At a high level, the process for creating a custom elastic image consists of taking one of the existing Amazon Machine Images (AMIs) available on Amazon EC2, starting an instance of the AMI, customising the instance and then creating an image from the customised instance. This image can then be used as an elastic image in your Bamboo installation.

Before you begin:

• This is not a trivial procedure. Instead of creating a custom elastic image, you may want to consider customising an existing Bamboo image by using Amazon's Elastic Block Store (EBS), as described in Configuring Elastic Instances to use the EBS. This is a much simpler option. If you are having problems, please don't hesitate to contact us for further help.
• Please note, Atlassian does not support custom elastic images. Consider customising the elastic agents started from your existing image instead.
• These instructions are written for Unix/Linux operating systems.
• A number of the EC2 commands in the steps below can be completed using the AWS console rather than command line tools (e.g. registering an image). However, we recommend that you follow the steps below unless you are an experienced AWS user.

Before you begin, you need to ensure that you have set up the following:

• Amazon Web Services (AWS) account with EC2 — if you are already using Elastic Bamboo, you should already have an AWS account with EC2 set up. If not, please read Getting Started with Elastic Bamboo.

• Amazon EC2 API Tools — you must install the EC2 API tools on your local machine, otherwise you will not be able to start and access your AMI instance. Note: you need Java Runtime Environment to run these tools. You can install the EC2 API tools by executing the following commands:

  wget http://s3.amazonaws.com/ec2-downloads/ec2-api-tools.zip unzip ec2-api-tools.zip

• Environment Variables— you must set up the following environment variables on your local machine before creating a custom elastic image:
  • EC2_HOME — set this to the path to the installed EC2 API Tools
  • EC2_CERT — set this to the path to the certificate assigned to EC2 account
  • EC2_PRIVATE_KEY — set this to the path to the private key assigned to your AWS account

• Registered Key Pair— you need a registered EC2 key pair, which consists of a private key file and certificate file, to use the EC2 API tools with your AMI instance. If you have previously generated and registered an EC2 key pair (e.g. to use the EC2 API tools), you can re-use it. If you need to generate a new key pair, you can use the following command to do so:

  ec2-add-keypair <key_pair_name>

  The content of the private key will be displayed in the command-line output on your console. Save this content in a file, starting with the line:

  "--BEGIN RSA PRIVATE KEY--"

  and ending with the line:

  "--END RSA PRIVATE KEY--"

  This private key file will be used to access your AMI instance. Set up the appropriate permissions on the private key file by executing the following command.

  chmod 600 <private_key_file>

We strongly recommend that you select an existing Linux/UNIX AMI to customise, rather than starting with a blank AMI. You can get the list of available AMIs by executing the following command:

ec2-describe-images -a

Please check whether you want to launch 32-bit or 64-bit instances from your custom elastic image before selecting an existing AMI. Elastic Bamboo creates "High-CPU Medium" instances (32-bit) by default.

We recommend the following existing Linux/UNIX AMIs for customisation (in order of preference):

• Atlassian's AMI — this AMI is the default image provided by Atlassian. It is an Amazon Linux image updated and prepared for Bamboo, i.e. you will not have to install any Bamboo prerequisites.
• Amazon's AMI — this is the base CentOS-based image provided by Amazon. It does not have any Bamboo prerequisites installed. Typically, you will be better off using the Atlassian AMI.
• Canonical's AMI — this is an Ubuntu image provided by Canonical. It does not have any Bamboo prerequisites installed. You can find AMI ids for Ubuntu images on the Canonical's site.

The names of Atlassian's and Amazon's AMIs (and hence, their IDs) may change with each release of Bamboo, including both major and minor (or point) releases. To quickly access Atlassian's AMI ID for a currently-running version of Bamboo, open that Bamboo site in a web browser and access its 'Image Configurations' page (see Managing your Elastic Image Configurations for more information). The AMI IDs of Atlassian's AMIs are labelled with "(stock image)".

If you want to find out the AMI IDs for a version of Bamboo you don't have running or you're starting an image from scratch and you need the image baseline, the procedure is more complicated:

1. Open the following URL: https://maven.atlassian.com/content/repositories/atlassian-public/com/atlassian/bamboo/atlassian-bamboo/ in a web browser.
2. On the resulting directory page, click the link that represents the version of Bamboo you are currently running. For example, if you are running Bamboo 3.4.4, click on the 3.4.4 link. Another directory page opens, listing a .pom and some additional checksum files.
   Do not click on a version number link that contains 'mX', 'rcX' or 'betaX' (where 'X' is a number), since these relate to publicly available developmental releases of Bamboo.
3. Open the atlassian-bamboo-x.x.x.pom file (where x.x.xis your version of Bamboo). The image version/baseline is stored in elastic-image.version tag. For example, for version 3.4.4, the baseline was 1.7 .
4. Open the following URL: https://maven.atlassian.com/content/repositories/atlassian-public/com/atlassian/bamboo/atlassian-bamboo-elastic-image/ in a web browser.
5. Click on the image baseline version you found in the elastic-image.version tag.
6. On the resulting directory page, the file with ami extension contains all stock image AMI ids.

After you have selected an existing AMI to customise, the next step is to start an instance of the AMI.

3.1 Starting an instance of Atlassian's default AMI

If you chose to customise Atlassian's default AMI, you must start the instance from the admin section of Bamboo. See Starting an Elastic Instance.

Note that Atlassian's default AMI cannot be started using the command line ec2 tools. This is because, on start up, the Bamboo agent on Atlassian's AMI checks to see if it was started from a Bamboo server, and immediately shuts itself down if it was not.

Once started, see Accessing an Elastic Instance for details on how to access the running instance.

3.2 Starting an instance from the command line

Use the ec2-run-instances command to start your instance, as follows:

ec2-run-instances <image_name> -k <key_pair_name>

where <image_name> is the name of the AMI selected in the previous step and <key_pair_name> is the name of the registered key pair generated in '1. Requirements' The public certificate of this key will be injected into your instance.

For example, if you wanted to start an instance of image ami-e55bbd8c using key pair my-keypair, you would run the following command:

ec2-run-instances ami-e55bbd8c -k my-keypair

This command would produce the following command-line output:

INSTANCE        i-25b86743   ami-e55bbd8c     running   my-keypair

i-25b86743 is the name of the new instance in the above example. You should note down the name of your new instance, as you will need that to access your instance in the next step.

3.3 Starting an instance from Bamboo

You can also start a fresh, uncustomised image from Bamboo and begin customisation. The drawback of this approach is that you have only 40 minutes before Bamboo shuts down your instance. The advantage is that you can customise the agent in a single step (as opposed to having to customise/create image/start from Bamboo/save image again).

Once your instance is running, you will need to obtain the address of the instance so you can access it. To do this, use the following command:

ec2-describe-instances <instance_name>

For example, if you wanted to find the address of instance i-25b86743, you would enter:

ec2-describe-instances i-25b86743

This command would produce the following command-line output similar to this:

RESERVATION	r-790f7210	121852097033	default
INSTANCE	i-25b86743	ami-e55bbd8c 	ec2-174-129-94-241.compute-1.amazonaws.com
domU-12-31-39-04-38-87.compute-1.internal	running	elasticbamboo	0		m1.small
2009-06-24T12:36:20+0000	us-east-1c	aki-a71cf9ce	ari-a51cf9cc
monitoring-disabled

The address for the instance in the above example is ec2-174-129-94-241.compute-1.amazonaws.com

You can use the address to access the instance via SSH. See Accessing an Elastic Instance for instructions. If you are using the example command text from that document, you will need to adjust it as follows:

• replace /opt/bamboo/home/xml-data/configuration/elasticbamboo.pk in the example command text with the private key file you generated in '1. Requirements'.
• replace ec2-68-111-185-197.compute-1.amazonaws.com in the example command text with the address of your instance.

5.1 Installing Bamboo Prerequisite Packages

If you selected Atlassian's AMI as your base AMI in '2. Selecting an Existing AMI', you can skip this step and go to '5.2 Adding Customisations' as this imagehas been pre-configured for Bamboo. If you have selected a different AMI, you will need to install the following packages onto your instance using the commands shown below:

1. Amazon EC2 API tools

   wget http://s3.amazonaws.com/ec2-downloads/ec2-api-tools.zip unzip ec2-api-tools.zip mv ec2-api-tools-* /opt/ec2-api-tools

   Note: if your distribution already contains ec2-api-tools package, you can install it instead.

2. Java JRE
   

   You need to install JRE (or JDK) on your instance to be able to run the agent. The preferred way of doing this is to install a package that came with your distribution. For a list of supported JREs, see supported platforms .
   

    

5.2 Adding User Customisations to your Instance

Adding your own customisations is quite a simple process, once you have made it this far.

To add user customisations to your instance:

1. Log into your elastic instance (as previously described in '4. Accessing your Instance').
2. Once you have logged into your elastic instance, you can treat it as a standalone machine and install anything you want. For example, if you want to install Tomcat on an Ubuntu instance you would run 'sudo apt-get install tomcat6', configure it, ensure that your startup scripts are in place, etc, just as you would when installing Tomcat on a standalone machine.
   Please note however, you cannot customise the operating system of a running instance. If you want to create an instance with a customised operating system (e.g. Ubuntu), you will need to select an AMI with that operating system installed (as previously described in '2. Selecting an Existing AMI').
3. Everything that you install will be saved in snapshot image created at the end of these instructions (see '6. Creating an Image of your Customised Instance'). Any instances started from this image will have all of your user customisations automatically installed.

5.3 Deploying Bamboo onto your Instance

Once you have installed the Bamboo pre-requisites on you instance and added your customisations, you can deploy Bamboo elastic bootstrap files onto your instance.

5.3.1

First, you need to create a 'bamboo' user on your instance by running the following command:

useradd -m bamboo

5.3.2

Then, install Bamboo Agent binaries as described below. In this case we're using image version 1.8, you can find out which version you need by following the steps in section "Selecting an Existing AMI". Bamboo 3.4 is a special case, you should use image version 1.8 with it.

imageVer=1.8
wget https://maven.atlassian.com/content/repositories/atlassian-public/com/atlassian/bamboo/atlassian-bamboo-elastic-image/${imageVer}/atlassian-bamboo-elastic-image-${imageVer}.zip
mkdir -p /opt/bamboo-elastic-agent
unzip atlassian-bamboo-elastic-image-${imageVer}.zip -d /opt/bamboo-elastic-agent

5.4 Instance Configuration

At this stage, you should have a customised instance with Bamboo deployed onto it. The last step in creating a customised instance is setting up an EC2 environment on your instance. Carry out the following steps to set this up:

1. Run the following command on your instance to set permissions on the bamboo user directory:

   chown -R bamboo:bamboo /home/bamboo/

2. Configure path variables
   Create a file profile.sh in your instance's /mnt directory. This file contains the default Elastic Bamboo path configuration settings, as seen below:

   export JAVA_HOME=<path to JRE used to start the agent> export EC2_HOME=<location of your EC2 tools installation> export EC2_PRIVATE_KEY=/root/pk.pem export EC2_CERT=/root/cert.pem export PATH=/opt/bamboo-elastic-agent/bin:$EC2_HOME/bin:$JAVA_HOME/bin:$M2_HOME/bin:$MAVEN_HOME/bin:$ANT_HOME/bin:$PATH

   If all of the tools on this page were installed in recommended locations, no changes are required. Otherwise, you can update the file as required.
   Once profile.sh is correctly customised for your instance, you need to copy it to the /etc/profile.d directory by running the following command on your instance in the /mnt directory:

   mv profile.sh /etc/profile.d/bamboo.sh

3. Configure automatic startup of the Bamboo agent
   You will need to configure your instance to start up the Bamboo agent automatically when the instance is started. You can do this by appending the rc.local file to the one that already exists on your instance, by running the following command on your instance in the mnt directory:

   cat /opt/bamboo-elastic-agent/etc/rc.local >> /etc/rc.d/rc.local

4. Final settings and cleanup
   Finally, create a Bamboo welcome screen and clean up keys on your instance by running the following command:

   cp /opt/bamboo-elastic-agent/etc/motd /etc/motd echo bamboo-<x.x.x> >> /etc/motd rm -f /root/firstlogin /etc/ssh/ssh_host_da_key /etc/ssh/ssh_host_dsa_key.pub /etc/ssh/ssh_host_key /etc/ssh/ssh_host_key.pub /etc/ssh/ssh_host_rsa_key /etc/ssh/ssh_host_rsa_key.pub /root/.ssh/authorized_keys touch /root/firstrun

   where <x.x.x> is the Bamboo version you are running (e.g. 3.4.4).

5. Now, follow the instructions from section "Creating an image of your Customised Instance" to create an AMI.
6. Start the image from Bamboo. The agent should come up and download all necessary data to the EC2 instance.
   
7. Run /opt/bamboo-elastic-agent/bin/prepareInstanceForSaving.sh .
8. Now, follow the instructions from section "Creating an image of your Customised Instance" to create an AMI. That's it, the newly created AMI contains everything you need to start Bamboo Agents.

Note: if you started your instance from Bamboo right at the start, instead of steps 5 & 6, you can just run:

su -c bamboo-elastic-agent - bamboo

The final step is to create an image from your customised instance. To do this, you will require the following information:

• Amazon Account Number
• Access Key ID
• Secret Access Key
• Amazon S3 bucket name that will be used to store image (if you don't have access to Amazon S3, you can sign up on this page.)

1. Transfer Amazon private key file and certificate to your instance
   Transfer the key files to your instance by running these commands on your local machine:

   scp -i <private_key_file> $EC2_PRIVATE_KEY root@<instance_address>:/mnt scp -i <private_key_file> $EC2_CERT root@<instance_address>:/mnt

   where <private_key_file> is the private key file from your local machine created in step 'Registered Key Pair' of 1. Requirements and the <instance_address> is the address of your instance from '4. Accessing your Instance'.

2. Set up EC2_HOME and JAVA_HOME environment variables
   Set up these environment variables by running the following commands on your instance:

   export EC2_HOME=<location of your EC2 tools installation> export EC2_PRIVATE_KEY=/mnt/<ec2_private_key_file> export EC2_CERT=/mnt/<ec2_certificate_file> export JAVA_HOME=<path to JRE used to start the agent>

3. You can create an image of your customised instance by using the ec2-bundle-vol command, as follows:
   
   

   ec2-bundle-vol -c $EC2_CERT -k $EC2_PRIVATE_KEY -u <amazon_account_number> -p <elastic_image_name> --batch --debug

   where <elastic_image_name> is the name that you want to assign to your custom image (e.g. 'CustomImage1')

4. Once the image is created, you need to upload it to Amazon S3 by running the command below:
   
   

   ec2-upload-bundle -b <s3_bucket_name> -m /tmp/<elastic_image_name>.manifest.xml -a <access_key_id> -s <secret_access_key> --retry --debug

   where <s3_bucket_name>, <access_key_id> and <secret_access_key> are the Amazon S3 bucket name, Access Key ID and Secret Access Key described previously, and <elastic_image_name> is the name that you want to assign to your custom image (e.g. 'CustomImage1')

5. You will then need to register your image with Amazon EC2 by using the ec2-register command:
   
   

   ec2-register <s3_bucket_name>/<elastic_image_name>.manifest.xml

   where <s3_bucket_name> is the Amazon S3 bucket name described previously and <elastic_image_name> is the name that you want to assign to your custom image (e.g. 'CustomImage1'). The output of this command will show the AMI ID of your custom image.

Now that you have created a custom elastic image, there are two more steps that you will need to complete before you can use it.

First, you will need to associate your custom elastic image with your Bamboo installation by creating an Elastic Image Configuration. Please note the AMI ID of your new custom image and read Managing your Elastic Image Configurations for further instructions.

Secondly, you will need to configure the capabilities of the elastic agents that will run on instances started from your image. This is done by adding the appropriate builder, JDK, Perforce and custom capabilities to your elastic image configuration, so that it reflects what your custom elastic image actually can do. For example, if you have created a custom elastic image with JDK 1.6 and Maven 2 installed, you will need to add capabilities for JDK 1.6 and Maven 2 to the elastic image configuration. Read Configuring Elastic Agent Capabilities for further instructions.

If you need more help, there are a number of resources that you can take advantage of:

• AWS Support Center — if you are having problems with any of your Amazon services, not specifically related to Bamboo, you can obtain basic support from the AWS Support Center. Note, you will need to sign up for Premium Support to get access to web/phone support.
• AWS Resource Center — the AWS Resource Center has links to online documentation, code samples and tools for AWS services.
• Bamboo Developer Forums — please feel free to discuss any useful tips or issues regarding this process in the Bamboo Developer Forums.