Logging

Logging is the practice of emitting log messages that describe what is happening in the system, or “events” in short. Log events can have varying degrees of severity, such as debug, info, warning, error or even critical. By default, Referentielijsten emits logs with level info and higher.

A collection of log events with a correlation ID (like a request or trace ID) allow one to reconstruct the chain of events that took place which lead to a particular outcome.

Referentielijsten emits structured logs in JSON format (unless explicitly configured otherwise), which should make log aggregation and analysis easier.

We try to keep a consistent log message structure, where the following keys are (usually) present:

source: The component in the application stack that produced the log entry. Typical values are uwsgi and app.
level: The severity level of the log message. One of debug, info, warning, error or critical.
timestamp: The moment when the log entry was produced, a string in ISO-8601 format. Most of the logs have microsecond precision, but some of them are limited to second precision.
event: The event that occurred, e.g. request_started or spawned worker (PID 123). This gives the semantic meaning to the log entry.

Other keys that frequently occur are:

request_id: Present for application logs emitted during an HTTP request, makes it possible to correlate multiple log entries for a single request. Not available in logs emitted by background tasks or logs emitted before/after the Referentielijsten app.

Tip

Certain log aggregation solutions require you to configure “labels” to extract for efficient querying. You can use the above summary of log context keys to configure this according to your needs.

Note

We can not 100% guarantee that every log message will always be JSON due to limitations in third party software/packages that we use. Most (if not all) log aggregation technologies support handling both structured and unstructured logs.

Logging

Format

Referentielijsten emits structured logs (using structlog). A log line can be formatted like this:

{
    "id": "1",
    "code": "ABC-1",
    "naam": "Tabel 1",
    "event": "tabel_created",
    "user_id": null,
    "request_id": "2f9e9a5b-d549-4faa-a411-594aa8a52eee",
    "timestamp": "2025-05-19T14:09:20.339166Z",
    "logger": "referentielijsten.api.signals",
    "level": "info"
}

Each log line will contain an event type, a timestamp and a level. Dependent on your configured LOG_LEVEL (see Environment configuration reference for more information), only log lines with of that level or higher will be emitted.

Referentielijsten log events

Below is the list of logging event types that referentielijsten can emit. In addition to the mentioned context variables, these events will also have the request bound metadata described in the django-structlog documentation.

API

tabel_created: created an Tabel via the API. Additional context: id, code, naam.
tabel_updated: updated an Tabel via the API. Additional context: id, code, naam.
tabel_deleted: deleted an Tabel via the API. Additional context: id, code, naam.
item_created: created an Item via the API. Additional context: id, code, naam.
item_updated: updated an Item via the API. Additional context: id, code, naam..
item_deleted: deleted an Item via the API. Additional context: id, code, naam..

Exceptions

Handled exceptions follow a standardized JSON format to ensure consistency and improve error tracking. Most fields are standard and include: title, code, status, event, source, user_id, request_id, exception_id, timestamp, logger and level.

A new field invalid_params has been added to provide detailed information about which input parameters caused the error in API calls.

name: name of the invalid parameter

code: specific error code

reason: explanation/message of the error

{
    "title": "'Invalid input.'",
    "code": "invalid",
    "status": 400,
    "invalid_params": [
        {
        "name": "tabel__code",
        "code": "required",
        "reason": "Dit veld is vereist."
        }
    ],
    "event": "api.handled_exception",
    "user_id": null,
    "exception_id": "e0207775-bfc7-4ba6-b0f5-489e6e93030e",
    "request_id": "c75121fb-ac9f-474b-9ba9-ef7272add573",
    "timestamp": "2025-10-03T12:09:49.156009Z",
    "logger": "vng_api_common.exception_handling",
    "level": "error"
}

Uncaught exceptions that occur via the API are logged as api.uncaught_exception events and contain the traceback of the exception.

{
    "message": "division by zero",
    "event": "api.uncaught_exception",
    "user_id": null,
    "source": "app",
    "request_id": "b1c4db0e-da7f-4a9e-a9fe-efbb6420a907",
    "timestamp": "2025-10-03T12:09:04.238909Z",
    "logger": "vng_api_common.views",
    "level": "error",
    "exception": "Traceback (most recent call last):\n  File \"/usr/local/lib/python3.12/site-packages/rest_framework/views.py\", line 506, in dispatch\n    response = handler(request, *args, **kwargs)\n               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n  File \"/app/src/referentielijsten/api/viewset.py\", line 63, in list\n    1 / 0\n    ~~^~~\nZeroDivisionError: division by zero"
}

Third party library events

For more information about log events emitted by third party libraries, refer to the documentation for that particular library

Django (via django-structlog)