Why am I getting the message, “fatal: This operation must be run in a work tree?”

Question

Just installed git on Windows. I set the GIT_DIR variable to be c:\\git\\ and verified that this environment variable is maintained by cygwin (i.e. echo $GIT_DIR is what it

Answer 1

In addition, apparently, this error will happen if you clone into NTFS Ram Drive.

Answer 2

Just in case what happened to me is happening to somebody else, I need to say this: I was in my .git directory within my project when I was getting this error. I searched and scoured for answers, but nothing worked. All I had to do was get back to the right directory. It was kind of a face-palm moment for me. In case there's anyone else out there as silly as me, I hope you found this answer helpful.

Answer 3

Create a bare GIT repository

A small rant: git is unable to create a normal bare repository by itself. Stupid git indeed.

To be precise, it is not possible to clone empty repositories. So an empty repository is a useless repository. Indeed, you normally create an empty repository and immediately fill it:

git init
git add .

However, git add is not possible when you create a bare repository:

git --bare init
git add .

gives an error "fatal: This operation must be run in a work tree".

You can't check it out either:

Initialized empty Git repository in /home/user/myrepos/.git/
fatal: http://repository.example.org/projects/myrepos.git/info/refs not found: did you run git update-server-info on the server?

git --bare init
git update-server-info # this creates the info/refs file
chown -R <user>:<group> . # make sure others can update the repository

The solution is to create another repository elsewhere, add a file in that repository and, push it to the bare repository.

mkdir temp; cd temp
git init
touch .gitignore
git add .gitignore
git commit -m "Initial commit"
git push (url or path of bare repository) master
cd ..; rm -rf temp

hope this can help u

Answer 4

Explicitly setting the GIT_DIR environment variable forces git to use the given directory as the git repository. It is never needed during normal use.

In your example, because have specified a GIT_DIR and it isn't named .git (the leading dot is important) and you haven't provided a --work-tree option or set the GIT_WORK_TREE environment variable, that you want a bare repository when you said git init.

Because a bare repository has no working tree a large selection of commands don't make sense with a bare repository. git add is just one.

Is there a particular reason that you need to use a non-standard location for your git repository, rather than in a .git subfolder under the working tree root? While it's possible to arrange this it tends to be more work and more liable to user mistakes.

Answer 5

The direct reason for the error is that yes, it's impossible to use git-add with a bare repository. A bare repository, by definition, has no work tree. git-add takes files from the work tree and adds them to the index, in preparation for committing.

You may need to put a bit of thought into your setup here, though. GIT_DIR is the repository directory used for all git commands. Are you really trying to create a single repository for everything you track, maybe things all over your system? A git repository by nature tracks the contents of a single directory. You'll need to set GIT_WORK_TREE to a path containing everything you want to track, and then you'll need a .gitignore to block out everything you're not interested in tracking.

Maybe you're trying to create a repository which will track just c:\www? Then you should put it in c:\www (don't set GIT_DIR). This is the normal usage of git, with the repository in the .git directory of the top-level directory of your "module".

Unless you have a really good reason, I'd recommend sticking with the way git likes to work. If you have several things to track, you probably want several repositories!

Answer 6

This should solve it:

git config --unset core.bare