* Simplify docker builds by consolidating all arch's into a single Dockerfile and using ARGS for various differences
* Introduce docker-compose based builds (build.yml) for simple management of the various args differences
Signed-off-by: Daniel <daniel@developerdan.com>
By default, dig will retry 2 times (for a total of 3 attempts) to get a response back. Each attempt defaults to 5 seconds. Before this change, a single docker healtcheck failure would really mean three failures and would take a total of 15 seconds before failing. By default, docker healthchecks will retry 3 times before considering a service unhealhy (with a 30 second interval). Combined with dig retries, this means it would take a total of 9 failed DNS responses before it considers the pihole to be unhealthy. Combining the retry between dig and docker, dig considers it a success if even 1/3 responses are recieved - and docker considers it a success if only 1/3 of those successes are successful. I'm not great at math - and order does make a difference - but I think that means as long as 1/9th of DNS queries are being answered - then docker thinks its healthy. Anyways, long story sort, dig doesn't need to have its own retry logic since docker already has a configuarable retry. I also disable recurse since the goal is to test this specific instance.
Also removed duplicate import statement.
Signed-off-by: Daniel <daniel@developerdan.com>
- Tox py3.7 + pipenv
- Python3 Dockerfile.py
- Dockerfile.py tags remote instead of just local image names now
- Circle.sh instead of in-line circle.yml script breakout
- probably other stuff I forgot
- Docker images build during the tests should hopefullly now be available at the deploy job workflow thanks to shared docker layers.
- Rename aarch64 to arm64 to reduce custom map