geckodriver is an implementation of WebDriver, and WebDriver can be used for widely different purposes. How you invoke geckodriver largely depends on your use case.

Running Firefox in a container-based package

When Firefox is packaged inside a container (e.g. Snap, Flatpak), it may see a different filesystem to the host. This can affect access to the generated profile directory, which may result in a hang when starting Firefox.

This is known to affect launching the default Firefox shipped with Ubuntu 22.04+.

There are several workarounds available for this problem:

  • Do not use container-packaged Firefox builds with geckodriver. Instead download a Firefox release from and a geckodriver release from

  • Use a geckodriver that runs in the same container filesystem as the Firefox package. For example on Ubuntu /snap/bin/geckodriver will work with the default Firefox.

  • Set the --profile-root command line option to write the profile to a directory accessible to both Firefox and geckodriver, for example a non-hidden directory under $HOME.


If you are using geckodriver through Selenium, you must ensure that you have version 3.11 or greater. Because geckodriver implements the W3C WebDriver standard and not the same Selenium wire protocol older drivers are using, you may experience incompatibilities and migration problems when making the switch from FirefoxDriver to geckodriver.

Generally speaking, Selenium 3 enabled geckodriver as the default WebDriver implementation for Firefox. With the release of Firefox 47, FirefoxDriver had to be discontinued for its lack of support for the new multi-processing architecture in Gecko.

Selenium client bindings will pick up the geckodriver binary executable from your system’s PATH environmental variable unless you override it by setting the webdriver.gecko.driver Java VM system property:

System.setProperty("webdriver.gecko.driver", "/home/user/bin");

Or by passing it as a flag to the java(1) launcher:

% java -Dwebdriver.gecko.driver=/home/user/bin YourApplication

Your mileage with this approach may vary based on which programming language bindings you are using. It is in any case generally the case that geckodriver will be picked up if it is available on the system path. In a bash compatible shell, you can make other programs aware of its location by exporting or setting the PATH variable:

% export PATH=$PATH:/home/user/bin
% whereis geckodriver
geckodriver: /home/user/bin/geckodriver

On Window systems you can change the system path by right-clicking My Computer and choosing Properties. In the dialogue that appears, navigate AdvancedEnvironmental VariablesPath.

Or in the Windows console window:

% set PATH=%PATH%;C:\bin\geckodriver


Since geckodriver is a separate HTTP server that is a complete remote end implementation of WebDriver, it is possible to avoid using the Selenium remote server if you have no requirements to distribute processes across a matrix of systems.

Given a W3C WebDriver conforming client library (or local end) you may interact with the geckodriver HTTP server as if you were speaking to any Selenium server.

Using curl(1):

% geckodriver &
[1] 16010
% 1491834109194   geckodriver     INFO    Listening on
% curl -H 'Content-Type: application/json' -d '{"capabilities": {"alwaysMatch": {"acceptInsecureCerts": true}}}' http://localhost:4444/session
% curl -H 'Content-Type: application/json' -d '{"url": ""}' http://localhost:4444/session/d4605710-5a4e-4d64-a52a-778bb0c31e00/url
% curl http://localhost:4444/session/d4605710-5a4e-4d64-a52a-778bb0c31e00/url
% curl -X DELETE http://localhost:4444/session/d4605710-5a4e-4d64-a52a-778bb0c31e00
% fg

Using the Python wdclient library:

import webdriver

with webdriver.Session("", 4444) as session:
    session.url = ""
    print "The current URL is %s" % session.url

And to run:

% geckodriver &
[1] 16054
% python
1491835308354   geckodriver     INFO    Listening on
The current URL is
% fg