GQLAlchemy install fails because of mgclient dependencies

[antejavor]

GQL is hard to install for usage on AWS Lambdas because of the dependencies, probably because OS environment is striped compared to conventional Linux images, so it probably requires a bit more steps.

The user tried to create an identical environment in custom docker image, running amazonlinux:2 :
pip3 install -t python gqlalchamy

First, the typical issues with OpenSSL not being visible popped out:

CMake Error at /usr/share/cmake3/Modules/FindPackageHandleStandardArgs.cmake:137 (message):
  Could NOT find OpenSSL, try to set the path to OpenSSL root folder in the
  system variable OPENSSL_ROOT_DIR (missing: OPENSSL_CRYPTO_LIBRARY) (found
  version "1.0.2k") 

After installing OpenSSL-static, zlib, and zlib-static, the OpenSSL issue disappeared, and a strange linker failure was hit:

The user moved to the Neo4j Python driver.

Full context: https://discord.com/channels/842007348272169002/842338690109931520/1146269544118227054

[katarinasupe]

Noticed similar when installing GQLAlchemy on Windows with Python 3.11 -> memgraph/gqlalchemy#290

[ad1arsh]

I would like to add that I am also facing the same issue on an EC2 instance with Amazon linux and a conda environment:

CMake Error at /usr/share/cmake3/Modules/FindPackageHandleStandardArgs.cmake:164 (message): Could NOT find OpenSSL, try to set the path to OpenSSL root folder in the system variable OPENSSL_ROOT_DIR (missing: OPENSSL_CRYPTO_LIBRARY) (found version "3.0.8")

However, after installing OpenSSL-static, zlib, and zlib-static, pymgclient could be installed. However, I am hit with a different error:

I feel this is again due to a mismatch in the used OpenSSL package.

I had used this command previously to explicitly define which OpenSSL to use in my conda environment:
CC=clang CXX=clang++ OPENSSL_INCLUDE_DIR=/opt/conda/envs/pytorch/include OPENSSL_ROOT_DIR=/opt/conda/envs/pytorch OPENSSL_LIBRARIES="/opt/conda/envs/pytorch/lib/libssl.so;/opt/conda/envs/pytorch/lib/libcrypto.so" pip install pymgclient --no-cache

Some guidance would be much appreciated on this, as a large part of the workflow relies on pymgclient (I was on an ubuntu server before migrating to AWS), and I would rather avoid the work needed to modify this for the neo4j driver.

These are my specs:

NAME="Amazon Linux"
VERSION="2"
ID="amzn"
ID_LIKE="centos rhel fedora"
VERSION_ID="2"
PRETTY_NAME="Amazon Linux 2"
ANSI_COLOR="0;33"
CPE_NAME="cpe:2.3⭕amazon:amazon_linux:2"
HOME_URL="https://amazonlinux.com/"
SUPPORT_END="2025-06-30"

[katarinasupe]

Additional context: import is not working using venv on Amazon Linux 2, so this is not only related to Conda.
Error when trying to import mgclient:
ImportError: /home/ec2-user/biomex/venv/lib/python3.10/site-packages/mgclient.cpython-310-x86_64-linux-gnu.so: undefined symbol: krb5_free_context

cc @ad1arsh

[katarinasupe]

Here are the steps I recommended @ad1arsh to try:

1. Uninstall pymgclient anywhere you have it with pip uninstall pymgclient
2. Create a new clean conda environment and run pip install pymgclient --no-cache (without any prerequisites installed)
3. Try importing mgclient

User tried this with both Python 3.9 and Python 3.10 envs and it produces the same error as before:
ImportError: /opt/conda/envs/newenv/lib/python3.10/site-packages/mgclient.cpython-310-x86_64-linux-gnu.so: undefined symbol: krb5_free_context

Additional context
When creating a new env, this is what gets installed:

_libgcc_mutex      conda-forge/linux-64::_libgcc_mutex-0.1-conda_forge 
  _openmp_mutex      conda-forge/linux-64::_openmp_mutex-4.5-2_gnu 
  bzip2              conda-forge/linux-64::bzip2-1.0.8-hd590300_5 
  ca-certificates    conda-forge/linux-64::ca-certificates-2023.11.17-hbcca054_0 
  ld_impl_linux-64   conda-forge/linux-64::ld_impl_linux-64-2.40-h41732ed_0 
  libffi             conda-forge/linux-64::libffi-3.4.2-h7f98852_5 
  libgcc-ng          conda-forge/linux-64::libgcc-ng-13.2.0-h807b86a_3 
  libgomp            conda-forge/linux-64::libgomp-13.2.0-h807b86a_3 
  libnsl             conda-forge/linux-64::libnsl-2.0.1-hd590300_0 
  libsqlite          conda-forge/linux-64::libsqlite-3.44.2-h2797004_0 
  libuuid            conda-forge/linux-64::libuuid-2.38.1-h0b41bf4_0 
  libzlib            conda-forge/linux-64::libzlib-1.2.13-hd590300_5 
  ncurses            conda-forge/linux-64::ncurses-6.4-h59595ed_2 
  openssl            conda-forge/linux-64::openssl-3.2.0-hd590300_1 
  pip                conda-forge/noarch::pip-23.3.2-pyhd8ed1ab_0 
  python             conda-forge/linux-64::python-3.10.13-hd12c33a_0_cpython 
  readline           conda-forge/linux-64::readline-8.2-h8228510_1 
  setuptools         conda-forge/noarch::setuptools-68.2.2-pyhd8ed1ab_0 
  tk                 conda-forge/linux-64::tk-8.6.13-noxft_h4845f30_101 
  tzdata             conda-forge/noarch::tzdata-2023c-h71feb2d_0 
  wheel              conda-forge/noarch::wheel-0.42.0-pyhd8ed1ab_0 
  xz                 conda-forge/linux-64::xz-5.2.6-h166bdaf_0

It seems that other similar issues were resolved with the correct version of pymgclient being used, but on Amazon Linux 2 this is potentially not a fix.

cc @as51340 @antoniofilipovic for awareness