Use GraphQL to migrate repositories from GitLab to GitHub Enterprise Cloud

Step 0: Get ready to use the GitHub GraphQL API

To make GraphQL queries, you'll need to write your own scripts or use an HTTP client like Insomnia.

To learn more about getting started with the GitHub GraphQL API, including how to authenticate, see Forming calls with GraphQL.

You will send all GraphQL queries to the destination of your migration. If you're migrating to GitHub Enterprise Cloud with data residency, make sure to send queries to the endpoint for your enterprise's subdomain of GHE.com.

Step 1: Get the ownerId for your migration destination

As an organization owner in GitHub Enterprise Cloud, use the GetOrgInfo query to return the ownerId, also called the organization ID, for the organization you want to own the migrated repositories. You'll need the ownerId to identify your migration destination.

GetOrgInfo query

query(
  $login: String!
){
  organization (login: $login)
  {
    login
    id
    name
    databaseId
  }
}

Query variable	Description
login	Your organization name.

GetOrgInfo response

{
  "data": {
    "organization": {
      "login": "Octo",
      "id": "MDEyOk9yZ2FuaXphdGlvbjU2MTA=",
      "name": "Octo-org",
      "databaseId": 5610
    }
  }
}

In this example, MDEyOk9yZ2FuaXphdGlvbjU2MTA= is the organization ID or ownerId, which we'll use in the next step.

Step 2: Identify where you're migrating from

You can set up a migration source using the createMigrationSource query. You'll need to supply the ownerId, or organization ID, gathered from the GetOrgInfo query.

Your migration source is your GitLab instance.

createMigrationSource mutation

mutation createMigrationSource($name: String!, $url: String!, $ownerId: ID!) {
  createMigrationSource(input: {name: $name, url: $url, ownerId: $ownerId, type: GITLAB}) {
    migrationSource {
      id
      name
      url
      type
    }
  }
}

Set url to the full URL of your GitLab instance, such as https://gitlab.com or https://gitlab.example.com. Make sure to use GITLAB for type.

Query variable	Description
name	A name for your migration source. This name is for your own reference, so you can use any string.
ownerId	The organization ID of your organization on GitHub Enterprise Cloud.

createMigrationSource response

{
  "data": {
    "createMigrationSource": {
      "migrationSource": {
        "id": "MS_kgDaACQxYmYxOWU4Yi0wNzZmLTQ3NTMtOTdkZC1hNGUzZmYxN2U2YzA",
        "name": "GitLab Source",
        "url": "https://gitlab.com",
        "type": "GITLAB"
      }
    }
  }
}

In this example, MS_kgDaACQxYmYxOWU4Yi0wNzZmLTQ3NTMtOTdkZC1hNGUzZmYxN2U2YzA is the migration source ID, which we'll use in a later step.

Step 3: Generate and host your migration archive

Migrations from GitLab are archive-based. Instead of connecting to your GitLab instance during the migration, GitHub Enterprise Importer imports a migration archive that you generate from your GitLab project. A GitLab archive is a single file that contains both the Git source and the repository's metadata.

Before you start the migration, you must:

1. Generate a migration archive for the GitLab project you want to migrate.
2. Host the archive at a URL that GitHub Enterprise Cloud can access.

You'll provide this URL as the gitArchiveUrl value in the next step.

Generating a migration archive

Use the GitLab project export API to export the project you want to migrate. The token you use must have the api scope and a role with permission to export the project. For more information, see Manage access for a migration from GitLab to GitHub.

In the following requests, set the GITLAB_PAT environment variable to the token you created in Manage access for a migration from GitLab to GitHub. Replace GITLAB-SERVER with the host of your GitLab instance, such as gitlab.com, and replace GROUP%2FPROJECT with the URL-encoded path of your project. For example, the project acme-group/my-project is encoded as acme-group%2Fmy-project. For nested subgroups, include the full path, such as parent-group%2Fsubgroup%2Fmy-project.

1. Schedule the export.

   curl --request POST \
     --header "PRIVATE-TOKEN: $GITLAB_PAT" \
     "https://GITLAB-SERVER/api/v4/projects/GROUP%2FPROJECT/export"
   

2. Check the status of the export. Repeat this request until export_status is finished.

   curl --header "PRIVATE-TOKEN: $GITLAB_PAT" \
     "https://GITLAB-SERVER/api/v4/projects/GROUP%2FPROJECT/export"
   

3. Download the archive.

   curl --location \
     --header "PRIVATE-TOKEN: $GITLAB_PAT" \
     --output archive.tar.gz \
     "https://GITLAB-SERVER/api/v4/projects/GROUP%2FPROJECT/export/download"
   

Hosting the archive

You must host the archive at a URL that GitHub Enterprise Cloud can access. You can either upload the archive to GitHub-owned blob storage or use an external blob storage provider. For information about external providers, see Configure blob storage.

To upload the archive to GitHub-owned blob storage, you'll need the database ID of your organization on GitHub Enterprise Cloud. Replace ORGANIZATION with the name of your organization to get this ID from the id field in the response.

curl --header "Authorization: Bearer YOUR-TOKEN" \
  "https://api.github.com/orgs/ORGANIZATION"

Upload the archive with a POST request, replacing ORGANIZATION-ID with your organization's database ID. This request works for archives up to 100 MiB. For larger archives, use an external blob storage provider.

curl --request POST \
  --header "Authorization: Bearer YOUR-TOKEN" \
  --header "Content-Type: application/octet-stream" \
  --data-binary @archive.tar.gz \
  "https://uploads.github.com/organizations/ORGANIZATION-ID/gei/archive?name=archive.tar.gz"

The response includes a uri in the format gei://archive/GUID. Use this value as the gitArchiveUrl in the next step.

{
  "guid":