com.docker-tc.duplicate=50%" \
	--label "com.docker-tc.corrupt=10%" \
	busybox \
	ping google.com

You should see output similar to shown below, ping correctly reports duplicates, packets are delayed and some of them lost.

PING google.com (216.58.215.78): 56 data bytes
64 bytes from 216.58.215.78: seq=0 ttl=54 time=1.010 ms
64 bytes from 216.58.215.78: seq=1 ttl=54 time=101.031 ms
64 bytes from 216.58.215.78: seq=2 ttl=54 time=101.045 ms
64 bytes from 216.58.215.78: seq=3 ttl=54 time=101.011 ms
64 bytes from 216.58.215.78: seq=4 ttl=54 time=101.028 ms
64 bytes from 216.58.215.78: seq=5 ttl=54 time=101.060 ms
64 bytes from 216.58.215.78: seq=5 ttl=54 time=154.685 ms (DUP!)
64 bytes from 216.58.215.78: seq=6 ttl=54 time=101.084 ms
64 bytes from 216.58.215.78: seq=8 ttl=54 time=101.085 ms
64 bytes from 216.58.215.78: seq=8 ttl=54 time=1001.130 ms (DUP!)
64 bytes from 216.58.215.78: seq=11 ttl=54 time=102.218 ms
64 bytes from 216.58.215.78: seq=15 ttl=54 time=114.437 ms
64 bytes from 216.58.215.78: seq=16 ttl=54 time=101.471 ms
64 bytes from 216.58.215.78: seq=17 ttl=54 time=101.068 ms
64 bytes from 216.58.215.78: seq=17 ttl=54 time=1001.162 ms (DUP!)
64 bytes from 216.58.215.78: seq=19 ttl=54 time=101.104 ms
^C
--- google.com ping statistics ---
20 packets transmitted, 13 packets received, 3 duplicates, 35% packet loss
round-trip min/avg/max = 1.010/152.299/1001.162 ms

HTTP API

API available via HTTP allows you to manage network control rules manually on chosen container.

If you have running containers already or do not want to add Docker Traffic Control labels, you can use the POST endpoint to set the rules manually or in an automated process.

HTTP was chosen for local management instead of docker exec so that you can still easily control Docker Traffic Control on Swarm Nodes, utilize any service discovery and manage remotely using any HTTP client.

GET

GET /<container-id|container-name> HTTP/1.1

Get traffic control rules for a single container.

curl localhost:4080/221517ae59d1
curl localhost:4080/my-container-name

LIST

LIST /<container-id|container-name> HTTP/1.1

List all traffic control rules for all running containers.

curl -X LIST localhost:4080

DELETE

DELETE /<container-id|container-name> HTTP/1.1

Delete all container's traffic control rules. Container is also added to the ignore list to prevent further changes whether it has proper labels set or not.

curl -X DELETE localhost:4080/my-container-name

PUT

PUT /<container-id|container-name> HTTP/1.1

Put back the container to the scanning poll and remove it from the ignore list.

curl -X PUT localhost:4080/221517ae59d1

POST

POST /<container-id|container-name> HTTP/1.1
<rate|delay|loss|corrupt|duplicate>=<value>&...

Update container's traffic control rules. Container does not have to have any labels set, it can be any running container. All previous rules for this container are removed.

curl -d'delay=300ms' localhost:4080/my-container-name
curl -d'rate=512kbps' localhost:4080/221517ae59d1
curl -d'rate=1mbps&loss=10%' localhost:4080/my-container-name

Command-line

Set up a command that may be available for all users...

echo 'curl -sSf -X "$1" "localhost:4080/$2?$3"' > /usr/bin/docker-tc
chmod +x /usr/bin/docker-tc

... or set up a local command alias:

alias docker-tc='curl -sSf -X "$1" "localhost:4080/$2?$3"'

Usage

docker-tc get 221517ae59d1
docker-tc list
docker-tc delete my-container-name
docker-tc put 221517ae59d1
docker-tc set my-container-name 'delay=300ms&rate=1000kbps'

Build

git clone https://github.com/lukaszlach/docker-tc.git
cd docker-tc
make
docker images | grep docker-tc

Supported platforms

Docker Traffic Control only works on Linux distributions like Debian, Ubuntu, CentOS or Fedora.

• MacOS - not supported due to lack of host network mode support
• Windows - not supported due to separate network stack between Linux and Windows containers

Deploy on Swarm

Warning: Although this project is prepared for Docker Swarm you will not be able to deploy it as a service because of the moby/moby#25885 issue. This section is theoretical.

Create a global service named docker-tc that will scan all containers and other services that are currently running or will start in the future.

docker service create \
    --name docker-tc \
    --mode global \
    --restart-condition any \
    --network host \
    --cap-add NET_ADMIN \
    --mount "type=bind,src=/var/run/docker.sock,dst=/var/run/docker.sock" \
    --mount "type=bind,src=/var/docker-tc,dst=/var/docker-tc" \
    lukaszlach/docker-tc

/var/docker-tc directory has to exist on Docker Swarm nodes before deploying the service

Contributors

Licence

MIT License

Copyright (c) 2018-2019 Łukasz Lach [email protected]

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.