Debug messages getting lost after adding seqlog

[webJose]

Hello! First and foremost: This is a very nice project, so thank you very much for your efforts.

Second: I'm just starting to learn Python. As I learn this language, I would like to be able to do the things that I do in other languages, and one of those things is structured logging + Seq. This is why I arrived to this library.

Third: My "problem", and please don't stress too much as it is a minor thing, is that some logging has disappeared once I started using seqlog, and I don't know why.

My Logging Setup Before seqlog

logging.basicConfig(
    level=logging.DEBUG, 
    format="%(asctime)s - [%(name)s] [%(levelname)s]: %(message)s",
)
logging.Formatter.converter = time.gmtime  # Use GMT/UTC instead of local timezone

My Logging Setup With seqlog

# This is the Python logging schema version (currently, only the value 1 is supported here).
version: 1

# Configure logging from scratch.
disable_existing_loggers: False

# Configure the root logger to use Seq
root:
  level: DEBUG
  handlers:
    - seq
    - console

# You can also configure non-root loggers.
# loggers:
#   another_logger:
#     propagate: False
#     level: DEBUG
#     handlers:
#       - seq
#       - console

handlers:
  # Log to STDOUT
  console:
    class: seqlog.structured_logging.ConsoleStructuredLogHandler
    formatter: seq

  # Log to Seq
  seq:
    class: seqlog.structured_logging.SeqLogHandler
    formatter: seq
    server_url: "http://localhost:5341"

formatters:
  seq:
    style: "{"
    format: "{asctime} - [{name}] [{levelname}]: {message}"

This is basically the example found in the documentation. I just added the format to the formatter, and I have played around with the value of disable_existing_loggers to see if this was the culprit of the log entries disappearing.

Which Logging is Missing

The debug logging that happens when issuing an HTTP request using requests. I used to see 2 DEBUG lines from one request I make to the Github API, and after adding seqlog, those lines have disappeared.

As clarification, the lines themselves are not important. What's important to me here is to understand if I did something wrong. I'm trying to learn Python logging the right way, and the right way says that if I use logging.getLogger(__name__), I adhere to the standards, and that a consequence of this standard is that all libraries can participate, and that participation has apparently been affected by my addition of seqlog.

So please, take this as an "academic" issue. The package itself seems to be working just fine.

Oh, side question: Is it possible to configure the printing of asctime as UTC?

[tintoy]

Hi - thanks for the detailed report!

I’ll take a look at this sometime today and see what I can find. 🙂

[tintoy]

In the meanwhile, could you try logging config using seqlog’s built-in config method (just to see if it’s something environmental, etc)?

https://seqlog.readthedocs.io/en/latest/usage.html#configure-logging-programmatically

[webJose]

Do you mean to do it programatically? Sure, I can try that. I'll let you know.

[webJose]

The missing log entries have reappeared!

This is how my configure_logging() function looks like now:

def configure_logging():
    global logger
    # seqlog.configure_from_file("log-config.yml")
    console_handler = seqlog.structured_logging.ConsoleStructuredLogHandler()
    console_handler.setFormatter(logging.Formatter(fmt="{asctime} - [{name}] [{levelname}]: {message}", style="{"))
    seqlog.log_to_seq(
        server_url="http://localhost:5341",
        level=logging.DEBUG,
        additional_handlers=[console_handler],
    )
    seqlog.set_global_log_properties(TransactionId=str(uuid.uuid4()), AppCode="CG", AppName="Curious GHeorge")
    # logging.basicConfig(
    #     level=logging.DEBUG, 
    #     format="%(asctime)s - [%(name)s] [%(levelname)s]: %(message)s",
    # )
    # logging.Formatter.converter = time.gmtime  # Use GMT/UTC instead of local timezone
    logger = logging.getLogger(__name__)

Can you assert that my "by code" configuration is equivalent to my YAML configuration? Here's my YAML:

# This is the Python logging schema version (currently, only the value 1 is supported here).
version: 1

# Configure logging from scratch.
disable_existing_loggers: False

# Configure the root logger to use Seq
root:
  level: DEBUG
  handlers:
    - seq
    - console

# You can also configure non-root loggers.
# loggers:
#   another_logger:
#     propagate: False
#     level: DEBUG
#     handlers:
#       - seq
#       - console

handlers:
  # Log to STDOUT
  console:
    class: seqlog.structured_logging.ConsoleStructuredLogHandler
    formatter: seq

  # Log to Seq
  seq:
    class: seqlog.structured_logging.SeqLogHandler
    formatter: seq
    server_url: "http://localhost:5341"

formatters:
  seq:
    style: "{"
    format: "{asctime} - [{name}] [{levelname}]: {message}"

I tried to code a perfect equivalent. Did I succeed?

For completeness, here are the log entries, in the console:

2025-07-25 19:47:08,173 - [__main__] [INFO]: Collecting data for topic: svelte
2025-07-25 19:47:08,176 - [urllib3.connectionpool] [DEBUG]: Starting new HTTPS connection (1): api.github.com:443
2025-07-25 19:47:09,283 - [urllib3.connectionpool] [DEBUG]: https://api.github.com:443 "GET /search/repositories?q=topic%3Asvelte&sort=updated&order=desc&per_page=50 HTTP/1.1" 200 None
2025-07-25 19:47:09,355 - [__main__] [DEBUG]: Response: 200 - {"total_count":12951,"incomplete_results":false,"items":[{"id":596825179,"node_id":"R_kgDOI5LUWw","n
2025-07-25 19:47:09,359 - [__main__] [DEBUG]: Database storage is enabled. (This feature is not implemented in this script.)
2025-07-25 19:47:09,359 - [__main__] [INFO]: Total repositories found for topic svelte: 12951

[tintoy]

Just out of curiosity, if you don’t configure a formatter, does the message template still match the message when sent to seq? Or are you only configuring the formatter on the console log handler (and not the seq log handler)?

[tintoy]

Just out of curiosity, if you don’t configure a formatter, does the message template still match the message when sent to seq? Or are you only configuring the formatter on the console log handler (and not the seq log handler)?

Sorry, posted this to the wrong issue (this is what happens when I reply to GitHub conversations on my phone 😂).

[tintoy]

Sorry, I forgot to add - yes, that by-code configuration is equivalent to the YAML (except for the Seq-specific bits that can’t be expressed in standard logging YAML).

[arisliang]

There seem to be some strange behavior with seqlog and urllib3 logs.
The 2 log can be seen in file and console handler, but not seq handler.

    urllib3_connectionpool_logger = logging.getLogger("urllib3.connectionpool")
    urllib3_connectionpool_logger.setLevel(logging.DEBUG)
    urllib3_connectionpool_logger.propagate = False
    urllib3_connectionpool_logger.addHandler(seq_handler)    # missing
    urllib3_connectionpool_logger.addHandler(file_handler)     # present
    urllib3_connectionpool_logger.addHandler(console_handler)    # present

urllib3.connectionpool - DEBUG - Starting new HTTP connection (1): localhost:5341
urllib3.connectionpool - DEBUG - http://localhost:5341 "POST /api/events/raw HTTP/1.1" 201 None

there seems to be no way to make seqlog to send these 2 log to seq, I wonder why?

[tintoy]

Does it work if you use the standard Seqlog setup method?

[arisliang]

both SeqLogHandler and log_to_seq setup seem to behave the same way, doesn't work.
Only 2 logs from urlib3 is missing, other loggers like __main__ can send to seq.

seq_handler = SeqLogHandler(
        server_url="http://localhost:5341",
        api_key=os.getenv("SEQ_API_KEY"),
    )

seq_handler = seqlog.log_to_seq(server_url="http://localhost:5341",
                                    api_key=os.getenv("SEQ_API_KEY"))

[tintoy]

I’m not sure what’s going on there, then, to be honest - the setup method replaces the default logger so it should just work; maybe try configuring a handler for logging-related errors using seqlog.set_callback_on_failure and see what exceptions (if any) get passed to it?

[tintoy]

Just out of curiosity, what do you see if you set propagate=True?

[arisliang]

failure callbacks is not triggered.
seqlog.set_callback_on_failure(lambda e: print(f"exception: {e}"))

by default, it is propagate=True for urllib3 logger; if switch that off, the 2 logs will be missing in other handlers (file, console) as well in addition to seq handler.