How I built Elastic Stack for Docker Swarm using Docker Application Packages(docker-app)

Table of Contents

Let’s begin with Problem Statement !

DockerHub is a cloud-based registry service which allows you to link to code repositories, build your images, test them, store manually pushed images so you can deploy images to your hosts. It provides a centralized resource for container image discovery, distribution and change management as well as workflow automation throughout the development pipeline. We share Docker images all the time, but let’s agree to the fact that we don’t have a good way of sharing the multi-service applications that use them.

Let us take an example of Elastic Stack. Built on an open source foundation, the Elastic Stack lets you reliably and securely take data from any source, in any format, and search, analyze, and visualize it in real time with the help of Elasticsearch, Logstash, Kibana and multiple other tools and technique. In order to build these tools in the form of containers, one need to start building Docker Image for each of these tools. The recommended way is constructing a Dockerfile for each of these tools. In turn, Docker Compose uses these images to build required services. Whenever the docker stack deploy CLI is used to deploy the application stack, these Docker images are pulled from Dockerhub for the first time and then picked up locally once downloaded to your system. What if you could upload your whole application stack to DockerHub? Yes, it’s possible today and docker-app is the tool which can make Compose-based applications shareable on Docker Hub and DTR.

Docker-app v0.5.0 is now Available !

Docker Application Package v0.5.0 is the latest offering from Docker, Inc. You can download it from this link. The binaries are available for Linux, Windows and MacOS Platform. If you are looking out for source code, this is the direct link.

The docker-app v0.5.0 comes with notable features and improvements which are listed below:

The improved docker-app inspect command to shows a summary of services, networks, volumes and secrets.

The docker-app push CLI now works on Windows and bypasses the local docker daemon by talking directly to the registry.
The docker-app save and docker-app ls have been obsoleted.
All commands now accept an application package as a URL.
The docker-app push command now accepts a custom repository name.
The docker-app completion command can generate zsh completion in addition to bash.

In my last blog post, I talked about docker-app for the first time and showcased its usage soon after I returned back from Dockercon. Under this post, I will show how I built Elastic Stack using docker-app for 5-Node Docker Swarm cluster.

Prerequisite:

Open https://labs.play-with-docker.com on your browser

Click on Icon near to Instance to choose 3 Managers & 2 Worker Nodes

Deploy 5 Node Docker Swarm Cluster

$ docker node ls
ID                            HOSTNAME            STATUS              AVAILABILITY        MANAGER STATUSENGINE VERSION
iy9mbeduxd4mmjxoikbn5ulds *   manager1            Ready               Active              Reachable18.03.1-ce
mx916kgqg6gfgqdr2gn1eksxy     manager2            Ready               Active              Leader18.03.1-ce
xaeq943o84g9spy6mebj64tw3     manager3            Ready               Active              Reachable18.03.1-ce
8umdv6m82nrpevuris1e45wnq     worker1             Ready               Active18.03.1-ce
o3yobqgg7wjvjw2ec5ythszgw     worker2             Ready               Active18.03.1-ce

Cloning the Repository

$ git clone https://github.com/ajeetraina/app
Cloning into 'app'...remote: Enumerating objects: 134, done.
remote: Counting objects: 100% (134/134), done.remote: Compressing objects: 100% (134/134), done.
remote: Total 14511 (delta 95), reused 0 (delta 0), pack-reused 14377Receiving objects: 100% (14511/14511), 17.37 MiB | 13.35 MiB/s, done.
Resolving deltas: 100% (5391/5391), done.

Install Docker-app

$ cd app/examples/elk/
[manager1] (local) root@192.168.0.30 ~/app/examples/elk$ ls
README.md          devel              elk.dockerapp      install-dockerapp  prod
[manager1] (local) root@192.168.0.30 ~/app/examples/elk
$ chmod +x install-dockerapp[manager1] (local) root@192.168.0.30 ~/app/examples/elk
$ sh install-dockerappConnecting to github.com (192.30.253.112:443)
Connecting to github-production-release-asset-2e65be.s3.amazonaws.com (52.216.131.187:443)docker-app-linux.tar 100% |*************************************************************|  8895k  0:00:00 ETA
[manager1] (local) root@192.168.0.30 ~/app/examples/elk

Verifying Docker-app Version

$ docker-app version
Version:      v0.4.0
Git commit:   525d93bc
Built:        Tue Aug 21 13:02:46 2018
OS/Arch:      linux/amd64
Experimental: off
Renderers:    none

I assume you have a docker compose file for ELK stack application already available with you. If not, you can download a sample file from this link. Place this YAML file under the same directory(app/examples/elk/). Now with docker-app installed, let’s create an Application Package based on this Compose file:

$ docker-app init elk

Once you run the above command, it create a new directory elk.dockerapp/ that contains three different YAML files:

docker-compose.yml  elk.dockerapp
[manager1] (local) root@192.168.0.30 ~/myelk
$ tree elk.dockerapp/
elk.dockerapp/
├── docker-compose.yml
├── metadata.yml
└── settings.yml

0 directories, 3 files

Edit each of these files as shown to look similar to what are placed under this link.

Rendering Docker Compose file

$ docker-app render elk
version: "3.4"
services:
  elasticsearch:    command:
    - elasticsearch    - -Enetwork.host=0.0.0.0
    - -Ediscovery.zen.ping.unicast.hosts=elasticsearch
    deploy:
      mode: replicated
      replicas: 2
    environment:
      ES_JAVA_OPTS: -Xms2g -Xmx2g
    image: elasticsearch:5
    networks:
      elk: null
    volumes:
    - type: volume
      target: /usr/share/elasticsearch/data
  kibana:
    deploy:
      mode: replicated
      replicas: 2
    environment:
      ELASTICSEARCH_URL: http://elasticsearch:9200
    healthcheck:
      test:
      - CMD-SHELL
      - wget -qO- http://localhost:5601 > /dev/null
      interval: 30s
      retries: 3
    image: kibana:latest
    networks:
      elk: null
    ports:
    - mode: ingress
      target: 5601
      published: 5601
      protocol: tcp
  logstash:
    command:
    - sh
    - -c
    - logstash -e 'input { syslog  { type => syslog port => 10514   } gelf { } } output
      { stdout { codec => rubydebug } elasticsearch { hosts => [ "elasticsearch" ]
      } }'
    deploy:
      mode: replicated
      replicas: 2
    hostname: logstash
    image: logstash:alpine
    networks:
      elk: null
    ports:
    - mode: ingress
      target: 10514
      published: 10514
      protocol: tcp
    - mode: ingress
      target: 10514
      published: 10514
      protocol: udp
    - mode: ingress
      target: 12201
      published: 12201
      protocol: udp
networks:
  elk: {

Setting the kernel parameter for ELK stack

sysctl -w vm.max_map_count=262144

Deploying the Application Stack


[manager1] (local) root@192.168.0.30 ~/app/examples/elk
$ docker-app deploy elk --settings-files elk.dockerapp/settings.yml
Creating network elk_elk
Creating service elk_kibana
Creating service elk_logstash
Creating service elk_elasticsearch
[manager1] (local) root@192.168.0.30 ~/app/examples/elk
$

Inspecting ELK Stack

[manager1] (local) root@192.168.0.30 ~/app/examples/elk
$ docker-app inspect elk
myelk 0.1.0
Maintained by: Ajeet_Raina <ajeetraina@gmail.com>

ELK using Dockerapp

Setting                       Default
-------                       -------
elasticsearch.deploy.mode     replicated
elasticsearch.deploy.replicas 2
elasticsearch.image.name      elasticsearch:5
kibana.deploy.mode            replicated
kibana.deploy.replicas        2
kibana.image.name             kibana:latest
kibana.port                   5601
logstash.deploy.mode          replicated

Verifying Stack services are up & running

[manager1] (local) root@192.168.0.30 ~/app/examples/elk/docker101/play-with-docker/visualizer
$ docker service ls
ID                  NAME                MODE                REPLICAS            IMAGE               PORTS
uk2whax6f3jq        elk_elasticsearch   replicated          2/2                 elasticsearch:5
nm4p3yswvh5y        elk_kibana          replicated          2/2                 kibana:latest       *:5601->56
01/tcp
g5ubng6rhcyp        elk_logstash        replicated          2/2                 logstash:alpine     *:10514->1
0514/tcp, *:10514->10514/udp, *:12201->12201/udp
[manager1] (local) root@192

Pushing the App Package to Dockerhub

Password:[manager1] (local) root@192.168.0.30 ~/app/examples/elk
$ docker login
Login with your Docker ID to push and pull images from Docker Hub. If you don't have a Docker ID, head over to
 https://hub.docker.com to create one.
Username: ajeetraina
Password:Login Succeeded

Pushing the App package to DockerHub


[manager1] (local) root@192.168.0.30 ~/app/examples/elk$ docker-app push --namespace ajeetraina --tag 1.0.2
The push refers to repository [docker.io/ajeetraina/elk.dockerapp]
15e73d68a400: Pushed
1.0.2: digest: sha256:c5a8e3b7e2c7a5566a3e4247f8171516033e7e9791dfdb6ebe622d3830884d9b size: 524
[manager1] (local) root@192.168.0.30 ~/app/examples/elk
$

Important Note: If you are using Docker-app v0.5.0, you might face issue related to pulling the image from Dockerhub as it report unsupported OS error message. Here’s a link to this open issue.

Testing the Application Package

Open up a new PWD window. Install docker-app as shown above and try to run the below command:

docker-app deploy ajeetraina/elk.dockerapp:1.0.2

This should bring up your complete Elastic Stack Platform.

Did you find this blog helpful? Feel free to share your experience. Get in touch with me at twitter @ajeetsraina.

If you want to keep track of latest Docker related information, follow me at https://www.linkedin.com/in/ajeetsraina/.