Skip to content
GitLab

Import your project from Bitbucket Server (FREE)

Introduced in GitLab 11.2.

NOTE: This process is different than importing from Bitbucket Cloud.

From Bitbucket Server, you can import:

  • Repository description
  • Git repository data
  • Pull requests
  • Pull request comments

When importing, repository public access is retained. If a repository is private in Bitbucket, it's created as private in GitLab as well.

Import your Bitbucket repositories

Ability to re-import projects introduced in GitLab 15.9.

Prerequisites:

  • An administrator must enable Bitbucket Server in Admin > Settings > General > Visibility and access controls > Import sources.
  • At least the Maintainer role on the destination group to import to. Using the Developer role for this purpose was deprecated in GitLab 15.8 and will be removed in GitLab 16.0.

To import your Bitbucket repositories:

  1. Sign in to GitLab.
  2. On the top bar, select New ({plus}).
  3. Select New project/repository.
  4. Select Import project.
  5. Select Bitbucket Server.
  6. Log in to Bitbucket and grant GitLab access to your Bitbucket account.
  7. Select the projects to import, or import all projects. You can filter projects by name and select the namespace for which to import each project.
  8. To import a project:
    • For the first time: Select Import.
    • Again: Select Re-import. Specify a new name and select Re-import again. Re-importing creates a new copy of the source project.

Items that are not imported

The following items aren't imported:

  • Pull request approvals
  • Attachments in Markdown
  • Task lists
  • Emoji reactions

Items that are imported but changed

The following items are changed when they are imported:

  • GitLab doesn't allow comments on arbitrary lines of code. Any out-of-bounds Bitbucket comments are inserted as comments in the merge request.
  • Multiple threading levels are collapsed into one thread and quotes are added as part of the original comment.
  • Declined pull requests have unreachable commits. These pull requests show up as empty changes.
  • Project filtering doesn't support fuzzy search. Only starts with or full match strings are supported.

User assignment

Prerequisite:

  • Authentication token with administrator access.

When issues and pull requests are importing, the importer tries to find the author's email address with a confirmed email address in the GitLab user database. If no such user is available, the project creator is set as the author. The importer appends a note in the comment to mark the original creator.

The importer creates any new namespaces (groups) if they don't exist. If the namespace is taken, the repository imports under the namespace of the user who started the import process.

User assignment by username

  • Introduced in GitLab 13.4 with a flag named bitbucket_server_user_mapping_by_username. Disabled by default.
  • Not recommended for production use.

FLAG: On self-managed GitLab and GitLab.com, by default this feature is not available. To make it available, ask an administrator to enable the feature flag named bitbucket_server_user_mapping_by_username. This feature is not ready for production use.

With this feature enabled, the importer tries to find a user in the GitLab user database with the author's:

  • username
  • slug
  • displayName

If no user matches these properties, the project creator is set as the author.

Troubleshooting

General

If the GUI-based import tool does not work, you can try to:

See the troubleshooting section for Bitbucket Cloud.

LFS objects not imported

If the project import completes but LFS objects can't be downloaded or cloned, you may be using a password or personal access token containing special characters. For more information, see this issue.

Pull requests are missing

Importing large projects spawns a process that can consume a lot of memory. Sometimes you can see messages such as Sidekiq worker RSS out of range in the Sidekiq logs. This can mean that MemoryKiller is interrupting the RepositoryImportWorker because it's using too much memory.

To resolve this problem, temporarily increase the SIDEKIQ_MEMORY_KILLER_MAX_RSS environment variable using custom environment variables from the default 2000000 value to a larger value like 3000000. Consider memory constraints on the system before deciding to increase SIDEKIQ_MEMORY_KILLER_MAX_RSS.

Related topics