Improve logging for failed health checks

[danielmursa-dev]

While testing my application on Kubernetes, I configured the /_healthz/ endpoint as the startupProbe.

The probe kept failing, but nothing was logged to indicate which health check was responsible.
The only way I could identify the issue was by manually running the checks from the shell:

>>> from health_check.checks import *; import asyncio; res = asyncio.run(DNS().get_result()); print(res)
HealthCheckResult(..., error=ServiceUnavailable("DNS resolution failed..."), ...)

• get_result() already captures the exception and returns it in a HealthCheckResult, but the error is never logged

Would it be possible to log failed health checks (e.g. at warning or error level)?

This would make debugging much easier and allow us to immediately see which checks fails.

Maybe, add a logger here:

django-health-check/health_check/views.py

Line 120 in 2d9b82d

has_errors = any(result.error for result in self.results)

something similar to

django-health-check/health_check/management/commands/health_check.py

Line 145 in 2d9b82d

async def _run_checks_directly(self, path):

Thanks
Daniel

[codingjoe]

Hi @danielmursa-dev,

Thanks for reaching out. Django's request logs should show you the 500 response, but with those disabled, there isn't much output.

If you want, feel free to submit a PR for this one. I think I'd prefer a single logger for all checks. It makes it easier to configure the logging. The log level would be arguable, it would start with WARNING, since we are expecting errors and are handling them.

Cheers!
Joe

[tomatotomata]

I will take this on as the requested focused change: one configurable shared logger at WARNING for failed checks, with regression coverage. I confirmed there is no competing PR and will keep the scope limited to the failure-reporting path.