You can also add repositories to Sourcegraph that are already cloned to disk on the host machine. This is useful for repositories requiring non-standard authentication to clone, or very large repositories on which cloning exceeds the resources available to the Docker container.
The steps documented here are intended for Sourcegraph instances running on a single node. The general process also applies for clustered deployments of Sourcegraph to Kubernetes, but you need to perform these steps on the underlying node hosting the
gitserver pod, or on the persistent volume used by the
If you’re using the default
--volume $HOME/.sourcegraph/data:/var/opt/sourcegraph argument to run the
sourcegraph/server Docker image, and the repository you want to add is named
github.com/my/repo, then follow these steps:
If Sourcegraph is running, ensure the repository is disabled so it doesn’t attempt to clone it.
On the host machine, ensure that a bare Git clone of the repository exists at
To create a new clone given its clone URL:
git clone --mirror YOUR-REPOSITORY-CLONE-URL $HOME/.sourcegraph/data/repos/github.com/my/repo/.git
Or, as an optimization, you can reuse an existing local clone to avoid needing to fetch all the repository data again:
git clone --mirror --reference PATH-TO-YOUR-EXISTING-LOCAL-CLONE --dissociate YOUR-REPOSITORY-CLONE-URL $HOME/.sourcegraph/data/repos/github.com/my/repo/.git
Ensure that the code host of the added repository is configured as an external service.
Enable the repository on the site admin repositories page.
If this repository exists on a code host that Sourcegraph directly integrates with, then use that code host’s configuration (as described in the external service documentation). After updating the external service configuration, if you used the correct repository path, Sourcegraph will detect and reuse the existing clone. (For example, if you’re working with a repository on GitHub.com, ensure that the repository path name you used is of the form