unable to download 'results'

[Umesh]

Hi,

I am currently running three experiments in my class. My students ‘published’ the three experiments and ran a couple of pilots. Unfortunately, we are not able to download the results files (for any of them). We get the following two warning messages.

1. While opening the results file in the browser
# Last submission only; create an account for full results file
# WARNING: could not generate full results file due to timeout or disconnection

2. After downloading the results file
# Published experiment
# WARNING: could not generate full results file due to timeout or disconnection

Am I doing something wrong?

Thanks!
Umesh.

[Jeremy]

Hi Umesh,

Thank you for your message, I found out that one service has been down for ~3h. I restarted it, and as far as I can tell, it has now caught up with queued operations, and I am able to generate results files again. Let me know if you still cannot access your students’ results files

Jeremy

[Umesh]

It’s working now.
Thanks a lot for the quick response, Jeremy!

-Umesh.

[JulianaNovo]

Hi Jeremy,

I am currently running another experiment 🙂 I need to say PCIbex has been working just fine.

Recently though, Some participants have been reporting to take the experiment, but the data does not show on the results file. I thought it could be one isolated case, but now there are a couple of participants that have reported taking it but still I can find their results.

I’ll share the demonstration link
https://farm.pcibex.net/r/qyIKwz/

Hope you can help me out, as always.

Thanks,
Juliana

[Jeremy]

Hi Juliana,

Your code seems fine, as far as I can tell (the lines PennController.SendResults won’t have any effect because there’s no () but that doesn’t matter since you have SendResults() directly in your Sequence)

Do you know if the participants took your experiment at a specific time? With a specific browser? From a specific region?

Also, how many lines do you have in your results file? The PCIbex Farm will stop serving data after 100K lines of results, so if you have more than that you will need to collect your results file in multiple times

Jeremy

[Danny]

Hello Jeremy,

I’m hijacking this thread to dig deeper into something you mentioned:

Do you know if the participants took your experiment at a specific time? With a specific browser? From a specific region?

I am located in Germany and advising my participants to do the experiment in the morning hours of Central European time. I do this because I assume most people in the US who run and take part in PCIbex experiments are still sleeping and thus don’t cause too much load on server bandwidth(?).

I also figured that browser support seems to matter (with Safari apparently the worst).

What else is it about these factors (time, browser, region) that affects smooth running of the experiment or data transfer to PCIbex servers? And how can we influence that overseas?

I appreciate all your good work and support here in this forum. All questions I’ve had so far could be answered with a simple string search here. So thank you!

Danny
@ Heidelberg University Language and Cognition Lab

• This reply was modified 2 years, 5 months ago by Danny.

[Jeremy]

Hello Danny,

Runtime is probably a factor that matters, for the reason you mention (server load). However, browser should not matter when it comes to sending the results to the server (it does matter if, for example, you use the MediaRecorder or EyeTracker elements, which are poorly supported by Safari, if supported at all)

The most important factor regarding data transfer is the amount of data being transferred: if a submission contains thousands of lines, storing them in the server’s database will take time, and as incoming results accumulate, the risk of crash due to overload increases

Jeremy

[Author]

Posts