Cannot configure ldap in Superset 4.1.0 rc2

[ilsaloving]

Bug description

Apparently as part of the database driver purge, ldap drivers have also been removed and have to be reinstalled by the user.

There are currently 2 options, neither of which are ideal:

• Install using pip
  • While this works, it requires compiling from source. This requires numerous build environment dependencies and greatly complicates the setup
• Install from system packages (Debian 12)
  • The python3-ldap package provided by debian is tied to python 3.11. Since Superset 4.1.0 uses python 3.10, it does not find the library.

It would be much better if Superset restored the necessary ldap libraries by default.

Screenshots/recordings

No response

Superset version

master / latest-dev

Python version

3.10

Node version

Not applicable

Browser

Not applicable

Additional context

The following appears in the logs:

2024-10-08 16:25:15,031:ERROR:flask_appbuilder.security.manager:python-ldap library is not installed

Checklist

• I have searched Superset docs and Slack and didn't find a solution to my problem.
• I have searched the GitHub issue tracker and didn't find a similar bug report.
• I have checked Superset's logs for errors and if I found a relevant Python stacktrace, I included it here as text in the "additional context" section.

[dosubot]

A similar issue was discussed in a closed thread regarding the installation of the python-ldap package ^[1]. Here's a suggested solution:

1. Access the Superset container:

   docker exec -it <container_name> /bin/bash

2. Install necessary packages:

   apt-get update -y && apt-get install -y python3-dev libldap2-dev libsasl2-dev ldap-utils tox lcov valgrind

3. Install python-ldap:

   pip install python-ldap

4. Restart the container if needed:

   docker-compose -f docker-compose-non-dev.yml restart

This should help resolve the issue with LDAP configuration in Superset 4.1.0 rc2.

To continue talking to Dosu, mention @dosu.

[ilsaloving]

Is there a reason superset isn't standardizing on python 3.11? It would save a great deal of frustration, given that most of the distro-supplied python packages are built for 3.11.

[ilsaloving]

I'm going to have to give up on rc2. I keep running into one problem after another. I bit the bullet and built the ldap module. Now when Superset starts up, it seems to not know if it should be using postgres or SQLite. superset db upgrade generates numerous foreign key errors, the postgres database is only partially populated, and I get a 500 error when I try to access the UI.

I can understand wanting to slim down the build, but there has to be a better way than just stripping out all the drivers and forcing the burden onto users. This is a mess.

[sfirke]

Thanks for testing it out. How are you deploying - with the PyPI package? I ran into some similar problems when I tried out 4.1.0rc2, see my comment here: #29999 (comment) I'm deploying from Docker and for the purposes of testing I just switched to 4.1.0rc2-dev which fixed things. But maybe we can/should do better to support folks who want a production-ready image without building their own.

In addition to python-ldap I think other missing packages include the Postgres and SQLite drivers, openpyxl (for Excel uploads), pillow (?) for screenshots of alerts & reports.

[ilsaloving]

I was creating a docker container, using the superset provided docker container as the base and adding the required drivers.

And yes, it seems like all of that is missing. I think sqlite may be included by default because I didn't install that myself, but everything else is missing. Seems like too much has been stripped out.

I think a better option would be to provide a curated superset build container, where a user just have to provide a list of the desired drivers, and the build container will take care of making sure everything is installed.