Monitoring: health, logs, reporting

The health endpoint

GET /health is the endpoint for liveness and readiness checks: public, without a key, without consuming quota — and therefore the right probe for a service or container health check.

{ "status": "Healthy", "schemas": "ok", "db": "ok", "version": "1.0.42", "buildDate": "…" }

Response	Meaning
200 + "Healthy"	the service is up and both checks passed
503 + "Degraded"	at least one check failed; schemas or db then reads "missing"

What exactly is checked — and why only two things

Each of them because its failure would be silent:

Field	Check	Why
schemas	is the EN 16931 code-list file present in the output directory?	If it is missing, the code-list checks quietly pass — an unvalidated invoice would look like a validated one
db	is the SQLite database present, readable and migrated?	on Linux this also covers the native libraries

version and buildDate belong in every support enquiry

They are in the response deliberately, now that the product runs on customer servers: the installer uses them to confirm that the new build really started. Invoiceflair.Api.exe version returns the same. A build without a version stamp reports "dev", and buildDate is then null.

Logs

Location	Content
<target directory>\logs\invoice-api-YYYYMMDD.log	the application log, rolled daily, 30 files retained
%TEMP%\Setup Log*.txt	the installer’s log
Service control	sc query Invoiceflair, sc stop …, sc start …

The minimum level is Information; the framework categories are dampened.

That is deliberate, not parsimony

At Information, ASP.NET logs every request including the full URL — and a sign-in link carries a still-redeemable token there. The /auth paths are additionally excluded from the request log. Turning the level up logs sign-in tokens along with everything else.

Usage reporting

Invoiceflair.Api.exe usage --days 7

This answers the operational questions from the log: how often each tool ran, how many calls failed, how many addresses were involved, and which API key consumed how much.

The usage log is not viewable through the interface

It sits in the installation’s file system. The activity log in the account area is something else — it shows a tenant’s administrative events, not tool counts.

What is worth monitoring

Check	How
the service is running	/health against the HTTP port and localhost — not through the proxy, otherwise a fresh certificate becomes monitoring’s first stumbling block
space in the data directory	the archive and the log grow
the log for warnings	in particular the warning about a missing public address behind a proxy