git-update-server-info

git update-server-info

git-update-server-info is a Git plumbing command used to generate or update essential auxiliary files in a bare Git repository. Specifically, it creates or refreshes the info/refs and objects/info/packs files.

These files are crucial for Git clients that interact with the repository using the 'dumb' HTTP protocol. The info/refs file provides a list of all available references (branches, tags) and their corresponding object IDs, allowing clients to discover what's in the repository. The objects/info/packs file lists all available packfiles, which contain the actual Git objects.

Without these files, a 'dumb' HTTP client cannot discover the repository's contents and thus cannot clone or fetch. 'Smart' protocols (like SSH, the native Git protocol, or smart HTTP) do not require these files as they negotiate references and transfer objects directly with the server.

This command is typically invoked automatically as part of a post-update hook after a successful push to ensure the auxiliary files are always up-to-date. It requires the repository to be bare; it will fail if run in a non-bare repository.

The git-update-server-info command is specifically designed for and only relevant to serving Git repositories via the 'dumb' HTTP protocol. It has no effect on smart protocols (e.g., Git protocol, SSH, or smart HTTP). It also requires the repository to be a bare repository; running it in a non-bare repository will result in an error. Care should be taken with symlinks, as certain configurations (e.g., http.serverAllowSymlinks) can expose them.

This command is most commonly executed automatically through a post-update hook in the Git repository. When a user pushes changes to a bare repository, the post-update hook triggers git-update-server-info to ensure that the info/refs and objects/info/packs files are refreshed, making the latest changes discoverable by 'dumb' HTTP clients.

It's crucial to understand that git-update-server-info solely serves the 'dumb' HTTP protocol. For users accessing your repository via SSH, the native Git protocol (git://), or 'smart' HTTP (which uses git-http-backend), this command's output is not used. These 'smart' protocols negotiate directly with the Git server for references and object data, making the auxiliary info files redundant for their operation.

git-update-server-info dates back to the early days of Git, playing a vital role in enabling web-based access to repositories before 'smart' HTTP protocol became common. It was essential for simple static HTTP servers to offer Git read-only access. While modern Git hosting often relies on smart protocols, this command remains relevant for basic, low-overhead HTTP setups, embodying Git's flexibility in server-side deployment.

git-http-backend(1), git-daemon(1), git-config(1), git-clone(1)