Provide a FileHealthCheckPublisher to support health check based on existence of a file

[scharnyw]

What would you like to be added:
Provide a FileHealthCheckPublisher which creates an empty temporary file (at a configured path) if health check passes, and removes the file if health check fails.

Why is this needed:
Using the existence of a file for health check purposes is a common practice in cloud-native scenarios. As an example, the Kubernetes documentation on health probes uses this approach for illustration purposes.

I can provide a PR to implement this if the idea is OK.

[sungam3r]

@scharnyw Do you mean FileHealthCheckPublisher in terms of IHealthCheckPublisher? I think I do not understand "publish" term here. I have read provided docs example. Indeed they create and remove file for demonstration purposes only.

[scharnyw]

@scharnyw Do you mean FileHealthCheckPublisher in terms of IHealthCheckPublisher? I think I do not understand "publish" term here. I have read provided docs example. Indeed they create and remove file for demonstration purposes only.

Yes, the "publish" action basically consists of creating a file (if health check passes) and deleting the file (if check fails).

I believe that using the existence of a file for health check purposes is fairly common, especially in the context of Kubernetes health probes. It's one way for non-web-facing applications to communicate their health status to Kubernetes, which is then used by Kubernetes for a wide range of purposes (reporting this to downstream, and making decisions about terminating an application, or network routing, application upgrade, etc.).

[sungam3r]

Yes, the "publish" action basically consists of creating a file (if health check passes) and deleting the file (if check fails).

It is not a job for IHealthCheckPublisher. IHealthCheckPublisher sends HealthCheckReport somewhere in one or another format.

I still don't understand what exactly you want to implement in this repi but you may go on and show implementation.

[fcastells]

This is a wild guess, because the description and documentation link are not clear, but maybe the proposal is to have a health check (not a publisher) that periodically:

1. checks if a file exists
2. if it doesn't exist, it returns unhealthy
3. if it exists, it deletes it and returns healthy

The only think I can think of is that there is a background process, external process, sidecar container or something else which detects when the file is deleted and creates it again. If that process hangs, the file won't be created and the health check will report that the application is unhealthy.