Skip to content

Storage: what counts and how to free space

Every account on Arkiviva has an amount of storage: your own account, and each Organization. This page says what counts against it, what still works when it is full, and how to get space back.

What counts

All the data an account stores counts towards one total:

  • its Git repositories, public and private;
  • Git LFS files;
  • files attached to issues, comments and releases;
  • packages;
  • wikis;
  • the logs of Actions runs.

A fork is a repository of its own, so it counts against the account that owns the fork.

Your storage page shows your total, what is free, and what your account uses, kind by kind. To open it, go to your user settings and choose Storage, or open arkiviva.com/user/settings/storage_overview. An Organization has a storage page of its own, under Storage in the Organization’s settings. The figures on a storage page can take up to an hour to show a change.

When an account is full

When an account has no storage left, uploads that would add data to it are refused: a push that adds commits, a new Git LFS file, an attachment, a release file or a package. The refusal begins with “Upload refused”, for example:

Upload refused: this account uses … of its … storage. Delete branches, tags, releases, packages, repositories or wiki pages, or review its storage page, then try again.

When some storage is left, but not enough for an upload, the refusal says how large an upload the account can take now.

Reading and deleting keep working. You can still clone repositories and download their files, and you can delete data to make room. A push that only deletes branches or tags is accepted. A push that adds a commit is refused, even a commit that only removes files.

Uploads that are still running count too, so an upload can be refused while another one is in progress, even if each of them would fit on its own.

When you push to a repository that belongs to someone else, its storage is the owner’s. The refusal then asks you to ask its owner to make room.

How to free space

Start with your storage page: it shows which kind of data takes the space.

Packages, attachments, releases, LFS files and repositories

Deleting a package, an attachment, a release with its files, a Git LFS file or a whole repository frees its space at once.

Git LFS files stay when you delete the branches or commits that use them. To free them, open the repository’s settings and choose LFS, where each LFS file can be deleted. Only the repository’s administrators can do this. Delete only the files that no branch you keep still needs: deleting an LFS file that a commit still uses can cause “object does not exist” errors when that commit is checked out.

Branches and tags

Git keeps every file in the history of the commits that contain it. A commit that deletes a file therefore frees nothing: the file is still in the commits before it.

What frees Git data is deleting the branches and tags that hold it. The data that no other branch or tag still holds is removed at the next clean-up. A clean-up runs every night, and also about five minutes after a deletion when the account is full. Data that is less than a day old when a clean-up runs is kept until a later one.

If a branch had a pull request, its commits also stay with that pull request. They are freed once the pull request is deleted as well, which a repository administrator can do.

The same holds for wikis: a wiki is a Git repository, and deleting a page keeps the page in the wiki’s history. Wiki pages can be deleted while the account is full, but deleting them does not free space.

When the data is in the default branch

If the data that fills the account is in the history of a repository’s default branch, deleting other branches does not help. The default branch cannot be deleted, and a full account cannot push a rewritten history either. Then the way out is to delete the repository and push it again without the large files, or to write to us.

Before you delete a repository, make sure you have a complete clone of it. Deleting a repository is permanent and cannot be undone: it deletes the repository’s code, issues, comments, wiki and collaborator settings. A push does not create a repository, so create it again on arkiviva.com before you push to it.

Keeping large files easy to remove

Large files are easier to free when they are stored with Git LFS: each LFS file can be deleted on its own, while a large file in Git’s history stays until every branch and tag that holds it is gone.