tailcove 0.1.4

Documentation  ·  Get started free  ·  View on PyPI  ·  Report a bug

Send Python logs to Tailcove with the standard logging module.

Install

pip install tailcove

Requires Python 3.9 or newer.

Quick start

1. Get a token. Create a source in your Tailcove dashboard and copy its ingest token. It is shown only once, so save it somewhere safe (an environment variable is ideal).

2. Add the handler.

import logging
import os

from tailcove import TailcoveHandler

logger = logging.getLogger("myapp")
logger.addHandler(TailcoveHandler(token=os.environ["TAILCOVE_TOKEN"]))
logger.setLevel(logging.INFO)

logger.info("user signed up", extra={"user_id": 42})
logger.error("payment failed", extra={"order_id": 99, "reason": "card_declined"})

3. Open Tailcove. Your records appear in search and live tail within a couple of seconds.

Buffered records are flushed when your program exits normally, because Python's own logging.shutdown() closes every handler.

Options

Records are sent once batch_size is reached or batch_seconds has passed, whichever comes first. They are never sent one HTTP request per log call. To send whatever is buffered right now, call handler.flush(). It blocks until the background thread confirms.

Structured context

Anything you pass in extra={...} is shipped as the line's context, with the same keys:

logger.info("order placed", extra={"order_id": 99, "total_cents": 4200})

Exceptions

Log with exc_info=True, or use logger.exception(...) inside an except block, and the exception is sent as structured data and not as a flattened string:

try:
    charge_card(order)
except CardDeclined:
    logger.error("payment failed", exc_info=True)

"exception": {
    "class": "billing.errors.CardDeclined",
    "message": "insufficient funds",
    "location": "/app/billing/charge.py:42",
    "trace": [
        "/app/billing/charge.py:42 in charge_card",
        "/app/orders/service.py:17 in place_order"
    ]
}

location is where the exception was raised, and trace is limited to trace_limit frames. It's the same shape the Laravel, Node and browser libraries use, so the same error groups into one issue whichever app reported it.

Frameworks

Django

Add the handler to LOGGING in settings.py:

LOGGING = {
    "version": 1,
    "disable_existing_loggers": False,
    "handlers": {
        "tailcove": {
            "class": "tailcove.TailcoveHandler",
            "token": env("TAILCOVE_TOKEN"),
        },
    },
    "root": {
        "handlers": ["console", "tailcove"],
        "level": "INFO",
    },
    "loggers": {
        "django.request": {
            "handlers": ["tailcove"],
            "level": "ERROR",
            "propagate": True,
        },
    },
}

Flask

Attach the handler to the app logger, or to the root logger if you want every logging.getLogger(__name__) in the app to ship too:

from flask import Flask
from tailcove import TailcoveHandler

app = Flask(__name__)
app.logger.addHandler(TailcoveHandler(token=app.config["TAILCOVE_TOKEN"]))
app.logger.setLevel("INFO")


@app.errorhandler(Exception)
def handle_exception(exc):
    app.logger.error("unhandled exception", exc_info=exc)
    raise

When delivery fails

Network errors and non-2xx responses never raise into your code. emit() follows the standard Handler convention and routes its own failures through handleError(). Errors on the background thread are caught and printed to stderr. Logging is best-effort by design.

Help and feedback

• Docs: tailcove.dev/docs/sending-logs/python
• Bugs and feature requests: open an issue
• Security problems: please email security@tailcove.dev instead of opening a public issue.

Documentation  ·  Get started free  ·  Report a bug  ·  MIT License

_{More Tailcove libraries: Laravel  ·  Node.js  ·  Browser}