Awesome
HAproxy-API
This application is a companion to Sidecar that allows you to run HAproxy and Sidecar in individual Docker containers that can be deployed separately. This has the advantage of not taking down HAproxy while redeploying Sidecar. It is not a general purpose API and relies heavily on the encoding of the state used by Sidecar.
This application will manage HAproxy by either running it or restarting it after templating out a configuration from the provided Sidecar state.
Generally you will run this on the same host as the Sidecar instance it is subscribed to. There might be situations like frontend gateways where it makes sense to subscribe it to a remote Sidecar instance. Its sole job is to receive state from Sidecar and configure/manage HAproxy.
You may instead also run HAproxy-API in a follower mode where it can manage a
local HAproxy while subscribing to a remote Sidecar. This is particularly
useful for local development work. It is probably not the best configuration
for production. This mode is triggered either by supplying a hostname:port
combination with the -F flag or by supplying the HAPROXY_API_FOLLOW
environment variable where the vlaue is hostname:port of a remote or local
Sidecar.
Configuration
There is a sample configuration in haproxy-api.toml,
including comments on what each item means. Generally you need to know the
remote URL of a Sidecar instance you will connect to. If you run in host
networking mode this will be http://localhost:7777/state.json if you have a
custom Docker network, then it will be somethign like http://<name of Sidecar container>:7777/state.json. This URL is used for bootstrapping the container
because we don't have any stored state. It will then receive updates from
Sidecar itself (if you've configured it to publish them), and won't make any
further calls to Sidecar.
Health Checking
haproxy-api can be health checked by sending a GET request to the /health
endpoint. This in turn simply checks to make sure that HAproxy is currently
running by shelling out to bash, ps, and grep.
Contributing
Contributions are more than welcome. Bug reports with specific reproduction steps are great. If you have a code contribution you'd like to make, open a pull request with suggested code.
Pull requests should:
- Clearly state their intent in the title
- Have a description that explains the need for the changes
- Include tests!
- Not break the public API
Ping us to let us know you're working on something interesting by opening a GitHub Issue on the project.