1.1 Installation

The pygitsync package is available from PyPI. Installing into a virtual environment is recommended.

python3 -m venv .venv; .venv/bin/pip install pygitsync

1.2 Credential management

One relatively straight-forward means of managing credentials for SSH based access to git repositories is using ssh-agent. Before starting the git-sync process, take the following steps.

eval $(ssh-agent)
ssh-add <path to keyfile to be used by pygitsync>
# type the passphrase for your SSH key as prompted
git-sync [options]

git-sync will now use the ssh agent session to acquire access to the SSH key.

Having a human type in a passphrase may not work very well for automation, so further effort will likely be needed to automatically acquire the passphrase from some kind of encrypted vault, and then apply that passphrase to the ssh agent session. A work-around may be to use an empty-passphrase SSH key, but this is not recommended at all for maintaining good security.

1.3 Getting started

Having just installed pygitsync you really just want to get a remote repo syncing, right? You need to create a configuration file - the default location is .pygitsync.yaml and it needs to look something like this. In this case we’re just syncing the default (main) branch of the pygitsync repo itself.

---
repo:
  pattern_type: branch
  pattern: main
  url: git@gitlab.com:ci-cd-devops/pygitsync.git
application:
  exception_sleep_seconds: 10
  is_daemon: false
  sleep_interval_seconds: 30

Using this configuration the remote repo will be pulled every 30 seconds to check for updates in the main branch. The pattern type “branch” means that the sync will always update to the latest HEAD pulled from the remote. Disabling the is_daemon setting means that the process will run indefinitely in the foreground (useful to run inside containers).

Running indefinitely means that the process will never exit for exceptions. If an exception occurs very early in the process then the exception handling loop could run very quickly, consuming large amounts of needless CPU. The exception_sleep_seconds setting puts a delay in that exception handling oop to prevent such a scenario from occurring.

The repo will be initially be cloned to the current working directory of the git-sync executable, or use the --working argument to specify an alternate location for the repo working directory.

Take a look at other configuration arguments using git-sync --help.