88

Is there a way to set up a git repository, so that git pull defaults to one remote and git push defaults to another? I know I can set both by changing the value of the remote variable in branch section of .git/config, but how to do it for each direction separately?

Improve this question

7 Answers 7

Reset to default
113

Since Git version 1.7.0, you can set this with:

git remote set-url --push origin https://your.push.com/blah/
Improve this answer
Sign up to request clarification or add additional context in comments.

8 Comments

3 years later than the question, but this should be the new accepted answer!
@Kevlar Why? Accepting is used to mark not always "the best" answers, but the one that worked out for OP (read FAQ for more information). At a time of asking question above answer wouldn't work (wouldn't even existed), as git was way earlier than 1.8. Accepted answer however did worked out for OP. What reason to you find for changing OP's decision after three years?
@trejder Stack Overflow is also a place to serve useful answers to future visitors, who find a question via a search engine or whatever. It's valuable to have the currently-best answer appear first. I'm not saying that OP must change the accepted answer, but that it would be perfectly reasonable (and in my opinion a net positive) to do so.
@trejder this also doesn't answer the question that was asked.
1. fetch is what is relevant here, pull is simply fetch+merge 2. You have to do it the other way around apparently
|
40

Since Git 1.8.3, you can use the remote.pushDefault option to do exactly what you want (i.e. having different default remotes for pull and push). You can set the option just like any other; for example, to set it to the pushTarget remote, use

git config remote.pushDefault pushTarget

This option will have the following effect:

  • git pull will pull from the remote specified by the remote option in the relevant branch section in .git/config, while
  • git push will push to the remote specified by remote.pushDefault.

Note that you need to specify the name of a remote, not an URL. This makes this solution more flexible than the solution involving remote.<name>.pushurl, because (for example) you will still have tracking branches for both remotes. Whether you need or want this flexibility is up to you.

The release notes say this option was added specifically to support triangular workflows.

Improve this answer

3 Comments

Strange: I thought you are pushing to upstream only, and you don't know how many downstream repo are pulling from you: see stackoverflow.com/a/2749166/6309
@VonC Ah, yes, I see why it's confusing. I usually call the remote I want to pull from by default upstream because... well... it's upstream of my repository during those pulls. But the option is pushDefault, not pullDefault, so I used downstream as the name in the example. It's probably a better idea to call it defaultPushTarget ;)
@MvanGeest I agree. But I confirm you generally push to "upstream". There one (or very few and known) upstreams. But there can be many (and unknown) downstreams. Such is the DVCS (as in "distributed") universe.
29

For Git 1.6.4 and later, set remote.<name>.pushurl with git config.

One might use this to pull using the read-only https: protocol and push using an ssh-based protocol.

Say origin's url (remote.origin.url) is https://git.example.com/some/repo.git. It is read-only, but you have write access through the ssh-based ‘URL’ [email protected]:some/repo.git. Run the following command to effect pushing over the ssh-based protocol:

git config remote.origin.pushurl [email protected]:some/repo.git
Improve this answer

1 Comment

How to have push going to a say 'development' branch and pull coming from say 'production' branch in same remote+bare Git repository?
8

Thanks to MvanGeest for linking to the git 1.8.3 release notes. Those release notes say:

  • A triangular "pull from one place, push to another place" workflow is supported better by new remote.pushdefault (overrides the "origin" thing) and branch.*.pushremote (overrides the branch.*.remote) configuration variables.

I use such a triangular workflow all the time for open-source contributions. For example: I have my own GitHub fork of llvm/llvm-project, and I want to keep my own main branch up-to-date with the upstream's main. So I frequently git pull upstream main; it would be convenient if I could just type git pull instead. But, I don't want any chance that I might fat-finger git push<return> instead of git push origin main<return> and accidentally push to the upstream project's repo before I intended to! So, before today, my .git/config looked like this:

[remote "origin"]
        url = [email protected]:Quuxplusone/llvm-project
        fetch = +refs/heads/*:refs/remotes/origin/*
[remote "upstream"]
        url = [email protected]:llvm/llvm-project
        fetch = +refs/heads/*:refs/remotes/upstream/*
[branch "main"]
        merge = refs/heads/main
        remote = origin

Based on the release note quoted above, I've just changed my local repo's .git/config to this:

[remote "origin"]
        url = [email protected]:Quuxplusone/llvm-project
        fetch = +refs/heads/*:refs/remotes/origin/*
[remote "upstream"]
        url = [email protected]:llvm/llvm-project
        fetch = +refs/heads/*:refs/remotes/upstream/*
[branch "main"]
        merge = refs/heads/main
        remote = upstream
        pushremote = origin

Now I can do a simple git checkout main ; git pull to pull from upstream/main, and a simple git checkout main ; git push to push to origin/main. This is the "triangular workflow" I want.

Improve this answer

2 Comments

Thanks a lot! This works for me! For those who are new to this like me and wondering where/how to update .git/config, here it is: git config --edit
@ZoeL: I suppose git config --edit works, but FWIW, I just do nano .git/config. :) The config file for a given checkout/repo is always stored in a file named $ROOT_OF_MY_CHECKOUT/.git/config.
5

From what I can gather from the git config man page, the upstream repo is:

  • by default origin
  • set by branch.remote
  • always for both git pull/fetch and git pull

For a given branch, I don't see any way to have two separate remote by default.

Improve this answer

2 Comments

This seems to be true in practice for git v1.8.3.2, after trying both the git config remote... and git remote set-url...` answers, for a single branch or for an entire copy of a repo.
to reset the default remote back to origin for the current branch and push/pull to/from the matching branch name: git push --set-upstream origin <current_branch_name>
3

user392887's answer is mostly correct, but:

  1. You should prefer to use SSH. According to GitHub, "We strongly recommend using an SSH connection when interacting with GitHub. SSH keys are a way to identify trusted computers, without involving passwords."

  2. Anyone using RHEL/CentOS 6 will be using git 1.7.1 by default, which supports set-url.

So, the preferred solution for git 1.7.1. and later is:

git remote set-url --push origin [email protected]:username/somerepo.git
Improve this answer

1 Comment

How do you use SSH with libgit2?
2

In case you also came here looking for a per-branch solution, here it is from the manual:

branch.<name>.pushRemote

When on branch , it overrides branch..remote for pushing. It also overrides remote.pushDefault for pushing from branch . When you pull from one place (e.g. your upstream) and push to another place (e.g. your own publishing repository), you would want to set remote.pushDefault to specify the remote to push to for all branches, and use this option to override it for a specific branch.

Improve this answer

1 Comment

So at the command line that would be git config branch.my_branch.pushRemote my_push_remote

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.