ChatGPT解决这个技术问题 Extra ChatGPT

How can I remove a commit on GitHub? [duplicate]

This question already has answers here: Undoing a 'git push' (16 answers) Closed 2 years ago. The community reviewed whether to reopen this question 10 months ago and left it closed: Original close reason(s) were not resolved

I "accidentally" pushed a commit to GitHub.

Is it possible to remove this commit?

I want to revert my GitHub repository as it was before this commit.

Word of caution: Do not ever do this when you have a lot of people following your repository, you will make their local repository go out of sync if they have pulled in the latest changes. If this concerns a mistake, you can just do another commit undoing the mistake. If this concerns a password, you might want to change the password instead and don't hurry to delete this. Forcing things does not go without drawbacks.
Word of caution 2: The commit can still be accessible directly via SHA1. Force push does not delete the commit, it creates a new one and moves the file pointer to it. To truly delete a commit you must delete the whole repo.
@Gustav "... you must delete the whole repo." - Or just force garbage collection to kick in.
With regard to WOC1, the next time followers pull, they will automatically get the new history and lose the old one, which seems quite acceptable behaviour. The problem is if other people may have committed new work after your commit: then you are causing a significant amount of hassle for them (they will need to cherry-pick their changes onto the new history). This makes a force push more acceptable after 1 minute than after 1 week (fewer followers exposed), and more acceptable for projects which people use but don't modify (they won't notice the timeline was changed).

K
Kenly

Note: please see an alternative to git rebase -i in the comments below— git reset --soft HEAD^

First, remove the commit on your local repository. You can do this using git rebase -i. For example, if it's your last commit, you can do git rebase -i HEAD~2 and delete the second line within the editor window that pops up.

Then, force push to GitHub by using git push origin +branchName --force

See Git Magic Chapter 5: Lessons of History - And Then Some for more information (i.e. if you want to remove older commits).

Oh, and if your working tree is dirty, you have to do a git stash first, and then a git stash apply after.


More accurately, you /have/ to stash because git rebase -i won't let you if you have a dirty tree.
Note that this will still leave the commit in the reflog. If you have sensitive data in there, you may have to delete the repo entirely.
I'm confused. Why is it not possible to uncommit with git reset --soft HEAD^ and then do git push origin +master? Why are we using git rebase -i HEAD^^ in this case?
@Dennis because I wasn't familiar with reset --soft 3.5 years ago. =)
Everyone beware. See subutux's comment below. Even after force pushing to GitHub, GH still caches your commit. From help.github.com/articles/remove-sensitive-data : "Danger: Once the commit has been pushed you should consider the data to be compromised. If you committed a password, change it! If you committed a key, generate a new one."
D
Dustin
git push -f origin HEAD^:master

That should "undo" the push.


This worked fine too! It removes the push from github but leaves my local repository intact. Thanks!
Well, yes. It only does what you asked for. :) Your repository and the remote repository don't have to have matching refs.
Note, however, that this only moves the branch pointer. The accidentally pushed commit is still present in the remote repo. In GitHub's case, this means that it can still be seen if you know the SHA-1 hash (from user activity history, for example).
do: git push -f origin HEAD^^:master to reverse the 2 last changes, works n times
@ianj Note that HEAD with n ^'s can be replaced by HEAD~n, e.g. HEAD~3 instead of HEAD^^^.
k
karan.dodia

For an easy revert if it's just a mistake (perhaps you forked a repo, then ended up pushing to the original instead of to a new one) here's another possibility:

git reset --hard 71c27777543ccfcb0376dcdd8f6777df055ef479

Obviously swap in that number for the number of the commit you want to return to.

Everything since then will be deleted once you push again. To do that, the next step would be:

git push --force

WARNING: This will rewrite your history, you will lose the commit and is generally not a very nice thing to do in a collaborative environment.
Yes this was the easiest and best for me. My dev fork needed to be reverted before I could send a PR for something else. I should have put my changes in a branch to begin with.
This works on Unprotected branches. If the Github branch is protected, forced push will fail.
Or: || git reset --hard HEAD^1 || (1 is the last commit)
As I couldn't edit I add another comment, in case someone does this silly thing without thinking before just like me, you can go back with a "git reflog" and then a "git reset --hard xxx" where xxx is the last commit you've made before the accidental reset...
k
kate

git log to find out the commit you want to revert git push origin +7f6d03:master while 7f6d03 is the commit before the wrongly pushed commit. + was for force push

And that's it.

Here is a very good guide that solves your problem, easy and simple!


I recommend that guide, super helpful in this.
This works for me for deleting a few latest published commits from a branch.
After some moment trying to discover why this command was not working, you clearly pointed out here to insert the last desirable commit. And I was trying to make the first commit which is why I received the message it was up-to-date already.
This should the accepted answer since the question did not specify which commit to delete. The accepted answer only deletes the last commit. This answer deletes any commit.
@kate this should be the accepted answer, so simple!! thank you
L
Loukan ElKadi

In case you like to keep the commit changes after deletion:

Note that this solution works if the commit to be removed is the last committed one.

1 - Copy the commit reference you like to go back to from the log:

git log

2 - Reset git to the commit reference:

 git reset <commit_ref>

3 - Stash/store the local changes from the wrong commit to use later after pushing to remote:

 git stash

4 - Push the changes to remote repository, (-f or --force):

git push -f

5 - Get back the stored changes to local repository:

git stash apply

7 - In case you have untracked/new files in the changes, you need to add them to git before committing:

git add .

6 - Add whatever extra changes you need, then commit the needed files, (or use a dot '.' instead of stating each file name, to commit all files in the local repository:

git commit -m "<new_commit_message>" <file1> <file2> ...

or

git commit -m "<new_commit_message>" .

Thanks! I was just happy to get it back to 5. Then I could work with changes in github desktop.
s
subutux

You'll need to clear out your cache to have it completely wiped. this help page from git will help you out. (it helped me) http://help.github.com/remove-sensitive-data/


+1: whilst removing the commit from your branch, it's still available on GitHub if you know the URL/SHA-1. The only way of removing the commit from the cache is by contacting GH support (see the 'Cached Data on Github' section in that link
I do want to note that it may have been a better idea to actually explain what's on the page, and quote the relevant sections. From the help guide, it says: "Links to external resources are encouraged, but please add context around the link so your fellow users will have some idea what it is and why it’s there. Always quote the most relevant part of an important link, in case the target site is unreachable or goes permanently offline." Rule of thumb: pretend the link isn't there, or they can't click on it.
There is another way to clear the cache: Delete the repository, then recreate and push. While this might not be possible for everyone, for many it is way easier and faster than contacting Github Support (and maybe you do not want to point to a third party [Github] that you pushed sensitive data).
E
Ekambaram E

Delete the most recent commit, keeping the work you've done:

git reset --soft HEAD~1

Delete the most recent commit, destroying the work you've done:

git reset --hard HEAD~1

what about the push afterwards?
a
alper
git reset HEAD^ --hard
git push origin -f

This work for me.


Says "Everything up-to-date" when commit is already pushed to GitHub...
If commit A is the top one, B is the second one. Which commit do you want to reset?
A ist the latest and I would like to go back to B.
git reset B --hard then git push origin -f
This worked best for me when I wanted to delete a commit from a branch I was working on.
J
Jyoti Prakash

Use git revert for reverting your push.

git-revert - Revert some existing commits git revert [--edit | --no-edit] [-n] [-m parent-number] [-s] ... git revert --continue git revert --quit git revert --abort

Revert the changes that the related patches introduce, and record some new commits that record them. This requires your working tree to be clean (no modifications from the HEAD commit).

Note: git revert is used to record some new commits to reverse the effect of some earlier commits (often only a faulty one). If you want to throw away all uncommitted changes in your working directory, you should see git-reset, particularly the --hard option.


That is the recommended way to avoid breaking other people's clones. It create a new commit that undoes an earlier commit. However it does not answer the question, which was to remove a commit from "history".
many of the other answers are good, but this is the only way that seems to work if you cannot do force pushes to master.
example usage would help, better than paste from man page. Help people wade through [] stuff.
C
Community

To delete the commit from the remote repository:

 git push -f origin last_known_good_commit:branch_name

In order delete the commit from your local repository:

git reset --hard HEAD~1

link


B
Benny Neugebauer

You need to know your commit hash from the commit you want to revert to. You can get it from a GitHub URL like: https://github.com/your-organization/your-project/commits/master

Let's say the hash from the commit (where you want to go back to) is "99fb454" (long version "99fb45413eb9ca4b3063e07b40402b136a8cf264"), then all you have to do is:

git reset --hard 99fb45413eb9ca4b3063e07b40402b136a8cf264
git push --force

o
orb

Find the ref spec of the commit you want to be the head of your branch on Github and use the following command:

git push origin +[ref]:[branchName]

In your case, if you just want to go back one commit, find the beginning of the ref for that commit, say for example it is 7f6d03, and the name of the branch you want to change, say for example it is master, and do the following:

git push origin +7f6d03:master

The plus character is interpreted as --force, which will be necessary since you are rewriting history.

Note that any time you --force a commit you could potentially rewrite other peoples' history who merge your branch. However, if you catch the problem quickly (before anyone else merges your branch), you won't have any issues.


l
loopbackbee

If you are doing this because you have sensitive data in a commit, using the other answers here is not safe (excepting subutux's, which I'll expand on).

The github guide on this recommends using a external tool, but I prefer using the built-in one.

Firstly, make a backup of your repository. Then:

git filter-branch --force --index-filter \
'git rm --cached --ignore-unmatch PATH-TO-YOUR-FILE-WITH-SENSITIVE-DATA' \
--prune-empty --tag-name-filter cat -- --all

After this, make sure the repository is in the state you want. You might want to diff against the backup.

If you're sure it's correct, then:

#get rid of old unreferenced commits (including the data you want to remove)
git gc --prune=now
git push origin --force --all

You might want to keep the local backup for a while, just in case.


Related: stackoverflow.com/questions/872565/… I think this method is not enough, the data is still accessible via the commit.
K
Kent Aguilar

Run this command on your terminal.

git reset HEAD~n

You can remove the last n commits from local repo e.g. HEAD~2. Proceed with force git push on your repository.

git push -f origin <branch>

Hope this helps!


C
Carlos Mafla

To preserve the branching and merging structure is important to use the --preserve-merges option when doing the rebase:

git rebase --preserve-merges -i HEAD^^

Note: it is not recommended to use --preserve-merges and --interactive together. See the BUGS section on rebase
M
M_R_K

For GitHub

Reset your commits (HARD) in your local repository

Create a new branch

Push the new branch

Delete OLD branch (Make new one as the default branch if you are deleting the master branch)


z
zOqvxf

Save your local changes first somewhere on the side ( backup )

You can browse your recent commits, then select a commit hash by clicking on "Copy the full SHA" button to send it to the clipboard.

If your last commit hash is, let's say g0834hg304gh3084gh ( for example )

You have to run:

git push origin +g0834hg304gh3084gh:master

Using the hash that you've copied earlier to make it the "HEAD" revision.

Add your desired local changes. Done ;)


M
Mohideen bin Mohammed

if you want to remove do interactive rebase,

git rebase -i HEAD~4

4 represents total number of commits to display count your commit andchange it accordingly

and delete commit you want from list...

save changes by Ctrl+X(ubuntu) or :wq(centos)

2nd method, do revert,

git revert 29f4a2 #your commit ID

this will revert specific commit


K
Keith

In GitHub Desktop you can just right click the commit and revert it, which will create a new commit that undoes the changes.

The accidental commit will still be in your history (which may be an issue if, for instance, you've accidentally commited an API key or password) but the code will be reverted.

This is the simplest and easiest option, the accepted answer is more comprehensive.


I have seen this cause problems with git flow. Example, you accidentally merge develop into master for a feature not to be released yet. If you just revert that merge, then when you merge master back down to develop it will remove the feature from there.
v
vivek_ganesan

It is not very good to re-write the history. If we use git revert <commit_id>, it creates a clean reverse-commit of the said commit id.

This way, the history is not re-written, instead, everyone knows that there has been a revert.


This older answer already says to user git revert.
That doesn't meet this requirement: "I want to revert my GitHub repository as it was before this commit"
C
Chiragkumar Thakar

Add/remove files to get things the way you want:

git rm classdir
git add sourcedir

Then amend the commit:

git commit --amend

The previous, erroneous commit will be edited to reflect the new index state - in other words, it'll be like you never made the mistake in the first place

Note that you should only do this if you haven't pushed yet. If you have pushed, then you'll just have to commit a fix normally.


This will create an additional revision ... with no changes