Troubleshooting

When migrating to GitLab, you might encounter the following issues.

Imported repository is missing branches

If an imported repository does not contain all branches of the source repository:

1. Set the environment variable IMPORT_DEBUG=true.
2. Retry the import with a different group, subgroup, or project name.
3. If some branches are still missing, inspect importer.log (for example, with jq).

Exception: Error Importing repository - No such file or directory @ rb_sysopen - (filename)

The error occurs if you attempt to import a tar.gz file download of a repository’s source code.

Imports require a GitLab export file, not just a repository download file.

Diagnosing prolonged or failed imports

If you’re experiencing prolonged delays or failures with file-based imports, especially those using S3, the following may help identify the root cause of the problem:

• Check import steps
• Review logs
• Identify common issues

Check import status

Check the import status:

1. Use the GitLab API to check the import status of the affected project.
2. Review the response for any error messages or status information, especially the status and import_error values.
3. Make note of the correlation_id in the response, as it’s crucial for further troubleshooting.

Review logs

Search logs for relevant information:

For GitLab Self-Managed instances:

1. Check the Sidekiq logs and exceptions_json logs.
2. Search for entries related to RepositoryImportWorker and the correlation ID from Check import status.
3. Look for fields such as job_status, interrupted_count, and exception.

For GitLab.com (GitLab team members only):

1. Use Kibana to search the Sidekiq logs with queries like:

   Target: pubsub-sidekiq-inf-gprd*

   json.class: "RepositoryImportWorker" AND json.correlation_id.keyword: "<CORRELATION_ID>"

   or

   json.class: "RepositoryImportWorker" AND json.meta.project: "<project.full_path>"

2. Look for the same fields as mentioned for GitLab Self-Managed instances.

Identify common issues

Check the information gathered in Review logs against the following common issues:

• Interrupted jobs: If you see a high interrupted_count or job_status indicating failure, the import job may have been interrupted multiple times and placed in a dead queue.
• S3 connectivity: For imports using S3, check for any S3-related error messages in the logs.
• Large repository: If the repository is very large, the import might time out. Consider using Direct transfer in this case.