Skip to content

First deployment

1. Tell git-ftp where the server is

Run this once inside your repository:

git config git-ftp.url "ftp://example.com/public_html"
git config git-ftp.user "alice"
git config git-ftp.password "s3cret"

The settings land in .git/config, which is not part of the repository, so the password is not committed. If you would rather not store it at all, see credentials for -P, --password-command, ~/.netrc, the macOS keychain and environment variables.

2. Upload everything once

If the server is empty, or its contents should be replaced by what is in Git:

git ftp init

This uploads every tracked file that is not ignored and writes the commit id into .git-ftp.log on the server.

If the files are already on the server and match your working tree, do not upload them again — just record the commit:

git ftp catchup

Look before you leap

--dry-run prints what would be transferred and touches nothing:

git ftp init --dry-run

3. Deploy your changes

From now on, every deployment is one command:

echo "new content" >> index.txt
git commit index.txt -m "Add new content"
git ftp push

1 file to sync:
[1 of 1] Buffered for upload 'index.txt'.
Uploading ...
Last deployment changed from 1f2a3b4 to ded01b2.

git-ftp reads the commit recorded on the server, diffs it against HEAD and transfers exactly the files that were added, changed or deleted. Only when every upload succeeded does it update the log, so an interrupted deployment never claims a commit it did not finish.

4. Check what is deployed

git ftp show     # git show of the deployed commit
git ftp log      # git log starting at the deployed commit

Going back, and other branches

Because the state on the server is just a commit, moving around in history is ordinary Git work:

# deploy the state of three commits ago
git checkout HEAD~3
git ftp push

# deploy another branch without checking it out
git ftp push -b staging

# upload the difference between develop and master
git checkout develop
git ftp push --commit master

Where to go next