- **IssueOps** - Its like ChatOps but instead of using a chat bot, commands are invoked by commenting on a pull request (PRs are issues under the hood) - Example: commenting `.deploy` on a pull request
- **Branch Deployment** - A branch deploy is a deployment methodology that enables you to deploy a branch (or pull request) to a desired environment before merging to `main` or `master` - More on this below
- **PR** - Short for pull request
### IssueOps 🗨️
The best way to define IssueOps is to compare it to something similar, ChatOps. You may be familiar with the concept ChatOps already but in case you aren't here is a quick definition below:
> ChatOps is the process of interacting with a chat bot to execute commands directly in a chat platform. For example, with ChatOps you might do something like `.ping example.org` to check the status of a website
IssueOps adopts the same mindset but through a different medium. Rather than using a chat service to invoke the commands we use comments on a GitHub Issue or Pull Request. GitHub Actions is the runtime which executes our desired logic
### Branch Deployments 🌲
Branch deployments are a battle tested way of deploying your changes to a given environment for a variety of reasons. Branch deployments allow you to do the following:
- Deploy your changes to production **before** merging
- Deploy changes to a staging, QA, or non-production environment
#### Branch Deployment Core Concepts ⭐
> Note: The `main` branch is considered the base repository branch for all examples below
- The `main` branch is always considered to be a stable and deployable branch
- All changes are deployed to production before they are merged to the `main` branch
- To roll back a branch deployment, you deploy the `main` branch
-`noop` deployments should not make changes but rather report what they "would" have done
#### Why use branch deployments?
> To put the *merge -> deploy* model in the past!
What if your changes are bad and you broke production with the *merge -> deploy* model? Well now you have to revert your PR, get passing CI/builds, and then re-merge your changes to get back to a stable environment. With the **branch deploy** model, this is almost never the case. The `main` branch is considered to be always safe and stable
> This section will go into detail about how this Action works and hopefully inspire you on ways you can leverage it in your own projects
Let's walk through a GitHub Action workflow using this Action line by line:
```yaml
# The name of the workflow, it can be anything you wish
name: "branch deploy demo"
# The workflow to execute on is comments that are newly created
on:
issue_comment:
types: [created]
```
It is important to note that the workflow we want to run IssueOps on is `issue_comment` and `created`. This means we will not run under any other contexts for this workflow. You can edit this as you wish but it does change how this model ultimately works. For example, `issue_comment` workflows **only** use files found on `main` to run. If you do something like `on: pull_request` you could open yourself up to issues as a user could alter a file in a PR and exfil your secrets for example. Only using `issue_comment` is the suggested workflow type
```yaml
# Permissions needed for reacting and adding comments for IssueOps commands
Sets up your `demo` job, uses an ubuntu runner, and checks out your repo - Just some standard setup for a general Action. We also add an `if:` statement here to only run this workflow on pull request comments to make it a little cleaner
> Note: The Action will check the context for us anyways but this can save us a bit of CI time by using the `if:` condition
As seen above, we have two steps. One for a noop deploy, and one for a regular deploy. For example, the noop deploy could trigger a `terraform plan` and the regular deploy could be a `terraform apply`. These steps are conditionally gated by two variables:
> Example: You comment `.deploy noop` on a pull request. A noop deployment is detected so this action outputs the `noop` variable to `true`. You also have the correct permissions to execute the IssueOps command so the action also outputs the `continue` variable to `true`. This will allow the "fake noop deploy" step seen above to run and the "fake regular deploy" step will be skipped
| reaction | no | eyes | If set, the specified emoji "reaction" is put on the comment to indicate that the trigger was detected. For example, "rocket" or "eyes" |
| trigger | no | .deploy | The string to look for in comments as an IssueOps trigger. Example: ".deploy" |
| noop_trigger | no | noop | The string to look for in comments as an IssueOps noop trigger. Example: "noop" - The usage would then be ".deploy noop" |
| lock_trigger | no | .lock | The string to look for in comments as an IssueOps lock trigger. Used for locking branch deployments on a specific branch. Example: "lock" |
| unlock_trigger | no | .unlock | The string to look for in comments as an IssueOps unlock trigger. Used for unlocking branch deployments. Example: "unlock" |
| lock_info_alias | no | .wcid | An alias or shortcut to get details about the current lock (if it exists) Example: ".info" - Hubbers will find the ".wcid" default helpful ("where can I deploy") |
| environment | no | production | The name of the default environment to deploy to. Example, "production" |
| environment_targets | no | production,development,staging | Optional target environments to select for use with deployments. Example, "production,development,staging" |
| update_branch | no | warn | Determine how you want this Action to handle "out-of-date" branches. Available options: "disabled", "warn", "force". "disabled" means that the Action will not care if a branch is out-of-date. "warn" means that the Action will warn the user that a branch is out-of-date and exit without deploying. "force" means that the Action will force update the branch. Note: The "force" option is not recommended due to Actions not being able to re-run CI on commits originating from Actions itself |
| required_contexts | no | false | Manually enforce commit status checks before a deployment can continue. Only use this option if you wish to manually override the settings you have configured for your branch protection settings for your GitHub repository. Default is "false" - Example value: "context1,context2,context3" - In most cases you will not need to touch this option |
| allow_forks | no | true | Allow branch deployments to run on repository forks. If you want to harden your workflows, this option can be set to false. Default is "true" |
| admins | no | false | A comma seperated list of GitHub usernames or teams that should be considered admins by this Action. Admins can deploy pull requests without the need for branch protection approvals. Example: "monalisa,octocat,my-org/my-team" |
| admins_pat | no | false | A GitHub personal access token with "read:org" scopes. This is only needed if you are using the "admins" option with a GitHub org team. For example: "my-org/my-team" |
| noop | The string "true" if the noop trigger was found, otherwise the string "false" - Use this to conditionally control whether your deployment runs as a noop or not |
| continue | The string "true" if the deployment should continue, otherwise empty - Use this to conditionally control if your deployment should proceed or not |
> This is useful to display to the user the status of your deployment. For example, you could display the results of a `terraform apply` in the deployment comment
You can use the GitHub Actions environment to export custom deployment messages from your workflow to be referenced in the post run workflow for the `branch-deploy` Action that comments results back to your PR
Simply set the environment variable `DEPLOY_MESSAGE` to the message you want to be displayed in the post run workflow
To add custom messages to our final deployment message we need to use the GitHub Actions environment. This is so that we can dynamically pass data into the post action workflow that leaves a comment on our PR. The post action workflow will look to see if this environment variable is set (`DEPLOY_MESSAGE`). If the variable is set, it adds to to the PR comment. Otherwise, it will use a simple comment body that doesn't include the custom message.
> If you are using environment rather than repo secrets, this section will be of interest to you
For those familiar with GitHub Actions, you have probably used environments before to store secrets and trigger deployments. The syntax for doing so is very simple and usually looks like this:
```yaml
jobs:
deploy:
environment: production # right here we use an environment
runs-on: ubuntu-latest
steps:
- name: deployment
run: terraform apply -auto-approve
```
However, this has a few limitations:
- When workflows finish, so does the deployment to that environment - This means that the little green rocket doesn't "stick" to your pull request
- It is tricky to tune in environment protection rules with a single environment when using IssueOps + branch-deployments
To get around these limitations with this branch-deploy action and IssueOps, we can use two different environments. One to store our environement secrets and another to use in our branch deployments.
> Yes this isn't the most elegant solution, but it works and is very easy to accomplish
Here is a proper example for using two environments with this action:
With this Action, you can also choose the environment you wish to deploy to. This is useful if you have multiple environments and want to deploy to a specific environment.
This can be achieved with the `environment_targets` input
The IssueOps + branch-deploy model is significantly more secure than a traditional "deploy on merge" or "run on commit" model. Let's reference the workflow trigger that the branch-deploy model uses:
```yaml
on:
issue_comment:
types: [created]
```
Unlike the `on: pull_request` trigger, the `on: issue_comment` trigger only uses Actions workflow files from the default branch in GitHub. This means that a bad actor cannot open a PR with a malicious workflow edit and dump secrets, trigger bad deployments, or cause other issues. This means that any changes to the workflow files can be protected with branch protection rules to ensure only verified changes make it into your default branch.
To further harden your workflow files, it is strongly suggested to include the base permissions that this Action needs to run:
It should also be noted that this Action has built in functions to check the permissions of a user who invokes a IssueOps command. If the user does not have `write` or greater permissions to the repository, their command will be rejected
This Action supports a configurable input called `admins` which can be used to specify a list of individual GitHub users or teams that should have elevated permissions when using this Action
The `admins` input option takes a comma seperated list of GitHub handles or GitHub org teams which can bypass branch protection rules related to approvals for deployments. For example, if you give the option `admins: monalisa`, the `monalisa` user will be able to deploy without needing approval on their pull requests. CI checks will still need to pass however.
It should be noted that if you do not have pull request approvals enabled in your branch protection rules, then this option will not make a difference either way
Here is a simple example using only handles below (the monalisa and octocat users will be treated as admins):
```yaml
- uses: github/branch-deploy@vX.X.X
id: branch-deploy
with:
admins: monalisa,octocat
```
Here is an example using a mix of GitHub handles and a GitHub org team below:
In this case, all users (and future users) in the `octo-awesome-org/octo-awesome-team` team will be treated as admins in addition to the monalisa and octocat users
It should be noted if you choose to use GitHub org teams for admin definitions, you **will** need a [GitHub Personal Access Token](https://docs.github.com/en/authentication/keeping-your-account-and-data-secure/creating-a-personal-access-token) with the `read:org` scope. This is because the Action will need to make API calls on behalf of an authenticated user in the org to retrieve team memberships. If you choose to only use GitHub handles for admin definitions, then the `admins_pat` input is not required
> Note: You can read more about the `admin` option under the **inputs** section in this readme
In order to ensure your usage of this action is stable, it is highly recommended that you use either pin your action to a SHA or use a specific release tag
### Actions Tag Pinning
You can easily select the exact version you want on the GitHub Actions marketplace seen in the screenshot below:
### Actions SHA Pinning
You can also pin to an exact commit SHA as well using a third party tool such as [mheap/pin-github-action](https://github.com/mheap/pin-github-action)
There are multiple ways to leverage this action for deployment locks! Let's take a look at each option
### Deployment Locks
The suggested way to go about deployment locking is to use the built in locking feature in this Action!
Just like how you can comment `.deploy` on a pull request to trigger a deployment, you can also comment `.lock` to lock deployments. This will prevent other users from triggering a deployment. The lock is associated with your GitHub handle, so you will be able to deploy any pull request in the repository and as many times as you want. Any other user who attempts a deployment while your lock is active will get a comment on their PR telling them that a lock is in effect
To release the deployment lock, simply comment `.unlock` on any pull request in the repository at anytime. Please be aware that other users can run this same command to remove the lock (in case you get offline and forget to do so 😉)
These deployment locks come in two flavors:
-`sticky`
-`non-sticky`
**sticky** locks are locks that presist until you remove them. As seen in the example above, the `.lock` command creates a **sticky** lock that will persist until someone runs `.unlock`
**non-sticky** locks are temporary locks that only exist during a deployment. This action will automatically create a **non-sticky** lock for you when you run `.deploy`. It does this to prevent another user from running `.deploy` in another pull request and creating a deployment conflict
#### Deployment Lock Core Concepts
Let's review the core concepts of deployment locks in a short summary:
- Deployment locks are used to prevent multiple deployments from running at the same time and breaking things
- Non-sticky locks are created automatically when running `.deploy` or `.deploy noop`
- Sticky locks are created manually by commenting `.lock` on a pull request - They will presist until you remove them with `.unlock`
- Locks are associated to a user's GitHub handle - This user can deploy any pull request in the repository and as many times as they want
- Any user can remove a lock by commenting `.unlock` on any pull request in the repository
- Details about a lock can be viewed with `.lock --details`
- Like all the features of this Action, users need `write` permissions or higher to use a command
#### How do Deployment Locks Work?
This Action uses GitHub branches to create a deployment lock. When you run `.lock` the following happens:
1. The Action checks to see if a lock already exists
2. If a lock does not exists it begins to create one for you
3. The Action creates a new branch called `branch-deploy-lock`
4. The Action then creates a lock file called `lock.json` on the new branch
5. The `lock.json` file contains metadata about the lock
Now when new deployments are run, they will check if a lock exists. If it does and it doesn't belong to you, your deployment is rejected. If the lock does belong to you, the deployment will continue.
> Note: Using the locking mechanism included in this Action (above) is highly recommended over Actions concurreny. The section below will be included anyways should you have a valid reason to use it instead of the deploy lock features this Action provides
If your workflows need some level of concurrency or locking, you can leverage the native GitHub Actions concurrency feature ([documentation](https://docs.github.com/en/actions/using-jobs/using-concurrency)) to enable this.
For example, if you have two users run `.deploy` on two seperate PRs at the same time, it will trigger two deployments. In some cases, this will break things and you may not want this. By using Actions concurrency, you can prevent multiple workflows from running at once
The default behavior for Actions is to run the first job that was triggered and to set the other one as `pending`. If you want to cancel the other job, that can be configured as well. Below you will see an example where we setup a concurrency group which only allows one deployment at a time and cancels all other workflows triggered while our deployment is running:
This section will cover a few suggestions that will help you when using this Action
1. Suggest Updating Pull Request Branches - You should absolutely use this option when using the `branch-deploy` Action. This option can be found in your repository's `/settings` page
