This directory is to demostrate the "container linking" feature of Docker, using wrk (written in C and Lua), Spray (written in Scala), Fluentd (written in Ruby), and Elasticsearch + Kibana (written in Java) as example.

Prepare environments for experiment.

CAUTION: This may take minutes. Do it at a place with good bandwidth.

$ docker pull williamyeh/wrk
$ docker pull williamyeh/spray-httpserver
$ docker pull williamyeh/fluentd
$ docker pull digitalwonderland/elasticsearch

1. Clear all unused container records:

   $ docker rm `docker ps --no-trunc -a -q`
   

2. Start the http server (with the container name spray):

   $ docker run -d --name spray  \
         williamyeh/spray-httpserver
   

3. Tail the logs of the http server, and keep this console open:

   $ docker logs -f spray
   

4. Open another console. Use wrk to benchmark the http server:

   $ docker run --link spray:httpserver \
         williamyeh/wrk  http://httpserver:8080/
   

5. Use wrk to benchmark the http server, with random IPs as input:

   $ docker run --link spray:httpserver  \
         -v `pwd`:/data  \
         williamyeh/wrk  -s wrk-script.lua  http://httpserver:8080/
   

6. Generate Fluentd conf file td-agent.conf for capturing logs produced by the spray container:

   $ ./gen-fluentd-conf.sh  spray  > td-agent.conf
   

7. Use fluentd to collect the spray logs:

   $ docker run -it                 \
        -v /var/lib/docker/containers:/var/lib/docker/containers:ro  \
        -v `pwd`:/etc/td-agent:ro   \
        -v `pwd`:/data              \
        --net=host                  \
        williamyeh/fluentd  start
   

8. Re-run previous wrk task to see the fluentd output!

9. Start the Elasticsearch server:

   $ docker run -d  --name elasticsearch  \
         -p 9200:9200  \
         digitalwonderland/elasticsearch
   

10. Use browser to open the Kibana page: http://localhost:9200/_plugin/kibana3/

11. You may also re-run previous wrk task to see the Kibana dashboard!