"Loss of connection with the browser" when trying to work with it

[Velinkton]

OpenRPA version: 1.4.56
OpenFlow version: 1.5.7
Using app.openiap.io or self hosted openflow: self hosted
Error message: none
Screenshot or video: none
Attach a simple workflow from OpenRPA or NodeRED that reproduces the error/issue: I can’t provide it because the robot works on a corporate resource

I have a problem where the robot stops detecting the browser in any way. That is, the robot cannot find the element, cannot enter data into the input fields and cannot even close the browser (I’m not sure, but it seems that you don’t even need to install an extension in the browser to close the browser)

The essence of the robot is as follows - it tries to log into the corporate system and make a couple of simple manipulations in the interface in order to make sure that this system is available for use (in other words, system monitoring). In case of an error, a corresponding notification will be sent by email and Telegram

There are several such corporate systems, and the verification of each system is carried out in subprocesses. Each such subprocess begins with the launch of the browser and ends with its closing (regardless of the result of the process)

So, we began to notice strange behavior - the robot opened the corporate system, did not enter anything into the authorization form, did not close the tab after completing the subprocess, and launched the next subprocess. Because of this, we began to open many tabs that clog the server’s memory.

This can happen once a day (that is, during one iteration of checking, nothing happened in the browser, and at the next, everything is fine and the browser closes), or it can continue to work incorrectly.

This monitoring process runs on a single robot user (HDRobots)

To solve this problem, I tried to set a pause of 3 seconds between subprocess calls. It seemed to help, but such a case still happened once.

Could this be a problem with the Google Chrome browser extension? Or maybe this is due to the fact that the browser processes do not have time to close completely when I call a new subprocess that ends up working with the wrong instance of the browser process?

[Allan_Zimmermann]

Closing browser requires the chrome extension to be working.
So if that “stop working” then you cannot close tabs or chrome either.

OpenURL how ever, first tryies to use the chrome extension, if it does not find it, it will launch chrome with the URL ( if chrome is running that will open a new tab )

I suggest you kill the chrome process at the end of each sub process and not depend on close tab.
Maybe use “start process” with taskkill /F /IM chrome /T ( not tested just copied from a google search )

[Velinkton]

This can be used, but there are a few “buts”:

1. I don’t know how killing a process will affect the site itself. In one of my cases, in this case, the session for the user was blocked on all devices, which was quite critical
2. In addition to closing the browser, there are other problems with working in the browser itself - values ​​are not entered into fields, elements are not found, etc.

[Allan_Zimmermann]

It’s not that I don’t want to help, but that is some super generic statements.
If i cannot reproduce it, i cannot troubleshoot it …
And without more details i cannot “guess” what a potential issue could be.

1. A website that cannot handle a user closing the webpage seems “broken” to me. I can not think of any use case where that would be a wanted feature unless we are talking anonymous users and you hit a rate limit.

2. I read it as, there is a problem with the chrome extension sometimes losing connection with OpenRPA. yes, that can happen, and the only way to fix it is to close all instances of chrome… if you want to do that from OpenRPA, you need to use kill.
   Throwing in “values ​​are not entered into fields, elements are not found, etc.” is confusing to me. Is that happening after it loses connection ? then yes, of course. it has no control over chrome anymore.
   If this is a separate issue, then please create that as an separate issue and include something i can reproduce or A LOT more details … else i have no chance of giving any kind of answer

[Velinkton]

You were able to give me the answer. As you said, the browser extension can sometimes lose connection with OpenRPA, and this is the reason for my problem that I wanted to find out. In my situation, I will simply need to handle such a case - I will have to kill all browser processes and wait for the next iteration. Thank you for the clarification

[Allan_Zimmermann]

Oki, good.

Just to elaborate. I have seen some rare cases were chrome send control charecters in the json that makes the native messaging application get corrupted json. It cannot recover from that, despite getting correct json after, hence we need to kill chrome.