Impossible to send results to the server

I just wanted to ask if people have ever encountered problems while saving data at the end of the experiment. Yesterday, a few of my subjects were complaining that an error would pop up while the orange window popped up. The error said something like:

when uploading participant’s results for experiment: [name]. Bad Gateway

In the end, it turned out that the data files for those subjects were indeed stored on github, but I was wondering if anyone knew of about this problem and a possible fix.


I haven’t seen this. @apitiot any ideas what would cause this?

Hello @rob-linguistics13,

Could you let me know which experiment you are running? I’ll have a look at the pavlovia logs and try to replicate a potential issue.


I’ll message you the link of the experiment. Thanks!

Hello @rob-linguistics13,

Checking the logs, I can confirm that the results were successfully saved even though your experiment did experienced a few git conflicts.
May I ask you whether you were perhaps updating the experiment on your end (or deleting results files from your local data repository) while participants were running it, e.g. around 12h30 UK time on the 1st of August?


Hi Alain,

thanks for your message. Around that time I was actually testing the script. I started collecting data around 18h00 UK time and after that a few people have notified me of the error I was mentioning. It is true though I deleted the local folder of the experiment (I asked in another topic yesterday, and I believed the local and the remote folders were independent of one another, so I trashed the locale folder to make some space on my local disk).

On the other hand, I confirm that the output data files of the very same subjects that notified me of the error were actually in the git folder, so at the end of the day I did not lose any data.

Just to clarify my previous post, if you delete the local repository, you will be unable to sync your changes to the online repo (until you clone it). If there were files being deleted locally, then you would have to push those changes up to the online repository in order for them to cause a git conflict. This could only happen if the local repository was still present.

1 Like

Thanks for your clarification. I confirm that I deleted the whole local folder was removed before data collection, so according to this, the absence of the local folder could not have caused the git conflict. Is that right?

Yes, that should be the case. Even if you still had the task open in PsychoPy whilst the task and folder was deleted, it would throw an directory not found error if you tried to sync.

I haven’t used PsychoPy to sync for a while. I am just using it to create new projects for new experiment, but then I would just work on the js scripts on github. So, the git conflicts @apitiot was referring to could not have possibly been due to syncing via PsychoPy.