WebDriver UDF - Help & Support (IV)

[CodeWriter]

My code was working fine this morning for Chrome/Edge. Returning this afternoon without making any changes in the code or any driver updates, however, _WD_CreateSession() is now unable to launch for Chrome/Edge and returns the errors:

__WD_Post ==> Webdriver Exception [10] : HTTP status = 500
_WD_CreateSession ==> Webdriver Exception [10]

Does anyone recognize what these errors mean? These error persist after rebooting. Thanks.

Edited March 8, 2023 by CodeWriter

[Danp2]

@CodeWriterYou should enable full debugging to get the greatest level of details related to this issue. I would hazard a guess that it's an issue related to the profile being used. You should double check that there aren't any "zombie" instances of the browser handing around.

[CodeWriter]

@Danp2Thanks for the quick reply as always. A couple of quick follow up questions:

1. What global variable should I change to enable full debugging? $_WD_DEBUG_FULL I’m guessing but it provides no new debugging information than what’s posted.

2. would “zombie” instances persist beyond a reboot cause these errors have lingered after several reboots.

Edited March 8, 2023 by CodeWriter

[Danp2]

1. Change $_WD_Debug or use the new _WD_DebugSwitch.

2. No

[CodeWriter]

Perhaps it has something to do with Chrome being updated to 111.0.5563.65 (64 bit) and the driver is now out of date. Strangely, when you try to download the latest chrome driver for 111.0.5563.64, it doesn’t offer a 64-bit windows version but only 32-bit (chromedriver_win32.zip)

[Danp2]

I think that's normal for chromedriver to only offer a 32-bit version for Windows.

[CodeWriter]

It would be nice to have a method that prevents chrome from auto updating until the corresponding driver is ready. Until Google provides control of when you want to update, it’s an ongoing frustration that what worked this morning is rendered useless by an auto update while you stepped out for lunch.

Edited March 8, 2023 by CodeWriter

[Danp2]

@CodeWriterThere is probably a way to disable the automatic update, but it isn't something that I've investigated.

[mLipok]

46 minutes ago, Danp2 said:

@CodeWriterThere is probably a way to disable the automatic update, but it isn't something that I've investigated.

Let me check if I had done it before.....

[mLipok]

54 minutes ago, mLipok said:

Let me check if I had done it before.....

nope.
but after some research :

https://stackoverflow.com/questions/16149610/how-to-override-default-set-of-chrome-command-line-switches-in-selenium

 

maybe this:
--safebrowsing-disable-auto-update

?

You can also take a look here:

https://www.chromium.org/administrators/turning-off-auto-updates/

 

[CodeWriter]

@Danp2I've been trying to get to the root of why running my code with Chrome or Edge takes > 20 sec to load the first page while it's nearly instantaneous with Firefox. Going back to your demo (wd_demo.au3), I've noticed the same issue running the "DemoNavigation". The demo takes 20-25 sec to load the first Google page with both Chrome and Edge but not with Firefox on my workstation. Any thoughts about why this delay occurs and how to eliminate it?

[Danp2]

@CodeWriterI just tested and the first page loads instantly for me with Chrome. Have you tried reviewing chrome.log to see if it provides any clues?

[CodeWriter]

@Danp2Thanks for the quick reply as always. Nothing obvious in the logs. Must be something unique to my workstation. Will continue digging around. 

[CYCho]

13 hours ago, CodeWriter said:

Must be something unique to my workstation.

Your worksation is not unique. I am experiencing the same phenomena, too. I have two computers with following hardware configurations:

1. Main computer
CPU: AMD Ryzen 5 3400G       3.70 GHz
Graphics: AMD Radeon(TM) RX Vega 11 Graphics
RAM: 8.00GB

2. Secondary computer
CPU: Intel(R) Celeron(R) N5095   2.00 GHz
Graphics: Intel(R) UHD Graphics
RAM: 8.00GB

The main computer takes less than 2 seconds to navigate to the first page, and the secondary computer always takes about 20 seconds more. There is no meaningful difference in performance, including navigation to a different website, after loading the the first page. Both computers run Chrome browsers in Windows 11.

However, I found out that the difference of speed existed regardless of webdriver: there was a similar time difference in loading the first instance of Chrome.exe without a webdriver. So my conclusion is that the Chrome.exe takes unreasonably long time to load in some computers.
I have not tested this with other browsers.

Edited March 18, 2023 by CYCho

[mLipok]

16 minutes ago, CYCho said:

1. Main computer
CPU: AMD Ryzen 5 3400G       3.70 GHz
Graphics: AMD Radeon(TM) RX Vega 11 Graphics
RAM: 8.00GB

2. Secondary computer
CPU: Intel(R) Celeron(R) N5095   2.00 GHz
Graphics: Intel(R) UHD Graphics
RAM: 8.00GB

Please suplement this information by adding:

1. how many cores are on each of this computer
2. what kind of disc are installed there (HDD, SSD, NVME, SATA2, SATA3, SCSI, SAS, or at least some kind of disc speed test results)

 

Edited March 18, 2023 by mLipok

[CYCho]

41 minutes ago, mLipok said:

1. how many cores are on each of this computer
2. what kind of disc are installed there (HDD, SSD, NVME, SATA2, SATA3, SCSI, SAS, or at least some kind of disc speed test results)

Both of them are quad-cores, installed on SSD's.

One solution suggested in internet was to disable hardware acceleration in Chrome settings. I tried it, but it was of no use.

Edited March 18, 2023 by CYCho

[mLipok]

2 hours ago, CYCho said:

Intel(R) Celeron(R) N5095   2.00 GHz

In the past I had similar problem with IntelCeleron N series system - Is he passively cooled ?
Check for any ECO settings, turn them off , and try again. The system should be set to the highest performance setting.

Edited March 18, 2023 by mLipok

[CYCho]

16 hours ago, mLipok said:

Is he passively cooled ?

No, it is not. The device is a Mini S computer made by Beelink in China and it is actively cooled by a pretty large fan compared to its overall size. I have not made any change to the device settings.

During the last couple of hours, I installed Firefox in both of my computers and did some tests.  The two computers, with or without webdriver, showed very similar speeds with Firefox, comparable to the speed of my main computer with Chrome.

Edit: I can't switch to Firefox because I have to use multiple sessions simultaneously, which is easy to implement in chromedriver but cumbersome in geckodriver.

Edited March 18, 2023 by CYCho

[naixZ94]

Hello,

i have a working script to track different values on different urls.
The script gets data from one site and changes the tab to the next site, but every time the tab changes chrome gets focused.
This is the same if i minimise chrome before, after the tab change it get maximised.
This is annoing if i want to work at the same time.

I tried different ways to change the tabs but the behavior is the same.

_WD_Attach($sSession, $aUrlListTrack[$j][0], 'URL')
_WD_Window($sSession, 'Switch', '{"handle":"' & $aUrlListTrack[$j][2] & '"}')

i know i can run the browser without a gui but this is not what i want.

Is there any way i can change tabs without chrome getting focused?
for example

_WD_Action($sSession, "refresh")

does not change the focus on chrome.

[mLipok]

Have you tried with FireFox or MS EDGE?
What about these browsers?

 

btw.
Welcome to AutoIt forum.