Log successful kube-probe requests at Trace lvl

[dnwe]

When deployed and running in Kubernetes, step-ca will output a log statement every 5 seconds at INFO level from the ongoing readiness/liveness probe requests. This PR changes the behaviour so that successful kube-probe requests will only be logged at Trace level by default.

Fixes #420

[codecov-io]

Codecov Report

Merging #421 into master will decrease coverage by 0.82%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master     #421      +/-   ##
==========================================
- Coverage   76.33%   75.50%   -0.83%     
==========================================
  Files          92       97       +5     
  Lines        9109     9271     +162     
==========================================
+ Hits         6953     7000      +47     
- Misses       1773     1880     +107     
- Partials      383      391       +8     

Impacted Files	Coverage Δ
logging/handler.go	66.66% <100.00%> (ø)
logging/logger.go	0.00% <0.00%> (ø)
logging/responselogger.go	50.00% <0.00%> (ø)
logging/context.go	0.00% <0.00%> (ø)
logging/clf.go	0.00% <0.00%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 98a5aa5...34fbb28. Read the comment docs.

[maraino]

I understand your issue, the main problem is that right now we don't have a way to define the log level. And those logs will be lost, not only that, I can change my user agent to hide my activity.

I have to discuss this internally.

[maraino]

The logging package is gonna be replaced by this package, and I've added an issue to be able to configure the debug level and to be able to hide liveness/readiness probes (smallstep/logging#2)

[dopey]

Hey @dnwe thanks again for the submission! We spent some time discussing this morning. As @maraino we're planning on switching up the logger soon which should make this type of thing easier. But we're not exactly sure when that will happen. Prioritization is hard 🤕

With regards to this submission we'd be open to pulling it in to fix the short term need but we'd like to see a few changes:

1. Condition on the Path /health rather than the user agent.
2. Put the whole section in a IF clause that's triggered by an environment variable e.g. STEP_*. This way users will need to explicitly ask to remove the logs.

If you're open to making those changes, we'll happily pull this in. With the caveat that sometime in the future this implementation will likely change.

[dnwe]

@dopey sure those recommended changes make sense 👍 I'll take a look at updating the PR

[dnwe]

@dopey apologies I'd forgotten about this open PR, I whipped up a draft replacement based on the discussion above, but wasn't sure what environment-variable naming you were looking for. I proposed STEP_LOGGER_ONLY_TRACE_HEALTH_ENDPOINT but feel free to suggest changes if you had a preferred one in mind

[dopey]

lgtm

[dopey]

Hey @dnwe thanks for the PR! Merging.