[EC2] What to do when selenium is stuck and processing does not proceed

[EC2] What to do when selenium is stuck and processing does not proceed

--When using selenium on EC2, it takes a long time to get the data. ――It went smoothly at first, but as the execution is repeated, the processing becomes heavier. --If you wait for a long time, you will finally get an error that you cannot reach chrome.

error

$python3 filename.py

selenium.common.exceptions.WebDriverException: 
Message: chrome not reachable

Cause

Multiple processes and drivers for the chrome browser have been launched. (Because it is headless, it cannot be visually confirmed ...)

approach

-Stop the running chrome browser & driver. -Write object.quit () at the end of the file.

Remedy procedure

1. 1. Check if multiple chrome drivers are launched.

python

#Show all running processes including "chrome"
$ ps aux | grep chrome


#If you see a lot of descriptions like the one below, multiple webdrivers are running.
Username pts/0    Sl   04:09   0:00 
chromedriver --port=49671

#If you see a lot of statements like the one below, multiple browser processes are running.
Username pts/0    S    04:32   0:00
/opt/google/chrome/chrome --type=broker

• Chrome is running multiple processes even when displaying one browser.

ps aux

--ps command: Display the command being executed --aux is an option --a: All user processes --u: Display user name and start time --x: Shows processes without control terminals

• Process Status -｜ Often used as a set with grep

-"| (Pipe)": Added command --grep text: Show only processes that include text

• Global Regular Expression Print

python

killall chrome
killall webdriver

Check the running file again.

python

$ ps aux | grep chrome


#Only the following was left and the others were finished.
Username pts/0    S+   04:35   0:00 
grep --color=auto chrome

python

#Assuming the following description
browser = webdriver.Chrome(options=options)

browser.quit()

If the variable for starting webdriver is described in another description such as driver, match it. Example: driver.quit ()

Supplement

In another article, there was something like stopping the sandbox and adding --no-sandbox as an option, but in my case, the sandbox was not started and was unnecessary.

If sandbox is displayed in the running process, it may be effective to try the following as well.

killall chrome-sandbox

chrome_options.add_argument("--no-sandbox")
chrome_options.add_argument("--disable-setuid-sandbox")

The original article is here

By the way, the sandbox is an isolated environment that does not affect outside programs. (Used to run programs that are likely to contain viruses, etc.)