If you ever tried to inspect your remotes in git, you probably have
notice that there are two distinct URLs for the push and the fetch
operations. Today I realized why.
My repository is localized on a virtual machine other people (work colleagues) can reach. I clearly don't put my own SSH private key on the machine: in principle, anyone accessing the server could steal it and push stuff on my account. However, both me and the mentioned colleagues need to pull from a remote repository, so the following configuration was devised:
A fake user has Read permission on the target remote (GitHub);
The fake user declared a public key in GitHub, and its private counterpart lives in the
.sshdirectory on the virtual machine (shared among colleagues);The Remote's Fetch URL points to
git@github.com/....git.
Git configures both the Push and the Fetch URLs to the same endpoint, and that's quite reasonable.
The use-case
What if we were in a rush, we decided to do some local hack and we want to push it? We stated that the fake user is not allowed to push.
The fix
Set-up the Push URL to the HTTPs endpoint!
git remote set-url --push <remote> https://github.com/...
Now your git push will ask you the authentication.
The caveat
Are you using a two-factor authentication in GitHub? Create an Access Token and use it as a password
Happy hacking