Setting up Jenkins-CI with the Docker-Plugin

Note:   I’m not using Docker at the moment, so I cannot guarantee that the following works anymore given the volume of updates to Docker and Jenkins.


This post describes how to:

What you’ll need

  • Two hosts running Linux. One for running Jenkins-CI the other for running Docker. I’ll be using Ubuntu for this how to but any distribution should do as long as it meets Jenkins and Docker’s requirements.
  • Internet access. Ubuntu and Docker both need Internet access.

While you can do everything this how to describes on one host, by using two hosts you will get a feel for the master-slave configuration we are creating.

Security Note

This how-tos’ servers are on a secure network, so the hosts’ firewalls are disabled to speed things up. If you are using CentOS, you will want to disable SELinux. In a production environment be sure to verify your environments security requirements before disabling any of these features.

On Ubuntu:

# sudo stop ufw
# sudo ufw disable

On CentOS:

# service iptables stop
# chkconfig iptales off
# setenforce 0
# sed -i 's/=enforcing/=disabled/g' /etc/sysconfig/selinux

Setting up the Docker host

  • Install the OS on what will be the Docker server. Make sure OpenSSH is installed, configured and running.
  • Install Docker by following the instructions in the documentation,
  • Once Docker is running properly, update the /etc/default/docker (CentOS: /etc/sysconfig/docker) file to include the following:

DOCKER_OPTS="-H unix:///var/run/docker.sock -H tcp://"

  • The -H unix:///var/run/docker.sock configures a local unix socket for Docker to use. This is the default.
  • The -H tcp:// option configures Docker to listen on a TCP port, a requirement for Jenkins to talk to Docker.

Restart Docker to enable the changes.

sudo restart docker

Optional:  if you feel confident, add your user to the docker group so you will not have to prefix your docker command with sudo all the time.

sudo usermod -aG docker myuser

You need to logout and login for the change to take effect.

Docker images

Our example uses a Docker image created by the author of the Jenkins Docker Plugin. Download the evarga/jenkins-slave image using the command:

sudo docker pull evarga/jenkins-slave

The time required to download the image depends on your Internet connection speed.

The evarga/jenkins-slave image is a Debian image with a pre-configured jenkins user account.

Setting up the Jenkins server

  • Install the OS on what will be the Jenkins-CI server. Make sure OpenSSH is installed, configured and running.
  • Go to and download the latest Jenkins-CI package. This how-to uses the current release, but the stable version should work.

Sample Jenkins installation process for Ubuntu:

wget -q -O - | sudo apt-key add -
echo 'deb binary/' > /etc/apt/sources.list.d/jenkins.list
sudo apt-get update
sudo apt-get install jenkins
sudo service jenkins start

This will:
– Install the Jenkins-CI package key.
– Configure the Jenkins-CI package repository.
– Install the Jenkins daemon and its dependencies.
– Start the Jenkins service.

Get the IP address (or hostname if DNS is configured) of your Jenkins host and using a browser connect to the Jenkins-CI host on port 8080. Example: If everything worked, you’ll be greeted with a Welcome to Jenkins! page.


Jenkins can take some time to start, so don't be surprised if you cannot connect right away.

By default, Jenkins-CI gives full access to the user. No user name or password is required to login. If you decide to roll-out your own setup in production, be sure to look into the various authentication and authorization options.

Installing and configuring the Jenkins Docker plugin

Installing the Docker plugin

  • Using your browser, connect to your Jenkins server.
  • From the main menu, select Manage Jenkins then select Manage Plugins.
  • If there are any updates available, install them before continuing.
  • Click on the Available tab.
  • In the Filter field at the top of the page, type in Docker. The list of
    matching plugins will update automatically.
  • Select Docker plugin from the list.
  • Click the Download now and install after restart button at the bottom of
    the page.
  • You must now restart Jenkins manually (service jenkins restart) or check off
    the Restart Jenkins when installation is complete and no jobs are running box
    which will restart Jenkins-CI for you.

Configuring the Docker plugin

Configuring the Docker plugin is a two-step process. The first step is configuring the plugin to use our Docker server. The second step is configuring the list of Docker images available to Jenkins-CI on that server.

Adding the Docker server (aka, the cloud)

I’m only describing the fields required to configure the plugin. Each field’s help is accessed by clicking the field’s help icon (?).

  • From the Manage Jenkins menu, select Configure System.
  • Scroll down to the Cloud section. (Usually at the bottom on the page.)
  • Click the Add a new cloud drop down and select Docker. A Docker sub-section is added to the Cloud section.
  • Now fill in the Docker cloud section.
  • Name: Enter a name for the docker cloud. This is purely a label and can be anything.
  • Docker URL: Enter the URL of our Docker server including the port number. This is the IP or hostname of the Docker server we created earlier. Example:
  • Credentials: We didn’t configure our Docker server to use credentials so leave it set to none.
  • Container Cap: The number of Docker containers we can run at any one time on the Docker server.
  • Click the Test Connection to test the configuration. The Docker version running on the server is returned if everything is correct.

The server is now configured and we can move on to configuring the images to use.

Configuring the Docker images to use

I’m only describing the fields required. Each field’s help can be accessed by clicking the field’s help icon (?).

  • In Cloud’s Docker sub-section, click the Add Docker Template drop down and select Docker Template. An empty Docker Template gets added to the section.
  • Fill in the Docker Template.
  • Docker Image: This is the name of the docker image to use. Our example uses the image we downloaded earlier,
  • Remote Filing System Root: This is the path to the Jenkins home directory inside the Docker image. If you use another Docker image, you probably need to adjust this value.
  • Labels: This is the label to assign to this image. This is the label that appears in other parts of Jenkins, including restrictions. I suggest a label based on the image name. Our example uses evarga-jenkins-slave.
  • Launch method: This is how Jenkins will log in to the running Docker image. By default there is only the SSH option which is what we’ll use.
  • Next to the empty Credentials drop down, click the Add button to open the Add credentials dialog. The username and password are already configured in the Docker image and have been provided by the image’s author.
  • Kind: Username with password
  • Username: jenkins
  • Password: jenkins
  • Description: Enter a description of the credentials. For example: evarga/jenkins-slave ssh credentials.
  • Click Add and the credentials will be added to Jenkins. Because this is the first set of credentials we created they will automatically be selected.
  • Click the Save button at the bottom of the page.

That’s it. The Docker plugin is now ready for use.

A quick example

Here is a quick example of how to use the docker plugin in a job.

  • Click New Item.
  • Enter a name for the job.
  • Select the Freestyle project option.
  • Click OK. You’ll be taken to the project’s configuration page.
  • Enable the Docker Container option.
  • Make sure Restrict where this project can be run is selected.
  • In the Label Expression field, enter evarga-jenkins-slave. Jenkins attempts to auto-complete the label from the list of labels previously defined. Jenkins will display an error if it cannot find the label you enter.
  • Click Add build step and select Execute shell.
  • We’ll use a simple script. Copy and paste the code into the Command area.
echo "hello from docker!"
exit $?
  • Click Save.
  • Click the Build Now link.

Unlike a locally run job, the Docker slave jobs can take a minute or two to
start up.

If everything worked, the output from the job should look like the following:

Started by user anonymous
Building remotely on our-docker-cloud-451be36e99da (evarga-jenkins-slave) in workspace /home/jenkins/workspace/testjob
[testjob] $ /bin/bash /tmp/
SSH_CLIENT= 60438 22
NODE_LABELS=evarga-jenkins-slave our-docker-cloud-451be36e99da
hello from docker!
Finished: SUCCESS

Congratulations! You know have a working Jenkins-CI setup using Docker for SSH build slaves.

Fixing Smokeping’s “Opening secrets file /etc/smokeping/smokeping_secrets: Permission denied” warning

I recently decided to start playing with smokeping, in particular a master/slave setup.

I started by keeping it simple, two Ubuntu 14.04 virtual machines on the same network. I decided to use Ubuntu because it already has smokeping packages and more importantly I had the ISO image handy.

I started by reading the documentation on the Smokeping home page and dove in.  Setting up the master went smoothly.   OK, technically there was nothing to do but install the smokeping package, but hey, baby steps.

Next came setting up the slave.

I installed smokeping on the second virtual machine that was to act as the slave.  I updated the slave’s default/smokeping file and created the secret file on the slave. I then updated the target entries on the master to expect the slave.

When I started the smokeping on the slave I was quickly presented with the following errors:

WARNING: Opening secrets file /etc/smokeping/smokeping_secrets: Permission denied

ERROR: we did not get config from the master. Maybe we are not configured as a slave for any of the targets on the master ?

I must have wasted two hours on this.   I kept looking at the configuration files on the slave for problems, after all it was reporting the error.

I ended up running tcpdump on the slave to see exactly what the master and slave were saying to each other.
Here is what I saw:


POST /smokeping/smokeping.cgi HTTP/1.1
TE: deflate,gzip;q=0.3
Connection: TE, close
User-Agent: smokeping-slave/1.0
Content-Length: 366
Content-Type: multipart/form-data; boundary=xYzZY

Content-Disposition: form-data; name=”slave”
Content-Disposition: form-data; name=”key”

Content-Disposition: form-data; name=”protocol”

Content-Disposition: form-data; name=”data”

Content-Disposition: form-data; name=”config_time”

HTTP/1.1 200 OK
Date: Tue, 14 Apr 2015 17:38:07 GMT
Server: Apache/2.4.7 (Ubuntu)
Vary: Accept-Encoding
Connection: close
Transfer-Encoding: chunked
Content-Type: text/plain

WARNING: Opening secrets file /etc/smokeping/smokeping_secrets: Permission denied


So, it wasn’t the slave that could not open file, it was apache on the master!  Why couldn’t the error message say that?!

A quick chgrp www-data /etc/smokeping/smokeping_secrets fixed the problem right away.

Now the slave reports “Sent data to Server and got new config in response.” when it starts.  Yea!

Find’s “or” option

I work with a group of web servers that regularly need to have their /tmp directories cleaned up.  At first we tried the following find(1) command:

/usr/bin/find /tmp -maxdepth 1 -type f -iname '*.jpg' -or -iname '*.gif' -or -iname '*.bmp' -or -iname '*.png' -or -iname '*.pdf' -or -iname '*.bmp' -print0 -mmin +1440 | xargs -r -0 rm -rf

While it looked right, no files were being removed.  Going back to find(1)s man page I noticed a key word, expression.

So what is a find(1) expression?  A find(1) expression is made up of two parts, what your looking for and what you want to do with it. “-iname ‘*.bmp'” only matched half of the requirements.

Adding the “what to do” part to each expression, the find command now looks like:

/usr/bin/find /tmp -maxdepth 1 -type f -iname '*.jpg' -print0 -mmin +1440 -o -iname '*.gif' -print0 -mmin +1440 -o -iname '*.bmp' -print0 -mmin +1440 -o -iname '*.png' -print0 -mmin +1440 -o -iname '*.pdf' -print0 -mmin +1440 | xargs -r -0 rm -rf

Now find is listing the files which are being fed into xargs/rm for deletion.

Configuring udev to not link MAC addresses to network cards

At work I have been making more and more use of virtualization.   To make life simple, I have a virtual machine image that I use as a template and clone it when I need a new virtual machine.   This works great, except that as part of the cloning process, the clone receives a new MAC address (which is expected).  The problem is,  udev has already linked eth0 to the templates MAC address, so the clone’s network card is configured as eth1, not eth0.  Not a big deal, but annoying (to me anyways).

Here is the workaround I found that tells udev to ignore KVM/Xen/Vmware issued MAC addresses.

In the file /lib/udev/rules.d/75-persistent-net-generator.rules, find the line:

Below that line add the following lines:
# Ignore KVM virtual interfaces
ENV{MATCHADDR}=="52:54:00:*", GOTO="persistent_net_generator_end"
ENV{MATCHADDR}=="54:52:00:*", GOTO="persistent_net_generator_end"
# This eems to be the range used by Xen, but also by virt-clone
ENV{MATCHADDR}=="00:16:36:*", GOTO="persistent_net_generator_end"
# Ignore VMware network interfaces identified by vendor part of MAC addr
ENV{MATCHADDR}=="00:0c:29:*", GOTO="persistent_net_generator_end"

Now that I’ve added these rules to my master image, all the copies I make have their NIC assigned to eth0.


  • Tested on Ubuntu Server 10.04 LTS, though it should work on any distro that uses udev in the same manner.
  • I’ve only used this trick on VMs with a single NIC.  I’m not sure what will happen on multihomed installs.