Fork of https://bitbucket.org/atlassian/dc-deployments-automation.git
This repository is a suite of Ansible roles, playbooks and support scripts to automate the installation and maintenance of Atlassian Data Center products in cloud environments.
On this page:
[TOC]
The usual scenario for usage as part of a cloud deployment is to invoke the script as part of post-creation actions invoked while a new product node is being brought up. In the case of AWS, this would usually be done by cfn-init/user-data. For example, the Jira quickstart template creates a per-node launch configuration that fetches this repository and runs the appropriate AWS/product playbook, which invokes the appropriate roles.
In practice, the Ansible roles require some information about the infrastructure
that was deployed (e.g. RDS endpoint/password). The way this is currently
achieved (on AWS) is that have the CloudFormation template dump this information
into the file /etc/atl as RESOURCE_VAR=<resource> lines. This can be then
sourced as environment variables to be retrieved at runtime. See the
helper-script bin/ansible-with-atl-env and the corresponding
groups_vars/aws_node_local.yml var-file.
To customise playbook behaviour, you can fork this repository and edit it as needed. However, for one-off tasks you can also override the default and calculated settings with special values. To do this, provide command-line overrides to ansible-playbook.
The most likely use-case for this is to download a custom product distribution
for testing (for example, a pre-release version of Jira). If you are running ansible-playbook
directly, the command for this would look like the following:
ansible-playbook \
-e atl_product_download_url=http://s3.amazon.com/atlassian/jira-9.0.0-PRE-TEST.tar.gz \
-e atl_use_system_jdk=true \
-e atl_download_format=tarball \
\
-i inv/aws_node_local aws_jira_dc_node.yml
You can also do this on a CloudFormation template where the stack details are in /etc/atl.
On such templates, the variable ATL_DEPLOYMENT_REPOSITORY_CUSTOM_PARAMS is added to the
ansible-playbook parameters in bin/ansible-with-alt-env. In this case you
need to set it to:
ATL_DEPLOYMENT_REPOSITORY_CUSTOM_PARAMS="-e atl_product_download_url=http://s3.amazon.com/atlassian/jira-9.0.0-PRE-TEST.tar.gz -e atl_use_system_jdk=true -e atl_download_format=tarball"
To set the same parameters in the AWS Quick Starts for
Jira Data Center,
Confluence Data Center, and
Bitbucket Data Center, enter
them in the Custom command-line parameters for Ansible field:
-e atl_product_download_url=http://s3.amazon.com/atlassian/jira-9.0.0-PRE-TEST.tar.gz -e atl_use_system_jdk=true -e atl_download_format=tarball
The roles in this repository currently target:
- Ansible-core 2.16
- Python >= 3.10 (as required by ansible-core 2.16)
- Amazon Linux 2023 and Debian 12+ (including derivatives, i.e., Ubuntu 22.04+) where the system-installable Python meets the above requirement
To use a previous version of this repository and the roles/playbooks within, your application nodes must clone/checkout
a previous commit that supports the desired OS and/or Ansible version. For instance, to continue using Ansible 2.13 on
Amazon Linux 2, use branch "ansible-core-2.13" and/or commit ID e5af2cf649f72bb5c9d50d0057ddae4a5c99b6f9. If using one
of the previously-provided AWS CloudFormation templates, you must set set the Deployment Automation Branch parameter
to "ansible-core-2.13" and/or manually set the stack's "pinned-ansible-sha" SSM Parameter to the referenced commit ID.
For more deployment customization options, consult the following files for parameters you can override:
More customizable parameters are defined in specific roles -- specifically, in the
role's defaults/main.yml file. Most of these parameters use the atl_ prefix. You can
use the following Bitbucket code search query
to find them:
repo:dc-deployments-automation repo:dc-deployments-automation path:*/defaults/main.yml atl
jira-config.properties: If this file exists in the shared home (default:/media/atl/jira/shared/) then it will be copied to the Jira local home directory (default:/var/atlassian/application-data/jira/). Note that this only happens on node creation; if you create or update the shared file it will need to be copied manually.
See Development for details on setting up a development environment and running tests.
This suite is intended to consist of many small, composable roles that can be combined together into playbooks. Wherever possible, roles should be product-agnostic (e.g. downloads) and platform-agnostic. Functions that are product-specific or platform-specific are split off into separate roles.
Roles should be reasonably self-contained, with sensible defaults configured in
/roles/<role>/defaults/main.yml. Like all playbook parameters, you can override
them at runtime.
Some roles implicitly depend on other variables beind defined elsewhere.
For example, the jira_config role depends on the atl_cluster_node_id
var being defined; on AWS this is provided by the aws_common role, which
should be run first.
- Helper scripts are in
bin/. In particular thebin/ansible-with-atl-envwrapper is of use during AWS node initialisation. Refer to the Usage section for more information. - Inventory files are under
inv/. For AWScfn-initthe inventoryinv/aws_node_localinventory is probably what you want.- Note that this expects the environment to be setup with infrastructure information. Refer to the Usage section for more information.
- Global group vars loaded automatically from
group_vars/<group>.yml. In particular notegroup_vars/aws_node_local.ymlwhich loads infrastructure information from the environment. - Roles are defined in
roles/- Platform specific roles start with
<platform-shortname>_..., e.g.roles/aws_common/. - Similarly, product-specific roles should start with
<product>_....
- Platform specific roles start with
If you find any bugs in this repository, or have feature requests or use cases for us, please raise them in our public Jira project.
Copyright © 2019 Atlassian Corporation Pty Ltd. Licensed under the Apache License, Version 2.0.