Chrome Extension Issue

[AidanMcArdle]

There are many threads existing on this topic, yet none seem to contain a workable solution. I see one with responses from six days ago, but it is suggested to start a new thread.
I have also tried support tickets to no avail.

This has become high-priority for us now that IE has reached end-of-lifecycle. We are currently migrating all processes from IE to Chrome and this issue has gone from occasional minor irritation to high priority. We are currently at a point of breakage where moving the entire system to a UIPath is being mentioned.



Issue: Blue Prism returns error: 'Chrome browser extension not detected' intermittently.
Extension is installed and enabled.
From what I can tell, this is due to the port connection not being correctly severed, and therefore a new connection on the same port causes confusion in the app.


Suggested solutions trialled so far;
1) Increase wait time after closing chrome.
Result: 6 seconds or 60, issue can persist.

2) Ensure fixed port is selected in Blue Prism settings (or not).
Result: We have seen the error continue whether port is set as fixed or variable.

3) Ensure Chrome setting 'Continue running background apps when Chrome is closed' is turned off.
Result: Issue persists.

4) Exit gracefully via Chrome hamburger & 'exit'. OR Exit ungracefully using 'X' to close, followed by killing Chrome process via utility/task manager.
Result: These both close all instances of chrome. Unworkable in a process with multiple (up to four) systems open in Chrome at once.

5) Await updated version of Blue Prism (Current version 6.10.2) 
Result: The issue has been present for multiple versions now and still persists. 

6) Restart Automate.exe OR robot VM. 
Result. These do work! But are also unworkable for obvious reasons. Stepping in manually 20+ times per day (or night) is not feasable.



Can anyone assist? I'm reaching the end of my tether with this bug.

[Joshna_16]

Hi Aidan,

Can you try launch Chrome Application through 
VBO : "Utility- Environment"
Action : Start Process

Inputs :
Application : Chrome.exe
Arguments : URL

After launching, attach the application.

Issue chrome extenstion not detected or installed error may not occur by this method. 



------------------------------
Joshna Dammala
Project Engineer
Asia/Kolkata
------------------------------

[AidanMcArdle]

Thanks for the suggestion Joshna. 
I have spent the morning tinkering with this method. No luck, launch and attach seems fine, but elements are no longer being identified/highlighted in the app model.

Steps followed using your method;
Pre) Object application model's application type has been changed to 'A browser which is already running'.

1) Launch chrome URL using 'Utility General - Start Process'
2) 'Utility General - Get Windows' in order to get process name / ID etc.
3) In the business object for this app, check connected. If no, then attach. (tested using Window Title, and also using Process ID)


This seems to result in successful attachment. However, on closer inspection not all elements are covered. Win32 elements can be identified and highlighted okay, but Browser features can not even be spied.

When attempting to highlight a previously identified feature, error is returned: 'No matching Web elements found'.
If attempting to identify a new element using 'identify', nothing is red-lined in Browser mode (Other modes work, but these only work on the outer frame etc.).
If attempting to identify using 'open application navigator', then the error 'An error occurred retrieving the application model' is returned. When attempting

------------------------------
- Aidan- Aidan
------------------------------

[Joshna_16]

Hi Aidan,
 
I think problem might be attach. Can you try giving child index as either 0 or 1 in attach.

------------------------------
Joshna Dammala
RPA Developer
Asia/Kolkata
------------------------------

[AidanMcArdle]

Thanks again Joshna, The quick responses are appreciated.
This is something I played around with this morning too, based on mentions of same in this thread (linked).

Let me be a little more clear.Following your method, I first use the 'Utility Environment - Start Process' to launch to the right Chrome URL
Then I'm using the 'Utility General - Get Windows' action allows me retrieve the process ID (PID) of the correct window before trying to attach.

Using this with Child Index of 0; Win32 elements are identified/highlighted, but browser are not. Spying new elements in Browser mode does not find anything.
Using this with Child Index of 1; Results are identical.


If instead I use the Window Title to attach, along with the Child Index, results are a little different for each.
Using this with Child Index of 0; Win32 elements are identified/highlighted, but browser are not. Spying new elements in Browser mode does not find anything.
Using this with Child Index of 1, no elements are identified. Spying elements continues to fail. (I expect in this case we attached to a non-GUI Chrome instance).



------------------------------
- Aidan
------------------------------

[Tejaskumar_Darji]

I was trying to create some demos with Chrome last week and I also faced all of the above issues. Again tried various workarounds found in the community.

There are many threads, workarounds, KB-articles etc but nothing solid on this yet. These workarounds work for some folks and not for all but it is still a workaround. 

Browser extension compatibility restricts the support of browser versions and extension versions this is also not a sustainable option because of the rising impacts of online malware/viruses etc, that browsers will be forever going to update very frequently. In this scenario, if an organization chooses to restrict the auto-update of the browser then it poses a security risk and if they update the browser there are high chances of all these issues popping back. Again for each browser version team will have to do rigorous testing on all objects. This will increase the manual efforts and that too when a new browser version is released frequently.

We surely need a master article and solution for this for a wider audience to sustain browser automation.

Surely Chrome/Edge works on different mechanisms than IE but with IE we all have experienced highly stable browser automation.

I definitely agree we need some concrete and sustainable solutions for the browser automation issues because even with the latest version of Blue Prism browser automation issues continues.

Anyone from the Product team can give more insights on how this issue is being handled right now and what is the priority.

Looking forward to stable Chrome/Edge browser automation using Blue Prism.
​​​

------------------------------
If I was of assistance, please vote for it to be the "Best Answer".

Thanks & Regards,
Tejaskumar Darji
Sr. Consultant-Technical Lead
------------------------------

[AidanMcArdle]

Thanks for the comments Tejaskumar, but to the matter at hand - did you find any workarounds not mentioned in my listing above?

------------------------------
- Aidan
------------------------------

[AdityaPatil2]

Are you using Edge browser(With extension installed) ?
If yes , then it interferes with the BP  port connection with Chrome browser.

Solution :- 1.  Try to run only chrome browser.
                   2. Kill edge processes from Background.
                   3. Disable "Startup Boost" in Edge (Setting > System)
                  Note : -After following above steps , if it doesn't work try to juggle between fixed and  variable. Hope this works along with above steps.



------------------------------
Aditya Patil
------------------------------

[Alexander.Duff]

Hi Aidan,

Due to its inevitable demise, we migrated everything from IE to Chrome last year, so don't envy the task.

This may have been covered in this/other threads but ill let you know how we manage with all of our chrome processes

We've found the following to be the most efficient way of launching chrome and subsequently mapping using 'Browser'

• Create Browser Based Application within Object Studio using an executable file (where chrome is stored)
• Use a navigate stage to 'Launch' the application passing through the [URL] as the input
• Check that the instance of chrome exists using Utility - General - Check if process exists

When carrying out processes within chrome run an 'Attach' page against each page to ensure it's attached to the instance of Chrome you've initiated.

So far we've not experienced any failure of the Chrome EXT.

We've recently moved from on-prem to BPC so this may have helped (we were previously running on an old extension of chrome but due to the upgrade we've changed ours to the one you mentioned).

Hope this helps,

------------------------------
Alex Duff
RPA Developer
Lowell Financial
Europe/London
------------------------------

[prateekgrover]

I am not sure what works universally.
But for us , We found out 2 things:
 
1. Extension Issue occurs as there is not enough time for the Extension to connect to chrome so we use waits and attach after launch if BP is not connected to browser.
2. When you close the Browser you are allowing the Browser to store the cache files which might cause extension issue in the subsequent runs.
For this , and to close all the windows elegantly, we use a PowerShell query.( Utility Environment - Run Process - powershell.exe)
Arguments:
"Get-Process ""chrome"" | For Each-Object($.CloseMainWindow() | Out-Null)"
Timeout: MakeTimeSpan(0,0,5,0)

Hope this helps someone.

Cheers,
Prateek.

------------------------------
prateek grover
------------------------------