Experiment fails
Hello,
I am having an issue with an experiment. About 1/4 of the participants tell me that it gets stuck in the middle, showing a blank white page, with nothing to do.
I know it is not a problem in the code because it works perfectly fine for the others.
When looking in the results page, I see it is written FAIL- It's not allowed to reload this component (ID: 72). Study (ID: 67) is finished. But I believe this message appears because they close the page after it is not responding.
I don't know what to do and would love to hear your suggestions.
Thank you very much!
Inbal.
Comments
Hi Inbal,
These cases are really hard to debug. From what you describe, it sounds to me like the script gets stuck somewhere, participants try to reload the page, and that's when the status gets set to FAIL.
If you are running the code on your own server, you might be able to access and check the logs to see if you see anything useful, or any indication of where your code is getting stuck. If you're running it on mindprobe, you first need a way to reproduce the problem. Once you have reproduced the problem (you, or or somebody with another system - have you asked which browser and OS these participants are using? Maybe it runs well on some browsers but not on others), you can look at the console to see the error you get.
Hope this helps.
Elisa
Thank you for your response!
It usually works fine when Iq other participants run it, so I asuume it is not a problem in the script, dont you agree?
When a participant reported it got stuck, I tried running it myself and then it happened to me to, showing these errors on the console:
What does it mean?
Thank you very much for your help!
Hi,
Those error messages mean that the WebSocket for the batch channel could not be opened. JATOS uses WebSockets for some fast data exchange between the server and the browsers. Somehow those could not be established. But since there are no other error messages (I assume) it means the normal network connection works fine, so it's only the WebSockets which do not "get through". Unfortunately I cannot do much from here since I have no access to your system. Your problem can have many reasons and maybe they were only temporary and are already solved. If you have an 'admin' account on your JATOS you could have a look in the logs for errors or exceptions and post them here.
Best,
Kristian
Hi Inbal, to add to Kristian's answer:
It's not necessarily the case that if the code runs for some participants, it's not a frontend problem (the code you wrote). This is because unlike in lab-based experiments, you control a lot less online. Some browsers are more forgiving than others when it comes to syntax errors. And once I had a problem with some kind of memory buildup on some versions of Chrome but not on others. It's also possible (just an example) that you are sending data to the server too often, leading to more stochastic failures.
Not a real answer, just a comment :)
Best