When the bot is triggered from Orchestrator, for some records the CAPTCHA value is not being entered into the input field.
However, when we connect to the machine and monitor execution, the bot:
Captures the CAPTCHA screenshot
Processes it using the API key
Successfully populates the CAPTCHA field
Additional Details:
While debugging the code or running the process from UiPath Studio on the same machine, the issue does not occur
API key limits are sufficient and not exceeded
No environment variables are used in file paths
The issue occurs intermittently only in unattended execution
Observation:
The automation works as expected when run from Studio or when the session is actively monitored, but fails to enter CAPTCHA values for some transactions when running fully unattended via Orchestrator.
Request
Could you please help with:
Possible reasons for different behavior between unattended and monitored/Studio sessions?
Best practices to handle CAPTCHA reliably in unattended automation?
Can you assist me in where are the I need to check the Resolution, In machine setting it’s Can’t be changeable it’s same for dev and prod. Only error While it’s running in unattended mode, even after i connect to machine while bot triggered in machine it’s running fine like partially unattended mode
.
It can be changed on robot account level and not machine level.
To do, follow below steps:
Tenant → Manage Access → Edit User (Access Rules) → Enterprise Advanced Robot options at bottom, if not Additional Setting tab → update width & height same as dev machine.
Also, make sure to log resolution in bot (heigh & width to track if it’s reflecting)
This can also be done from Folder → Settings ->Manage Access.
I believe CAPTCHA on website is being read using OCR, so ensure that the captcha is visible before doing an OCR based extraction..
ensure the website window is maximized while reading the captcha.. and also once the website opens, send a “Ctrl+0”, this ensure the website zoom is reset to 100%.
if there’s a static element near the captcha, try and wait till that appears.
use captcha extraction under retry scope,(captcha is not completely legible at times, retry ensures it reads a different captcha if its wrongly extracted earlier.) set the retry for finite number of times and give a wait till the new captcha loads and then extract.
Like Rohit mentioned, screen resolution also matters since captcha is read with OCR. Ensure you use the same resolution in DEV, QA and PROD instances especially when you are using image based actions.
