Enable Logging

dataversefs uses loguru for structured logging. Logging is disabled by default — importing the package produces no output. This guide shows how to turn it on and control what you see.

---

Basic usage

import sys
from loguru import logger

# Remove the default handler to avoid duplicate output in Jupyter/IPython
logger.remove()

# Add a handler and enable the dataversefs namespace
logger.add(sys.stdout, level="INFO", filter="dataversefs", colorize=False)
logger.enable("dataversefs")

After this, every routing table fetch is logged to stdout. To go back to silence:

logger.disable("dataversefs")

---

Log levels

Level	What is logged
INFO	Routing table fetch start and finish (with elapsed time); number of files registered
DEBUG	Every HEAD request to resolve a 303 redirect and every GET Range request for a file chunk

INFO gives you a timing overview with minimal output. DEBUG produces one line per network request — useful for diagnosing slow reads or unexpected fetches, but verbose during Zarr computation.

---

Jupyter note

In Jupyter, loguru's default handler writes to stderr, which means calling logger.add() without first calling logger.remove() creates a second handler and you see duplicate lines. Always call logger.remove() first.

Use colorize=False to avoid ANSI escape codes appearing as raw characters in notebook output cells.

# Correct pattern for Jupyter
logger.remove()
logger.add(sys.stdout, level="INFO", filter="dataversefs", colorize=False)
logger.enable("dataversefs")

---

Selective DEBUG output

DEBUG logs one line per HEAD and GET request. For a Zarr compute that touches thousands of chunks, this floods the notebook. A clean pattern is to enable DEBUG only for a specific block and then drop back to INFO:

# Temporarily enable DEBUG for one operation
logger.remove()
logger.add(sys.stdout, level="DEBUG", filter="dataversefs", colorize=False)
logger.enable("dataversefs")

ds.isel(time=0).load()  # watch every chunk request

logger.remove()
logger.add(sys.stdout, level="INFO", filter="dataversefs", colorize=False)