Skip to content

Deduplicate git objects across forks

Created by: mike-r-mclaughlin

Feedback

For large customers, there can be hundreds of repos that are forks of a single repository. It would be great to take advantage of something like what GitLab does using Git's alternates mechanism. This would drastically reduce the amount of storage needed for these repository forks.

In https://github.com/sourcegraph/accounts/issues/6716's case, it would save over 7 TB of disk space for just the top 1,000 repositories (out of > 170k).

Customer

https://github.com/sourcegraph/accounts/issues/6716 via Slack

/cc @jplahn @ryphil