Run Exasol locally¶
Local deployments run on supported macOS, Linux, and Windows computers.
- On macOS, the launcher manages a virtual machine containing Podman and the database. An Apple Silicon Mac with at least 8 GB of memory is required.
- On Linux, the database runs directly in Podman on the host. Install Podman and make sure it is
available in
PATH. - On Windows, the database runs through Podman's default machine. Windows Package Manager
(
winget) and the prerequisites for a Podman machine are required.
After installing the launcher, run:
The launcher starts a single-node Exasol database. Run exasol info at any time to display
connection information.
The launcher selects and saves a database port during initialization, preferring 8563, and keeps
it stable across restarts. To choose one explicitly, run exasol install local --ports db:9563.
To change the port later, stop the deployment first, update the configuration, and restart it:
If another process claims the saved port while the deployment is stopped, startup reports the conflict and shows the same explicit and automatic replacement commands.
On macOS, the managed virtual machine uses about half of the detected host memory by default. An
explicitly configured memory value must be at least 4096 MB. Linux and Windows use host-managed
resources, so the virtual-machine sizing options do not apply. The initial local database credentials
are sys / exasol.
Windows host preparation¶
If Podman is missing, the launcher can offer to install it. It also creates or starts Podman's default machine when necessary, without changing the configuration of an existing machine. Because installing Podman changes shared host state, the launcher displays the exact command and asks for approval.
Use --auto-approve for unattended preparation. A command that cannot prompt refuses this host
change unless the option is present, unlike other confirmations, which an unattended run approves
automatically. Stopping or destroying an Exasol deployment leaves the shared Podman machine
running.
Next steps¶
- Connect to the database.
- Load sample data.
- Install a script language container to run UDFs.
- Manage the deployment lifecycle.
If the local deployment does not behave as expected, see Troubleshooting.