Failing to install superset locally on a windows 11 pc in a python virtual environment

[lgiammattei]

Bug description

Hi everyone, I'm not sure if this is the right place to report this and ask for help, or if my request is legitimate after all. If it isn't, I apologize in advance for your time and attention.
I'm trying to install superset locally on my PC to study it and see if it might be suitable for some work I have planned.
PC with Windows 11 fully updated and with Python 3.14
After creating a virtual Python environment, I proceed to install superset with the command
pip install apache-superset
but the installation always fails when trying to installa numpy 2.2.6 package:

Collecting numpy<2.3,>1.23.5 (from apache-superset)
  Downloading numpy-2.2.6.tar.gz (20.3 MB)
     ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 20.3/20.3 MB 3.9 MB/s  0:00:05
  Installing build dependencies ... done
  Getting requirements to build wheel ... done
  Installing backend dependencies ... done
  Preparing metadata (pyproject.toml) ... error
  error: subprocess-exited-with-error

  × Preparing metadata (pyproject.toml) did not run successfully.
  │ exit code: 1
  ╰─> [21 lines of output]
      + C:\superset_project\superset-env\Scripts\python.exe C:\Users\l.giammattei\AppData\Local\Temp\pip-install-b3mpgmja\numpy_d25dccc0223c4e099c44c7f3ab8507fd\vendored-meson\meson\meson.py setup C:\Users\l.giammattei\AppData\Local\Temp\pip-install-b3mpgmja\numpy_d25dccc0223c4e099c44c7f3ab8507fd C:\Users\l.giammattei\AppData\Local\Temp\pip-install-b3mpgmja\numpy_d25dccc0223c4e099c44c7f3ab8507fd\.mesonpy-m0kwdtfv -Dbuildtype=release -Db_ndebug=if-release -Db_vscrt=md --native-file=C:\Users\l.giammattei\AppData\Local\Temp\pip-install-b3mpgmja\numpy_d25dccc0223c4e099c44c7f3ab8507fd\.mesonpy-m0kwdtfv\meson-python-native-file.ini
      The Meson build system
      Version: 1.5.2
      Source dir: C:\Users\l.giammattei\AppData\Local\Temp\pip-install-b3mpgmja\numpy_d25dccc0223c4e099c44c7f3ab8507fd
      Build dir: C:\Users\l.giammattei\AppData\Local\Temp\pip-install-b3mpgmja\numpy_d25dccc0223c4e099c44c7f3ab8507fd\.mesonpy-m0kwdtfv
      Build type: native build
      Project name: NumPy
      Project version: 2.2.6
      WARNING: Failed to activate VS environment: Could not parse vswhere.exe output

      ..\meson.build:1:0: ERROR: Unknown compiler(s): [['icl'], ['cl'], ['cc'], ['gcc'], ['clang'], ['clang-cl'], ['pgcc']]
      The following exception(s) were encountered:
      Running `icl ""` gave "[WinError 2] Impossibile trovare il file specificato"
      Running `cl /?` gave "[WinError 2] Impossibile trovare il file specificato"
      Running `cc --version` gave "[WinError 2] Impossibile trovare il file specificato"
      Running `gcc --version` gave "[WinError 2] Impossibile trovare il file specificato"
      Running `clang --version` gave "[WinError 2] Impossibile trovare il file specificato"
      Running `clang-cl /?` gave "[WinError 2] Impossibile trovare il file specificato"
      Running `pgcc --version` gave "[WinError 2] Impossibile trovare il file specificato"

      A full log can be found at C:\Users\l.giammattei\AppData\Local\Temp\pip-install-b3mpgmja\numpy_d25dccc0223c4e099c44c7f3ab8507fd\.mesonpy-m0kwdtfv\meson-logs\meson-log.txt
      [end of output]

  note: This error originates from a subprocess, and is likely not a problem with pip.
error: metadata-generation-failed

× Encountered error while generating package metadata.
╰─> numpy

note: This is an issue with the package mentioned above, not pip.
hint: See above for details.

Is there a way to overcome this problem, or is it always better to install via Docker to have a test environment?

Screenshots/recordings

No response

Superset version

master / latest-dev

Python version

3.14

Node version

24.15

Browser

Firefox

Additional context

No response

Checklist

• [ x] I have searched Superset docs and Slack and didn't find a solution to my problem.

[dosubot[bot]]

Hi @lgiammattei! I'm Dosu and I'm helping the Apache Superset team.

The root cause of your error is that numpy 2.2.6 is being compiled from source on your machine, and its Meson-based build system can't find a C compiler. The error Unknown compiler(s): [['icl'], ['cl'], ['cc'], ['gcc'], ...] confirms no compatible C toolchain is installed. This happens because pre-built numpy wheels for Python 3.14 on Windows likely don't exist yet, so pip falls back to building from source.

Two important things to note:

1. Windows is not officially supported. The Superset docs state: "Superset does not have official support for Windows." [1]. You will likely encounter further issues even if you get past this numpy hurdle.

2. Python 3.14 is too new. Python 3.14 is currently in pre-release, and many packages (including numpy) don't ship pre-built wheels for it yet — which is exactly why pip is trying to compile from source. Superset itself hasn't been tested against 3.14 either. You should use Python 3.11 instead.

Your best options:

• Docker (recommended): Use Docker Desktop with the provided docker-compose.yml. This is the smoothest path and sidesteps all Windows/compiler issues. You can have a fully functional Superset running with a single docker compose up command [1].

• WSL2 (Windows Subsystem for Linux): Install WSL2 with Ubuntu, then follow the standard Linux pip install instructions with Python 3.11. This gives you a proper Linux environment without leaving Windows.

• Fix the native install (not recommended): If you still want a native Windows venv, install Microsoft Visual C++ Build Tools (which provides cl.exe) and downgrade to Python 3.11. That said, you're likely to hit more Windows-specific issues downstream.

For a quick study/evaluation environment, Docker Compose is genuinely the easiest path.

To reply, just mention @dosu.

---

Share context across your team and agents. Try Dosu.

[rusackas]

As stated, Windows is not officially supported. We'd welcome PRs/docs to MAKE it official, but meanwhile I'll move this to a Help discussion.

[dosubot[bot]]

Hi @lgiammattei! The root cause of your error is that numpy 2.2.6 is being compiled from source, and its Meson-based build system can't find a C compiler. The error Unknown compiler(s): [['icl'], ['cl'], ['cc'], ['gcc'], ...] confirms no compatible C toolchain is installed. This happens because pre-built numpy wheels for Python 3.14 on Windows don't exist yet, so pip falls back to building from source.

Two important things to note:

1. Windows is not officially supported. As the maintainer @rusackas confirmed, Superset doesn't have official Windows support [1], so you may hit further issues even if you get past this numpy hurdle.

2. Python 3.14 is too new. Python 3.14 is pre-release, and many packages (including numpy) don't yet ship pre-built wheels for it — which is exactly why pip tries to compile from source. Superset itself hasn't been tested against 3.14 either.

Your best options:

• Docker (recommended for your use case): Use Docker Desktop with the provided docker-compose-image-tag.yml to pull a pre-built release image [2]. This sidesteps all Windows/compiler issues and gets you a fully functional Superset with a single docker compose up command.

• WSL2: Install WSL2 with Ubuntu, then follow the standard Linux pip install with Python 3.11. This gives you a proper Linux environment without leaving Windows.

• Fix the native install (not recommended): Install Microsoft Visual C++ Build Tools (which provides cl.exe) and downgrade to Python 3.11. That said, you're likely to hit more Windows-specific issues downstream.

For a quick study/evaluation environment, Docker Compose is genuinely the easiest path.

To reply, just mention @dosu.

---

Share context across your team and agents. Try Dosu.

[lgiammattei]

I was able to get Apache Superset up and running with Docker and WSL2, and get it working by accessing data from a DB2 for IBM i (aka AS/400). I thought I'd leave a few notes here that might be helpful to anyone else who might be interested in a similar work environment (whether for development or production).

I installed Superset in "dev mode" via Docker Compose (built from source), with the IBM i Access ODBC driver integrated so you can connect to Db2 for IBM i through the sqlalchemy-ibmi SQLAlchemy dialect.

1. Prerequisites

• Windows 10/11 with WSL2 already installed and enabled (check with wsl --status in PowerShell)
• A Linux distribution installed in WSL2 (e.g. Ubuntu) — check with wsl -l -v, it must show VERSION 2
• Docker Desktop installed on Windows
• Access to an IBM ID to download the "IBM i Access Client Solutions" ODBC driver (Linux)
• Your Db2 for i connection details (host/IP, port, user, password, default library or *LOCAL)
• Git installed inside the WSL2 distribution (usually already present on Ubuntu; check with git --version)

---

2. Configure Docker Desktop to use WSL2

• Open Docker Desktop
• Go to Settings (gear icon) > Resources > WSL Integration
• Enable "Enable integration with my default WSL distro"
• Also enable the specific toggle next to your distro's name (e.g. Ubuntu)
• Click Apply & Restart
• Open a WSL2 terminal (from PowerShell: wsl, or open the "Ubuntu" app from the Start menu)
• Verify: docker --version should return a version number with no errors
• If you get permission denied ... docker.sock: close and reopen the WSL2 session (this
  alone is often enough — the integration adds your user to the docker group, but a new
  session is required for it to take effect). If it persists: sudo usermod -aG docker $USER,
  then close and reopen the terminal again

---

3. Get the IBM i Access ODBC driver for Linux

• Log in to the IBM site (search for IBM i Access Client Solutions download) with your IBM ID
• The file you need to download is called IBMiAccess_v1r1_LinuxAP.zip (not a .deb
  directly — it's a zip archive that contains both the .deb and the .rpm packages inside it, e.g.
  ibm-iaccess-1.1.0.29-1.0.amd64.deb)
• Extract the zip and locate the .deb file inside it
• Copy the extracted .deb file into WSL2 (no need to leave it on Windows): from the
  WSL2 shell, e.g. cp /mnt/c/Users/<youruser>/Downloads/ibm-iaccess-*.deb ~/

---

4. Clone the Superset repository inside WSL2 (not on C:)

• Open the WSL2 terminal
• Create a working folder in the native Linux filesystem: mkdir -p ~/dev && cd ~/dev
• Clone the official repository: git clone https://github.com/apache/superset.git
• Enter the folder: cd superset

5. Add the Db2 for i driver to the repository

• Create a folder for the driver: mkdir -p ~/dev/superset/docker/vendor

• Copy the downloaded .deb file: cp ~/ibm-iaccess-*.deb ~/dev/superset/docker/vendor/

• Open the Dockerfile at the root of the repo: ~/dev/superset/Dockerfile

• Find, in the dev stage (the one used by docker-compose.yml via
  target: ${SUPERSET_BUILD_TARGET:-dev}), the block that looks like:
  dockerfile # Debian libs needed for dev RUN /app/docker/apt-install.sh \ git \ pkg-config \ default-libmysqlclient-dev

• Replace it with (adding unixodbc/unixodbc-dev and installing the IBM driver):
  ```dockerfile
  # Debian libs needed for dev
  RUN /app/docker/apt-install.sh
  git
  pkg-config
  default-libmysqlclient-dev
  unixodbc
  unixodbc-dev

  # IBM i Access Client Solutions ODBC driver (for Db2 for i)
  COPY docker/vendor/ibm-iaccess-1.1.0.29-1.0.amd64.deb /tmp/ibm-iaccess.deb
  RUN /app/docker/apt-install.sh /tmp/ibm-iaccess.deb
  ```
  (adjust the exact `.deb` filename to match the version you downloaded)
  

• This block must come before the USER superset line in the same stage, otherwise
  the apt install will fail due to insufficient permissions

• Create (or edit) the file ~/dev/superset/docker/requirements-local.txt, adding:
  pyodbc sqlalchemy-ibmi
  (this file is read automatically at runtime by docker-bootstrap.sh and installed via
  pip on every startup of the Python containers — no rebuild needed if you edit it later)

---

6. Build and start the stack

• From the repo root (~/dev/superset), build the Python services involved:
  docker compose build superset superset-init superset-worker superset-worker-beat
• Start the whole stack:
  docker compose up -d
• Follow the initialization logs:
  docker compose logs -f superset-init
  Wait for Init Step 4/4 [Complete] to appear and for the container to exit successfully
  (inside native WSL2 this typically takes 1-3 minutes, not more than 15)
• Check the status of all services: docker compose ps — everything should show Up
  (or healthy for superset/superset-worker after a minute or two)

---

7. First login

• Open your browser at http://localhost (port 80, via nginx) or http://localhost:8088
  (the Superset service's direct port)
• Log in with the default credentials: admin / admin (customizable in docker/.env
  or docker/.env-local via the ADMIN_USERNAME/ADMIN_PASSWORD variables)

---

8. Configure the Db2 for i connection

• Go to Settings > Database Connections > + Database
• If available, pick the "IBM Db2 for i" type from the preset list; otherwise choose
  "Other"
• Enter as the SQLAlchemy URI:
  ibmi://USER:PASSWORD@IBM_I_HOST/LIBRARY_NAME
  (omit /LIBRARY_NAME to use *LOCAL, the user's default library)
• Click Test Connection — you should see "Connection looks good"
• Save the connection
• Verify with a test query in SQL Lab, e.g.:
  sql SELECT * FROM SCHEMANAME.TABLENAME FETCH FIRST 10 ROWS ONLY