Skip to content

Troubleshooting repository mirroring

DETAILS: Tier: Free, Premium, Ultimate Offering: SaaS, self-managed

When mirroring fails, project maintainers can see a link similar to {warning-solid} Pull mirroring failed 1 hour ago. on the project details page. Select this link to go directly to the mirroring settings, where GitLab displays an Error badge for the mirrored repository. You can hover your mouse cursor over the badge to display the text of the error:

Error message shown on hover

Received RST_STREAM with error code 2 with GitHub

If you receive this message while mirroring to a GitHub repository:

13:Received RST_STREAM with error code 2

One of these issues might be occurring:

  1. Your GitHub settings might be set to block pushes that expose your email address used in commits. To fix this problem, either:
  2. Your repository exceeds GitHub's file size limit of 100 MB. To fix this problem, check the file size limit configured for on GitHub, and consider using Git Large File Storage to manage large files.

Deadline Exceeded

When upgrading GitLab, a change in how usernames are represented means that you must update your mirroring username and password to ensure that %40 characters are replaced with @.

Connection blocked because server only allows public key authentication

The connection between GitLab and the remote repository is blocked. Even if a TCP Check is successful, you must check any networking components in the route from GitLab to the remote server for blockage.

This error can occur when a firewall performs a Deep SSH Inspection on outgoing packets.

Could not read username: terminal prompts disabled

If you receive this error after creating a new project using GitLab CI/CD for external repositories:

  • In Bitbucket Cloud:

    "2:fetch remote: "fatal: could not read Username for 'https://bitbucket.org':
    terminal prompts disabled\n": exit status 128."
  • In Bitbucket Server (self-managed):

    "2:fetch remote: "fatal: could not read Username for 'https://lab.example.com':
    terminal prompts disabled\n": exit status 128.

Check if the repository owner is specified in the URL of your mirrored repository:

  1. On the left sidebar, select Search or go to and find your project.

  2. Select Settings > Repository.

  3. Expand Mirroring repositories.

  4. If no repository owner is specified, delete and add the URL again in this format, replacing OWNER, ACCOUNTNAME, PATH_TO_REPO, and REPONAME with your values:

    • In Bitbucket Cloud:

      https://OWNER@bitbucket.org/ACCOUNTNAME/REPONAME.git
    • In Bitbucket Server (self-managed):

      https://OWNER@lab.example.com/PATH_TO_REPO/REPONAME.git

When connecting to the Cloud or self-managed Bitbucket repository for mirroring, the repository owner is required in the string.

Pull mirror is missing LFS files

In some cases, pull mirroring does not transfer LFS files. This issue occurs when:

  • You use an SSH repository URL. The workaround is to use an HTTPS repository URL instead. An issue exists to fix this problem for SSH URLs.
  • You're using GitLab 14.0 or older, and the source repository is a public Bitbucket URL. Fixed in GitLab 14.0.6.
  • You mirror an external repository using object storage. An issue exists to fix this problem.

Pull mirroring is not triggering pipelines

Pipelines might not run for multiple reasons:

The repository is being updated, but neither fails nor succeeds visibly

In rare cases, mirroring slots on Redis can become exhausted, possibly because Sidekiq workers are reaped due to out-of-memory (OoM) events. When this occurs, mirroring jobs start and complete quickly, but they neither fail nor succeed. They also do not leave a clear log. To check for this problem:

  1. Enter the Rails console and check Redis' mirroring capacity:

    current = Gitlab::Redis::SharedState.with { |redis| redis.scard('MIRROR_PULL_CAPACITY') }.to_i
    maximum = Gitlab::CurrentSettings.mirror_max_capacity
    available = maximum - current
  2. If the mirroring capacity is 0 or very low, you can drain all stuck jobs with:

    Gitlab::Redis::SharedState.with { |redis| redis.smembers('MIRROR_PULL_CAPACITY') }.each do |pid|
      Gitlab::Redis::SharedState.with { |redis| redis.srem('MIRROR_PULL_CAPACITY', pid) }
    end
  3. After you run the command, the background jobs page should show new mirroring jobs being scheduled, especially when triggered manually.

Invalid URL

If you receive this error while setting up mirroring over SSH, make sure the URL is in a valid format.

Mirroring does not support the short version of SSH clone URLs (git@gitlab.com:gitlab-org/gitlab.git) and requires the full version including the protocol (ssh://git@gitlab.com/gitlab-org/gitlab.git).

Make sure that host and project path are separated using / instead of :.

Host key verification failed

This error is returned when the target host public SSH key changes. Public SSH keys rarely, if ever, change. If host key verification fails, but you suspect the key is still valid, you can refresh the key's information.

Prerequisites:

  • You must have at least the Maintainer role for a project.

To resolve the issue:

  1. Verify the host key.

  2. On the left sidebar, select Search or go to and find your project.

  3. Select Settings > Repository.

  4. Expand Mirroring repositories.

  5. To refresh the keys, either:

    • Select Detect host keys for GitLab to fetch the host keys from the server, and display the fingerprints.
    • Select Input host keys manually, and enter the host key into the SSH host key field.
  • Select Mirror repository.

Transfer mirror users and tokens to a single service account in Rails console

This requires access to the GitLab Rails console.

Use case: If you have multiple users using their own GitHub credentials to set up repository mirroring, mirroring breaks when people leave the company. Use this script to migrate disparate mirroring users and tokens into a single service account:

WARNING: Commands that change data can cause damage if not run correctly or under the right conditions. Always run commands in a test environment first and have a backup instance ready to restore.

svc_user = User.find_by(username: 'ourServiceUser')
token = 'githubAccessToken'

Project.where(mirror: true).each do |project|
  import_url = project.import_url

  # The url we want is https://token@project/path.git
  repo_url = if import_url.include?('@')
               # Case 1: The url is something like https://23423432@project/path.git
               import_url.split('@').last
             elsif import_url.include?('//')
               # Case 2: The url is something like https://project/path.git
               import_url.split('//').last
             end

  next unless repo_url

  final_url = "https://#{token}@#{repo_url}"

  project.mirror_user = svc_user
  project.import_url = final_url
  project.username_only_import_url = final_url
  project.save
end

The requested URL returned error: 301

When mirroring using the http:// or https:// protocols, be sure to specify the exact URL to the repository: https://gitlab.example.com/group/project.git

HTTP redirects are not followed and omitting .git can result in a 301 error:

13:fetch remote: "fatal: unable to access 'https://gitlab.com/group/project': The requested URL returned error: 301\n": exit status 128.

Push mirror from GitLab instance to Geo secondary fails: The requested URL returned error: 302

Push mirroring of a GitLab repository using the HTTP or HTTPS protocols fails when the destination is a Geo secondary node due to the proxying of the push request to the Geo primary node, and the following error is displayed:

13:get remote references: create git ls-remote: exit status 128, stderr: "fatal: unable to access 'https://gitlab.example.com/group/destination.git/': The requested URL returned error: 302".

This occurs when a Geo unified URL is configured and the target host name resolves to the secondary node's IP address.

The error can be avoided by:

  • Configuring the push mirror to use the SSH protocol. However, the repository must not contain any LFS objects, which are always transferred over HTTP or HTTPS and are still redirected.
  • Using a reverse proxy to direct all requests from the source instance to the primary Geo node.
  • Adding a local hosts file entry on the source to force the target host name to resolve to the Geo primary node's IP address.
  • Configuring a pull mirror on the target instead.