All components needed to run Zipkin in Docker containers (Cassandra, Collector, Query, Web, and FB-Scribe) We build a Zipkin base container installing scala and zipkin. From this docker image, we derive the specific zipkin parts to combine a zipkin infrastructure. Directorey fb-scribe will generate a facebook scribe service ready to communicat with zipkin collector. The same functionallity can be obtained by using a Scala Akka Tracing mechanism.
The ports and how to relate them is described in file Ports.md.
As cassandra uses many ports a possible AWS security group is described by DATASTAX.
Building the scribe container takes between 10 and 15 minutes on a 2GHz CPU depending on the memory available.
The resulting image is about 800MB. We had to start an AMI of type t2 to not run out of memory.
Two scripts build and deploy zipkin containers. To build them freshly just call ./build.sh -f and to pull them from registry call ./build.sh -r.
To deploy and run zipkin container on one host call ./deploy.sh -r and to cleanup old containers call ./deploy.sh -c.
######TODO
- Check for container link between web and query?
- Configure linking between fb-scribe (or akka) and collector (
COLLECTOR_PORT) running on different docker hosts.
Docker-Zipkin starts the services in their own container: zipkin-cassandra, zipkin-collector, zipkin-query, zipkin-web and only link required dependencies together.
The started Zipkin instance would be backed by a single node Cassandra. By default, the collector port is not mapped to public. You will need to link containers that you wish to trace with zipkin-collector or you may change the respective line in deploy.sh to map the port.
All images with the exception of zipkin-cassandra are sharing a base image: base. zipkin-base and zipkin-cassandra are built on ubuntu:12.04.
Once the containers are running you can connect to the collector on port 9410 via akka-tracing or other libraries that support Zipkin tracing. https://github.com/levkhomich/akka-tracing
We try to place as many installation commands into Docker files, to ease the build process:
- Each build step is frozen as an intermediate container
- Errors during build don`t require rerunning it completely
- Testing new install instructions start from a partly build process and thus run faster
- base docker container with scala and zipkin installed
- cassandra docker container with cassandra installed
- collector docker container with zipkin collector installed
- query docker container with zipkin query installed
- web docker container with zipkin web
- script contains utilites to manage (build, push, pull, start, deploy ) the zipkin container.
This repo is cloned from https://github.com/lispmeister/docker-zipkin.git Zipkin base installation is describe in the twitter zipkin repo and further configuration in detail in Zipkin, from Twitter.
Michael Klöckner mkl@im7.de
Markus Fix lispmeister@gmail.com