unknown error: DevToolsActivePort file doesn't exist when running in Docker

Question

I am trying to run chrome using Python and selenium in a Docker container running alpine. It was running fine until one day when it started throwing the following error when I instantiated chrome.

Message: unknown error: Chrome failed to start: crashed.
(unknown error: DevToolsActivePort file doesn't exist)
(The process started from chrome location /usr/lib/chromium/chrome is no longer running, so ChromeDriver is assuming that Chrome has crashed.)

Code

Here is the code that creates this error:

from selenium import webdriver


def generate_plugin():
    pluginfile = 'proxy_auth_plugin.zip'

    # manifest_json, background_js same as https://mcmap.net/q/377877/-how-to-set-proxy-with-authentication-in-selenium-chromedriver-python

    with zipfile.ZipFile(pluginfile, 'w') as zp:
        zp.writestr("manifest.json", manifest_json)
        zp.writestr("background.js", background_js)
    return pluginfile


chrome_options = webdriver.ChromeOptions()
prefs = {"profile.default_content_setting_values.notifications": 2}
chrome_options.add_experimental_option("prefs", prefs)
plugin = generate_plugin()
chrome_options.add_extension(plugin)
chrome_options.add_argument('--no-sandbox')
chrome_options.add_argument('--disable-dev-shm-usage')
chrome_options.add_argument("--window-size=1920,1080")
chrome_options.add_argument("--remote-debugging-port=9222")
driver = webdriver.Chrome(chrome_options=chrome_options)
driver.get("https://google.com")

The full traceback of the exception it generates

Full traceback

driver = webdriver.Chrome(chrome_options=chrome_options)
|   File "/usr/local/lib/python3.6/site-packages/selenium/webdriver/chrome/webdriver.py", line 81, in __init__
|     desired_capabilities=desired_capabilities)
|   File "/usr/local/lib/python3.6/site-packages/selenium/webdriver/remote/webdriver.py", line 157, in __init__
|     self.start_session(capabilities, browser_profile)
|   File "/usr/local/lib/python3.6/site-packages/selenium/webdriver/remote/webdriver.py", line 252, in start_session
|     response = self.execute(Command.NEW_SESSION, parameters)
|   File "/usr/local/lib/python3.6/site-packages/selenium/webdriver/remote/webdriver.py", line 321, in execute
|     self.error_handler.check_response(response)
|   File "/usr/local/lib/python3.6/site-packages/selenium/webdriver/remote/errorhandler.py", line 242, in check_response
|     raise exception_class(message, screen, stacktrace)

Message: unknown error: Chrome failed to start: crashed.
(unknown error: DevToolsActivePort file doesn't exist)
(The process started from chrome location /usr/lib/chromium/chrome is no longer running, so ChromeDriver is assuming that Chrome has crashed.)

I am using pyvirtualdisplay for the display for webdriver but even if I add the --headless option in my code, it throws the same error.

Dockerfile

This is my Dockerfile:

FROM python:3.6-alpine

RUN apk add xvfb

# update apk repo
RUN echo "http://dl-4.alpinelinux.org/alpine/v3.7/main" >> /etc/apk/repositories && \
    echo "http://dl-4.alpinelinux.org/alpine/v3.7/community" >> /etc/apk/repositories

# install chromedriver
RUN apk update
RUN apk add chromium=93.0.4577.82-r2 --repository=http://dl-cdn.alpinelinux.org/alpine/v3.15/community chromium-chromedriver=93.0.4577.82-r2

/usr/lib/chromium # ./chrome --version
Chromium 93.0.4577.82
/usr/lib/chromium # ./chromedriver --version
ChromeDriver 93.0.4577.82 (e3a25d9b9e2d0b728e045ec87c0aa4942aa46e4e-refs/branch-heads/4577@{#1237})
/usr/lib/chromium #

Original issue

I was originally using RUN apk add chromium=86.0.4240.111-r0 --repository=http://dl-cdn.alpinelinux.org/alpine/v3.13/community chromium-chromedriver=86.0.4240.111-r0

It had been running fine until it started throwing that exception. So I went to my docker container and tried to run Chrome manually. It threw this error

Error relocating ./chrome: hb_subset_input_set_drop_hints: symbol not found
Error relocating ./chrome: hb_subset_input_set_retain_gids: symbol not found
Error relocating ./chrome: hb_subset: symbol not found

Once I upgraded the Chrome version to 93.0.4577.82, I can manually start both Chrome and Chromedriver, but running the code produces the same traceback.

OS and Docker version

I am using Macbook Pro M1 (2020) with MacOS 12.1

Docker version is the latest: 4.3.2 (72729)

Things I have tried

I have looked at almost all the other questions asked on StackOverflow and other forums but none of the advices have worked for me till now.

• I tried all the Chrome and Chromedriver versions available on alpine from 72.0.3626.121-r0 to the edge version

• I have even purged all the images, containers, and volumes from Docker and reinstalled it. It did not help either.

Update:

Passing the --headless argument makes it run. This makes me assume that the issue is somewhere related to using pyvirtualdisplay

Update 2

I missed to add that I was creating an extension zip file to add it to Chrome on the fly.

Here is what's happening now:

1. Pass the argument "--headless" and disable adding extension, it works completely fine.
2. Pass the argument "--headless" and enable adding extension, it throws, understandably selenium.common.exceptions.WebDriverException: Message: Message: unknown error: failed to wait for extension background page to load: chrome-extension://ilkejjjhcpjkcecgdkkdfaocminnclpo/_generated_background_page.html
3. Don't pass the --headless argument and disable extension, it works completely fine
4. Don't pass the --headless argument and enable extension, it throws the same exception as in point 2.

Quentin

Answer 1

I had the same issue on M1 with official selenium docker image.

But there are paragraph about how to fix it:

For experimental docker container images, which run on platforms such as the Mac M1 or Raspberry Pi, see the community driven repository hosted at seleniumhq-community/docker-seleniarm. These images are built for three separate architectures: linux/arm64 (aarch64), linux/arm/v7 (armhf), and linux/amd64.

Furthermore, these experimental container images are published on Seleniarm Docker Hub registry.

See issue #1076 for more information on these images.

If you're working on an Intel or AMD64 architecture, we recommend using the container images in this repository (SeleniumHQ/docker-selenium) instead of the experimental ones.

Amimia

While this link may answer the question, it is better to include the essential parts of the answer here and provide the link for reference. Link-only answers can become invalid if the linked page changes. - From Review – Paget

I switched to the seleniarm builds and I continue to get the same DevToolsActivePort file doesn't exist. I still haven't figured it out, but I think this coincidentally fixed it for you (perhaps the browser<=>webdriver version differences between the two images) – Amanda

I'm using seleniarm images, but I'm still getting the same error. So doesn't look like this answer solves the problem. – Crossley

Answer 2

I was getting this error after upgrading my chromedriver version to 86 and Python runtime to 3.8 from 3.6 on AWS Lambda (Amazon Linux 2) run in a docker container. I played whack a mole for hours with chrome/chromedriver starting issues.

Eventually I found this actively maintained min miplementation of python+selenium+docker. https://github.com/umihico/docker-selenium-lambda/ The setup in their Dockerfile and test.py chrome_options worked.

Benbena