High RAM usage with SeleniumBase - how to properly free resources between iterations?

[sebastian-hermix]

I'm processing several thousand URLs using SeleniumBase with undetected-chromedriver mode, but I'm experiencing significant memory accumulation over time. Even though I'm calling driver.close() and driver.quit() in the finally block, CPU* usage keeps growing with each iteration.
Here's my current code structure:

for url in urls:  # Processing thousands of URLs
    driver = Driver(uc=True)
    driver.uc_open_with_reconnect(url, 4)
    time.sleep(1)
    try:
        # Business logic: bypass Cloudflare antibot and scrape data
        pass
    except Exception:
        raise
    finally:
        driver.close()    
        driver.quit()

Questions:

1. What's the proper way to completely free RAM and CPU resources after each iteration in SeleniumBase?
2. Are there any additional cleanup methods I should be calling beyond close() and quit()?
3. Should I be using a context manager or different approach for better resource management?
4. Is there a way to force garbage collection or clear browser cache/processes between iterations?

Any guidance on preventing memory leaks when processing large numbers of URLs would be appreciated.

[mdmintz]

The Driver() isn't a context manager format, so you can't guarantee all resources being freed until the Python program terminates.

Use the SB() format to guarantee resources being freed at the end of the with block:

from seleniumbase import SB

with SB(uc=True) as sb:
    sb.open(URL)
    # ...

You can also try the Pure CDP Mode, which has less memory usage because it doesn't use Selenium/WebDriver at all:

from seleniumbase import sb_cdp

sb = sb_cdp.Chrome(url)
sb.open(URL)

You can use objgraph to profile the memory usage. Here's an example with the Pure CDP (sb_cdp) format:

import objgraph
from seleniumbase import sb_cdp

print("\nBefore opening the browser:")
objgraph.show_growth(limit=50)

url = "https://seleniumbase.io/demo_page"
sb = sb_cdp.Chrome(url)
for i in range(10):
    sb.open(url)
    sb.sleep(0.1)

print("\nAfter browser is open:")
objgraph.show_growth(limit=50)

sb.driver.stop()

print("\nAfter closing the browser:")
objgraph.show_growth(limit=50)

Final Output: (After closing the browser)

After closing the browser:
tuple        7307        +3
cell         4328        +3
function    19303        +2
Context        45        +2
Future         11        +1
Handle          8        +1
Close           1        +1

(The +3 max memory increase is a small amount remaining that frees up after Python is exited)