Have you done a git push --force and overwritten some work that you don’t have backed up anywhere else? I just experienced this today but I was able to recover my work.

Fortunately, git is pretty good about not throwing away anything, so it is possible your work still exists on the server.

In my case, I used git push --force to erase some work on the server knowing that I still had the work on a branch on my local machine. Unfortunately, this doesn’t protect from hardware failure or unintended deletion - which is what happened in my case.

1. Find the lost SHA

Is there anywhere that you can reference a SHA pointing to the lost work? Possible places to look might be:

  • Email (Git messages)
  • Chat
  • Build service (Jenkins, etc)

I was saved by Jenkins because a previous build contained a reference the SHA that initiated it.

Once you find the SHA, place it in the URL to your GitHub repository like this:

https://github.com/coryfklein/myrepo/commit/<SHA_HERE>

And voila! You have a link to your lost work.

You can click on “parent” links to open any other related commits you may have lost as well.

2. Create Patch Files

For each commit you can get a patch file by appending .patch to the above URL like so:

https://github.com/coryfklein/myrepo/commit/<SHA_HERE>.patch

Save this file to disk, repeating as necessary for each commit you need to recover.

3. Apply the Patch Files

Use git am to apply your patch files to your repository.

git am 001-work.patch 002-work.patch 003-work.patch

4. Save Your Work

When you’re ready, push your recovered work to master:

git push

Otherwise, save your work to a branch and push that to the server so it is saved both locally and remotely:

git co -b my-saved-work
git push --set-upstream origin my-saved-work