Segmentation fault using parsers in Docker

[hawkeye-bot]

Hi,

I've made a small script that does nothing more basically than call Parser().parse_one(fp=path, fn=filename, save_history=False). It runs fine with unicorn from the terminal on a Mac. When I package it in docker however, I get a segmentation fault. I've already included ulimit -s 32768000 in the docker file, but still getting the segmentation fault. Does anybody have an idea about what could be the cause, or how to investigate?

[JessBerl]

Hi,

We discovered a conflict with llmware and the latest version of the mongoc driver that was released 2 days ago and this can cause a segmentation fault. To see if this is what you are encountering, please downgrade the mongo-c-driver to v1.24. Or we'll be releasing an llmware update today that addresses the issue.

Or, if you share your docker file, we can take a look to see what else might be happening.

[hawkeye-bot]

Hi @JessBerl,

thanks for your fast response. Good to hear that it's actually something that's broken, can you let me know/indicate here when a fixed version is released and the corresponding version nr?

This is the docker file I use which produces the error:

FROM python:3.10-bullseye
WORKDIR /abc
COPY requirements.txt /abc/requirements.txt
RUN pip install --no-cache-dir --upgrade -r /abc/requirements.txt
RUN ulimit -s 32768000
COPY ./de /abc/de/
CMD ["uvicorn", "de.app:app", "--host", "0.0.0.0", "--port", "80"]

[JessBerl]

@hawkeye-bot - thanks for sharing you docker file.

A quick update that llmware v0.1.6 is now available, but it's likely not the source of the seg fault you are seeing with docker. @turnham will be following up with you to share some suggestions for your docker file.

[turnham]

Hi @hawkeye-bot Thanks for trying us out and the feedback! The issue in this case, is that setting the ulimit in the Dockerfile is not sufficient and it needs to be set when running the container. I've attached a complete working example, but the key part is the --ulimit setting you'll see in docker-run.sh:

docker run --rm --ulimit stack=32768000:32768000 --name docker-test docker-test:latest 

With that set the parsing can succeed:

> ./docker-run.sh

 > Parsing documents in /abc/documents

First block found on page 1:
EXECUTIVE EMPLOYMENT AGREEMENT    THIS EXECUTIVE EMPLOYMENT AGREEMENT (this Agreement) is entered into this 2nd day of   April, 2012, by and between Aphrodite Apollo (   Executive) and TestCo Software, Inc. (the Company or    Employer), and shall become effective upon Executive's commencement of employment (the  Effective Date)   which is expected to commence on April 16, 2012. The Company and Executive agree that unless Executive has   commenced employment with the Company as of April 16, 2012 (or such later date as agreed by each of the   Company and Executive) this Agreement shall be null and void and of no further effect.

docker-test.zip

[hawkeye-bot]

Thanks a lot @turnham , I can confirm this solves my issue. Might be good to take this up on the readme file where the ulimit is mentioned that this is the way to do it for docker

[turnham]

Yes we'll do that for sure. Thanks again