Exporting migration data from GitHub.com

Preparing the source organization on GitHub

1. Ensure that you have owner permissions on the source organization's repositories.

2. Generate an access token with the repo and admin:org scopes on GitHub.com.

3. To minimize downtime, make a list of repositories you want to export from the source instance. You can add multiple repositories to an export at once using a text file that lists the URL of each repository on a separate line.

Exporting the organization's repositories

To export repository data from GitHub.com, use the Migrations API.

The Migrations API is currently in a preview period, which means that the endpoints and parameters may change in the future. To access the Migrations API, you must provide a custom media type in the Accept header: application/vnd.github.wyandotte-preview+json. The examples below include the custom media type.

Generating a migration archive

1. Notify members of your organization that you'll be performing a migration. The export can take several minutes, depending on the number of repositories being exported. The full migration including import may take several hours so we recommend doing a trial run in order to determine how long the full process will take. For more information, see "About Migrations."

2. Start a migration by sending a POST request to the migration endpoint. You'll need:

   • Your access token for authentication.
   • A list of the repositories you want to migrate:

     curl -H "Authorization: token GITHUB_ACCESS_TOKEN" -X POST \
     -H "Accept: application/vnd.github.wyandotte-preview+json" \
     -d'{"lock_repositories":true,"repositories":["orgname/reponame", "orgname/reponame"]}' \
     https://api.github.com/orgs/orgname/migrations

   • If you want to lock the repositories before migrating them, make sure lock_repositories is set to true. This is highly recommended.
   • You can exclude file attachments by passing exclude_attachments: true to the endpoint. File attachments can be large and may needlessly bloat your final migration archive. The final archive size must be less than 20 GB.

   This request returns a unique id which represents your migration. You'll need it for subsequent calls to the Migrations API.

3. Send a GET request to the migration status endpoint to fetch the status of a migration. You'll need:

   • Your access token for authentication.
   • The unique id of the migration:

     curl -H "Authorization: token GITHUB_ACCESS_TOKEN" \
     -H "Accept: application/vnd.github.wyandotte-preview+json" \
     https://api.github.com/orgs/orgname/migrations/id

   A migration can be in one of the following states:

   • pending, which means the migration hasn't started yet.
   • exporting, which means the migration is in progress.
   • exported, which means the migration finished successfully.
   • failed, which means the migration failed.

4. After your migration has exported, download the migration archive by sending a GET request to the migration download endpoint. You'll need:

   • Your access token for authentication.
   • The unique id of the migration:

     curl -H "Accept: application/vnd.github.wyandotte-preview+json" \
     -u GITHUB_USERNAME:GITHUB_ACCESS_TOKEN \
     -L -o migration_archive.tar.gz \
     https://api.github.com/orgs/orgname/migrations/id/archive

5. The migration archive is automatically deleted after seven days. If you would prefer to delete it sooner, you can send a DELETE request to the migration archive delete endpoint. You'll need:

   • Your access token for authentication.
   • The unique id of the migration:

     curl -H "Authorization: token GITHUB_ACCESS_TOKEN" -X DELETE \
     -H "Accept: application/vnd.github.wyandotte-preview+json" \
     https://api.github.com/orgs/orgname/migrations/id/archive

6. To prepare the archived migration data for import into a GitHub Enterprise Server instance, see "Preparing to migrate data to your enterprise".