Awesome

Table of contents

• Introduction
• Reference
• Contributing

Introduction

Overview

git-xargs is a command-line tool (CLI) for making updates across multiple GitHub repositories with a single command. You give git-xargs:

1. a script or a command to run
2. a list of repos

and git-xargs will:

1. clone each repo
2. run your specified script or command against it
3. commit any changes
4. open pull requests
5. provide a detailed report of everything that happened

Git-xargs leverages goroutines to perform the repo-updating work in parallel, so it is very fast.

For example, have you ever needed to add a particular file across many repos at once? Or to run a search and replace to change your company or product name across 150 repos with one command? What about upgrading Terraform modules to all use the latest syntax? How about adding a CI/CD configuration file, if it doesn't already exist, or modifying it in place if it does, but only on a subset of repositories you select? You can handle these use cases and many more with a single git-xargs command.

Example: writing a new file to every repo in your GitHub organization

As an example, let's use git-xargs to create a new file in every repo:

git-xargs \
  --branch-name test-branch \
  --github-org <your-github-org> \
  --commit-message "Create hello-world.txt" \
  touch hello-world.txt

Here's what it looks like in action:

In this example, every repo in your org will have a new file named hello-world.txt written to it with the contents "Hello, World!". You'll then receive an easy-to-read printout of exactly what happened on STDOUT:

*****************************************************************
  GIT-XARGS RUN SUMMARY @ 2021-04-12 23:05:18.478435534 +0000 UTC
  Runtime in seconds: 4
*****************************************************************


COMMAND SUPPLIED

[touch hello-world.txt]

 REPOS SUPPLIED VIA --repos FILE FLAG
│────────────────────────│────────────────────────│
│ ORGANIZATION NAME (5)  │ URL                    │
│────────────────────────│────────────────────────│
│ zack-test-org          │ terraform-aws-asg      │
│ zack-test-org          │ terraform-aws-vpc      │
│ zack-test-org          │ terraform-aws-security │
│ zack-test-org          │ terraform-aws-eks      │
│ zack-test-org          │ circleci-test-1        │
│────────────────────────│────────────────────────│

 ALL REPOS THAT WERE TARGETED FOR PROCESSING AFTER FILTERING MISSING / MALFORMED REPOS
│───────────────────│────────────────────────────────────────────────────│
│ REPO NAME         │ REPO URL                                           │
│───────────────────│────────────────────────────────────────────────────│
│ terraform-aws-vpc │ https://github.com/zack-test-org/terraform-aws-vpc │
│ terraform-aws-eks │ https://github.com/zack-test-org/terraform-aws-eks │
│ circleci-test-1   │ https://github.com/zack-test-org/circleci-test-1   │
│───────────────────│────────────────────────────────────────────────────│


 REPOS THAT WERE SUCCESSFULLY CLONED TO THE LOCAL FILESYSTEM
│───────────────────│────────────────────────────────────────────────────│
│ REPO NAME         │ REPO URL                                           │
│───────────────────│────────────────────────────────────────────────────│
│ terraform-aws-eks │ https://github.com/zack-test-org/terraform-aws-eks │
│ circleci-test-1   │ https://github.com/zack-test-org/circleci-test-1   │
│ terraform-aws-vpc │ https://github.com/zack-test-org/terraform-aws-vpc │
│───────────────────│────────────────────────────────────────────────────│


 REPOS THAT SHOWED FILE CHANGES TO THEIR WORKING DIRECTORY FOLLOWING COMMAND EXECUTION
│───────────────────│────────────────────────────────────────────────────│
│ REPO NAME         │ REPO URL                                           │
│───────────────────│────────────────────────────────────────────────────│
│ terraform-aws-eks │ https://github.com/zack-test-org/terraform-aws-eks │
│ terraform-aws-vpc │ https://github.com/zack-test-org/terraform-aws-vpc │
│ circleci-test-1   │ https://github.com/zack-test-org/circleci-test-1   │
│───────────────────│────────────────────────────────────────────────────│


 REPOS THAT WERE SUPPLIED BY USER BUT DON'T EXIST (404'D) VIA GITHUB API
│────────────────────────│──────────│
│ REPO NAME              │ REPO URL │
│────────────────────────│──────────│
│ terraform-aws-asg      │          │
│ terraform-aws-security │          │
│────────────────────────│──────────│


 REPOS WHOSE SPECIFIED BRANCHES DID NOT EXIST ON THE REMOTE, AND SO WERE FIRST CREATED LOCALLY
│───────────────────│────────────────────────────────────────────────────│
│ REPO NAME         │ REPO URL                                           │
│───────────────────│────────────────────────────────────────────────────│
│ terraform-aws-eks │ https://github.com/zack-test-org/terraform-aws-eks │
│ terraform-aws-vpc │ https://github.com/zack-test-org/terraform-aws-vpc │
│ circleci-test-1   │ https://github.com/zack-test-org/circleci-test-1   │
│───────────────────│────────────────────────────────────────────────────│


*****************************************************
  PULL REQUESTS OPENED
*****************************************************
│───────────────────│────────────────────────────────────────────────────────────│
│ REPO NAME         │ PR URL                                                     │
│───────────────────│────────────────────────────────────────────────────────────│
│ circleci-test-1   │ https://github.com/zack-test-org/circleci-test-1/pull/82   │
│ terraform-aws-eks │ https://github.com/zack-test-org/terraform-aws-eks/pull/81 │
│ terraform-aws-vpc │ https://github.com/zack-test-org/terraform-aws-vpc/pull/77 │
│───────────────────│────────────────────────────────────────────────────────────│

Getting started

Installation option 1: Homebrew

If you are Homebrew user, you can install by running

$ brew install git-xargs

Installation option 2: Installing published binaries

1. Download the correct binary for your platform. Visit the releases page and download the correct binary depending on your system. Save it to somewhere on your PATH, such as /usr/local/bin/git-xargs.

2. Set execute permissions. For example, on Linux or Mac, you'd run:

   chmod u+x /usr/local/bin/git-xargs
   

3. Check it's working. Run the version command to ensure everything is working properly:

   git-xargs --version
   

Installation option 3: Run go install or go get

1. Ensure you have Golang installed and working properly on your system. Follow the official Golang install guide to get started.

2. Run go install to install the latest release of git-xargs:

   go install github.com/gruntwork-io/git-xargs@latest
   

3. Alternatively, use go install to select a specific release of git-xargs:

   go install github.com/gruntwork-io/git-xargs@v0.0.5
   

4. If you have Go 1.16 or earlier, you can use get

   go get github.com/gruntwork-io/git-xargs
   go get github.com/gruntwork-io/git-xargs@v0.0.5
   

Try it out!

1. Export a valid GitHub token. See the guide on Github personal access tokens for information on how to generate one. For example, on Linux or Mac, you'd run:

   export GITHUB_OAUTH_TOKEN=<your-secret-github-oauth-token>
   

2. Setup authentication with your Github Enterprise server. To use a Github Enterprise server, set the GITHUB_HOSTNAME environment variable:

   export GITHUB_HOSTNAME=<your-ghe-hostname.your-domain.com>
   

3. Provide a script or command and target some repos. Here's a simple example of running the touch command in every repo in your GitHub organization. Follow the same pattern to start running your own scripts and commands against your own repos!

   git-xargs \
     --branch-name "test-branch" \
     --commit-message "Testing git-xargs" \
     --github-org <enter-your-github-org-name> \
     touch git-xargs-is-awesome.txt
   

Reference

How to supply commands or scripts to run

The API for git-xargs is:

git-xargs [-flags] <CMD>

Where CMD is either the full path to a (Bash, Python, Ruby, etc) script on your local system or a binary. Note that, because the tool supports Bash scripts, Ruby scripts, Python scripts, etc, you must include the full filename for any given script, including its file extension.

In other words, all the following usages are valid:

git-xargs --repo gruntwork-io/cloud-nuke \
   --repo gruntwork-io/terraform-aws-eks \
   --branch-name my-branch \
   /usr/local/bin/my-bash-script.sh

git-xargs --repos ./my-repos.txt \
  --branch-name my-other-branch \
  touch file1.txt file2.txt

git-xargs --github-org my-github-org \
  --branch-name my-new-branch \
  "$(pwd)/scripts/my-ruby-script.rb"

Using git-xargs environment variables in commands or scripts

When executing commands or scripts, git-xargs will register the following environment variables for use by commands or scripts based on the arguments and flags provided:

Debugging runtime errors

By default, git-xargs will conceal runtime errors as they occur because its log level setting is INFO if not overridden by the --loglevel flag.

To see all errors your script or command may be generating, be sure to pass --loglevel DEBUG when running your git-xargs command, like so:

git-xargs --loglevel DEBUG \
	--repo zack-test-org/terraform-aws-eks \
	--branch-name master \
	--commit-message "add blank file" \
	--skip-pull-requests touch foo.txt

When the log level is set to debug you should see new error output similar to the following:

Total 195 (delta 159), reused 27 (delta 11), pack-reused 17  Repo=terraform-aws-eks
[git-xargs] DEBU[2021-06-29T12:11:31-04:00] Created branch                                Branc
h Name=refs/heads/master Repo=terraform-aws-eks
[git-xargs] DEBU[2021-06-29T12:11:31-04:00] Error creating new branch                     Error
="a branch named \"refs/heads/master\" already exists" Repo=terraform-aws-eks
[git-xargs] DEBU[2021-06-29T12:11:31-04:00] Error encountered while processing repo       Error
="a branch named \"refs/heads/master\" already exists" Repo name=terraform-aws-eks

Rate Limiting

git-xargs attempts to be a good citizen as regards consumption of the GitHub API. git-xargs conforms to GitHub's API integration guidelines.

In addition, git-xargs includes several features and flags to help you:

1. run jobs without tripping GitHub's rate limits
2. recover when rate limited by automatically retrying failed pull requests again, all while honoring the GitHub rate limits

Distinct processing channels for expensive and non-expensive work

git-xargs distinguishes between work that is safe to perform in parallel, such as certain git operations, and work that must be done with consideration of resource constraints, such as issuing open pull requests to GitHub's API. Therefore, git-xargs is able to perform all concurrency-safe work as quickly as possible by leveraging goroutines, while treating the more expensive open pull request API calls separately.

Pull requests are handled on a separate channel so that they can be buffered and retried in accordance with rate limiting feedback git-xargs is receiving from GitHub's API.

This means that git-xargs performs all the work upfront that it can as quickly as possible, and then moves on to serially process the pull request jobs that have resulted from the concurrency-safe work of cloning repositories, making file changes, checking the git worktree, etc.

Limit amount of parallel git clones

GitHub has a rate limiting on GitHub clones. Which is why per default only 4 clones happen in parallel. This setting can be adjusted with the --max-concurrent-clones flag. Setting it to 0 or lower means no limit is applied.

Automatic pull request retries when rate limited

By default, git-xargs will re-attempt opening a pull request that failed due to rate limiting. The --max-pr-retries flag allows you to specify how many times you'd like a given pull request to be re-attempted in case of failure due to rate limiting. By default, the value is 3, meaning that if you do not pass this flag, git-xargs will retry all rate-limit-blocked pull requests 3 times.

Automatic backoff when rate limiting is detected

When git-xargs detects that it has been rate limited by GitHub, it begins requeuing failed pull requests for retry, but with an additional, larger buffer of time in between the next attempt. This larger buffer of time is intended to allow GitHub rate limit status to return to baseline for the git-xargs client. The --seconds-to-wait-when-ratelimited flag specifies the number of seconds to wait when git-xargs detects it has been rate limited. Note that this extra buffer of time is in addition to the value specified by the --seconds-between-prs flag.

When they are provided by GitHub, git-xargs will instead use the values of any Retry-After header, or the delta of seconds between the current time and the rate limit error's reset time. When these two values are not available, git-xargs falls back to the user-specified value of --seconds-to-wait-when-ratelimited, if the flag was passed, or the default value, which is 60 seconds.

Specifying the pause between pull requests

The GitHub's API integration guidelines specify that clients should pause at least 1 second in between consecutive requests against expensive endpoints, such as the one for opening a new pull request. As a result, git-xargs defaults to pausing 1 second in between each pull request that is opened. The flag --seconds-between-prs allows you to modify this value. For example, if you were to pass --seconds-between-prs 30, then git-xargs will sleep for half a minute between issuing pull request API calls to GitHub.

Guidelines and observations from testing

In local testing, the actual thresholds for when GitHub's secondary rate limits kick in can vary depending on the number of repos you're targeting and how long your script or command takes to complete. Secondary rate limits have been observed on jobs targeting as few as 10 repositories. If you are using the rate limiting flags with reasonable values, your job should never be rate-limited in the first place.

If your job is consistently being rate-limited, try incrementally increasing the value you pass with the --seconds-between-prs flag. Passing a higher value will increase the overall time your job takes to complete, but it will also greatly decrease the likelihood of your job tripping GitHub's rate limits at all. Passing a lower value, or not passing the flag at all, will greatly increase the likelihood that your job is rate limited by GitHub.

Branch behavior

Passing the --branch-name (-b) flag is required when running git-xargs. If you specify the name of a branch that exists on your remote, its latest changes will be pulled locally prior to your command or script being run. If you specify the name of a new branch that does not yet exist on your remote, it will be created locally and pushed once your changes are committed.

Default repository branch

Any pull requests opened will be opened against the repository's default branch (whether that's main, or master or something else). You can supply an additional --base-branch-name flag to change the target for your pull requests. Be aware that this will override the base branch name for ALL targeted repositories.

Git file staging behavior

Currently, git-xargs will find and add any and all new files, as well as any existing files that were modified, within your repo and stage them prior to committing. If your script or command creates a new file, it will be committed. If your script or command edits an existing file, that change will also be committed.

Paths and script locations

Scripts may be placed anywhere on your system, but you are responsible for providing absolute paths to your scripts when invoking git-xargs:

git-xargs \
  --branch-name upgrade-tf-14 \
  --commit-message "Update modules to Terraform 0.14" \
  --repos data/batch3.txt \
  $(pwd)/scripts/my-ruby-script.rb

or

git-xargs \
  --branch-name upgrade-tf-14 \
  --commit-message "Update modules to Terraform 0.14" \
  --repos data/batch3.txt \
  /usr/local/bin/my-ruby-script.rb

If you need to compose more complex behavior into a single pull request, write a wrapper script that executes all your commands, or place all your logic into one script.

How to target repos to run your scripts against

git-xargs supports four methods of targeting repos to run your selected scripts against. They are processed in the order listed below, with whichever option is found first being used, and all others after it being ignored.

Option #1: GitHub organization lookup

If you want the tool to find and select every repo in your GitHub organization, you can pass the name of your organization via the --github-org flag:

git-xargs \
  --commit-message "Update copyright year" \
  --github-org <your-github-org> \
  "$(pwd)/scripts/update-copyright-year.sh"

This will signal the tool to look up, and page through, every repository in your GitHub organization and execute the scripts you passed.

Option #2: Flat file of repository names

Oftentimes, you want finer-grained control over the exact repos you are going to run your script against. In this case, you can use the --repos flag and supply the path to a file defining the exact repos you want the tool to run your selected scripts against, like so:

git-xargs \
  --commit-message "Update copyright year" \
  --repos data/batch2.txt \
  "$(pwd)/scripts/update-copyright-year.sh"

In this example, batch2.txt looks like this:

gruntwork-io/infrastructure-as-code-training
gruntwork-io/infrastructure-live-acme
gruntwork-io/infrastructure-live-multi-account-acme
gruntwork-io/infrastructure-modules-acme
gruntwork-io/infrastructure-modules-multi-account-acme

Flat files contain one repo per line, each repository in the format of <github-organization>/<repo-name>. Commas, trailing or preceding spaces, and quotes are all filtered out at runtime. This is done in case you end up copying your repo list from a JSON list or CSV file.

Option #3: Pass in repos via command line args

Another way to get fine-grained control is to pass in the individual repos you want to use via one or more --repo arguments:

git-xargs \
  --commit-message "Update copyright year" \
  --repo gruntwork-io/terragrunt \
  --repo gruntwork-io/terratest \
  --repo gruntwork-io/cloud-nuke \
  "$(pwd)/scripts/update-copyright-year.sh"

Option #4: Pass in repos via stdin

And one more (Unix-philosophy friendly) way to get fine-grained control is to pass in the individual repos you want to use by piping them in via stdin, separating repo names with whitespace or newlines:

echo "gruntwork-io/terragrunt gruntwork-io/terratest" | git-xargs \
  --commit-message "Update copyright year" \
  "$(pwd)/scripts/update-copyright-year.sh"

Notable flags

git-xargs exposes several flags that allow you to customize its behavior to better suit your needs. For the latest info on flags, you should run git-xargs --help. However, a couple of the flags are worth explaining more in depth here:

Best practices, tips and tricks

Write your script to run against a single repo

Write your script as if it's operating on a single repo, then target many repos with git-xargs. Remember that at runtime, each of the scripts you select will be run, in the order you specify, once per repo that you've targeted.

Handling prerequisites and third party binaries

It is currently assumed that bash script authors will be responsible for checking for prerequisites within their own scripts. If you are adding a new bash script to accomplish some new task across repos, consider using the Gruntwork bash-commons assert_is_installed pattern to ensure the operator has any required binaries installed.

Grouping your repos into separate batches

This is a pattern that ended up working out well for us as we wrote and executed more and more ambitious scripts across our many repos as a team: By breaking your target repos into separate batches, (batch1.txt, batch2.txt, batch3.txt) and starting with a few repos (or even one repo!) in the initial batches, and then gradually expanding the batches in size, you can easily test your new scripts against a few repos and double check the generated pull requests for any issues prior to widening your target batches.

How git-xargs works

This section provides a more in-depth look at how the git-xargs tool works under the hood.

1. git-xargs will clone each of your selected repos to your machine to the /tmp/ directory of your local machine. The name of each repo, plus a random number each run, are concatenated together to form the local clone name to make the local repo easier to find in case you need to debug your script locally, e.g., terraform-aws-module-security3978298.
2. it will checkout a local branch (whose name you must specify with the --branch-name flag)
3. it will run all your selected scripts against your selected repos
4. it will commit any changes in each of the repos (with a commit message you can optionally specify via the --commit-message flag)
5. it will push your local branch with your new commits to your repo's remote
6. it will call the GitHub API to open a pull request with a title and description that you can optionally specify via the --pull-request-title and --pull-request-description flags, respectively, unless you pass the --skip-pull-requests flag
7. it will print out a detailed run summary to STDOUT that explains exactly what happened with each repo and provide links to successfully opened pull requests that you can quickly follow from your terminal. If any repos encountered errors at runtime (whether they weren't able to be cloned, or script errors were encountered during processing, etc) all of this will be spelled out in detail in the final report, so you know exactly what succeeded and what went wrong.

Tasks this tool is well-suited for

The following is a non-exhaustive list of potential use cases for git-xargs:

• Add a LICENSE file to all of your GitHub repos, interpolating the correct year and company name into the file
• For every existing LICENSE file across all your repos, update their copyright date to the current year
• Update the CI build configuration in all of your repos by modifying the .circleci/config.yml file in each repo using a tool such as yq
• Run sed commands to update or replace information across README files
• Add new files to repos
• Delete specific files, when present, from repos
• Modify package.json files in-place across repos to bump a node.js dependency using jq https://stedolan.github.io/jq/
• Update your Terraform module library from Terraform 0.13 to 0.14.
• Remove stray files of any kind, when found, across repos using find and its exec option
• Add baseline tests to repos that are missing them by copying over a common local folder where they are defined
• Refactor multiple Golang tools to use new libraries by executing go get to install and uninstall packages, and modify the source code files' import references

If you can instrument the logic in a script, you can use git-xargs to run it across your repos!

Contributing

Contributing scripts to this project

We hope that this tool will help save you some time as you apply it to your own automations and maintenance tasks. We also welcome the community to contribute back scripts that everyone can use and benefit from.

Initially, we'll add these scripts to the ./scripts directory in this repository and will eventually organize them into sub-folders depending on their purposes / use cases. If you would like to have your script considered for inclusion in this repo, please first ensure that it is:

• High quality: meaning free of typos, any obvious bugs or security issues
• Generic: meaning that it is likely to be of general use to many different people and organizations, and free of any proprietary tooling or secrets

Once you've done this, please feel free to open a pull request adding your script to the ./scripts directory for consideration.

Thanks for contributing back! Our hope is that eventually this repo will contain many useful generic scripts for common maintenance and upgrading tasks that everyone can leverage to save time.

Building the binary from source

Clone this repository and then run the following command from the root of the repository:

go build

The git-xargs binary will be present in the repository root.

Running the tool without building the binary

Alternatively, you can run the tool directly without building the binary, like so:

./go run main.go \
  --branch-name test-branch \
  --commit-message "Add MIT License" \
  --repos data/test-repos.txt \
  $(pwd)/scripts/add-license.sh

This is especially helpful if you are developing against the tool and want to quickly verify your changes.

Running tests

Tests are included within their respective packages.

go test -v ./...

License

This code is released under the Apache 2.0 License. See LICENSE.txt