SoSci Survey questionnaire crashes to blank screen at random points

Question

Dear SoSci Survey team,

I am currently running an experiment using SoSci Survey and have already collected some data. It has always worked fine and I have not had major problems so far. However, in the past few days I have repeatedly encountered the following issue:

For some participants, at a seemingly random point during the experiment, the screen turns completely white after clicking “Weiter” on a rating page. Reloading the page or reopening the link (also on another laptop) does not solve the problem. For some participants, the page stays completely blank; for others, they always return to the same point (the same page) and cannot progress.

This has now happened with several participants at different points in the experiment (e.g., for some around 25%, for others around 89% progress).

Could you please explain why this might be happening and how we can fix or avoid this issue?

Thank you in advance for your help!

Comments

> Reloading the page or reopening the link (also on another laptop) does not solve the problem.

Do you have access to the URL shown in the browser at this point? If it does not work on another PC as well, something's wrong with the experiment. If it works on another device, then its most likely a device or connectivity issue.

---

I tried to reproduce the problem by going through the experiment myself as a participant. At a certain point in the survey I received the following error message in the browser:

"“Looks like there’s a problem with this site
https://www.soscisurvey.de/my_experiment_link/index.php?i=N31QPPJ7IB03&rnd=TWVP
 might have a temporary problem or it could have moved.
Error code: 500 Internal Server Error
The site could be temporarily unavailable or too busy. Try again in a few moments.”"

If I reload this URL, I continue to get the same error.
If I instead reopen the experiment using the general experiment link, the survey works again for a while, but then the same error appears again.

I don’t think this is a device or connectivity issue because I have now tested it on multiple devices with stable internet connections.
The experiment itself has not been changed recently, and it ran without problems until a few days ago. The issue only started occurring recently, although the setup is the same.

Thank you for your help!

---

Do you have a (as accurate as possible) time for that? The STARTED value if the respective case should be a good start.

---

I'm not entirely sure I understand what you're asking for. However, I can provide the following information:

The STARTED value for the interview that crashed this afternoon after I tested was 15:05:26.
For another participant, who reported the issue this morning, the experiment started at 08:00:31.
I hope this helps! Let me know if you need any more details.

---

> The STARTED value for the interview that crashed this afternoon after I tested was 15:05:26

Which time zone? As you're writing in English, I do not see it as granted that we're talking about CET? Are we talking about the project /magco_fmri_...?

Also, pleasee see the answer below: https://support.soscisurvey.de/?qa=66257/sosci-survey-questionnaire-crashes-blank-screen-random-points&show=66264#a66264

---

yes, the times I provided are in CET and "magco_fmri.." is the project I´m talking about.

---

Hi @s293761, as we seemed to have a similar issue - has yours been resolved? And if so, just by itself, or did you change anything? I just tried our survey again as it seems fine - but as we will be applying it again with the same participant that it had issues with last time, I want to make sure that there is no issue this time.
Thanks in advance!

---

In the above case, the problem was that the questionnaire generation exceeded the memory limits. The issue was likely caused by storing too much data in the background - via registerVariable() in that case and/or by loops that eventually resulted in deadloops.

---

Hi. In my case the problem was indeed caused by exceeding the memory limits due to saving too much data via registerVariable().
I managed to solve it by avoiding saving all those data. After removing the unnecessary registerVariable() calls, the survey has been running smoothly so far.

---

Thanks, unfortunately the problem occured again, and we had to carry out the data collection on paper which is really not ideal. The project it concerns is 'Resilienz (BRS)'. How can I find out if the problem is also the exceeding memory, or something else?

---

Please send the project's URL and the (as exact as possible) time when the problem occured to info@soscisurvey.de. We can than check against our logfiles.

Answer 1

Hi,
we had the same issue yesterday and today after the survey has been working fine for months: the screen goes blank at different points during the survey.

During three interviews, it repeatedly happened that the connection to the server was interrupted and the questionnaire could no longer be continued afterwards. The internet connection was stable in all cases, and the problem occurred both on my own computer and on an in-house PC. Today, the first interview even had to be interrupted three times and restarted completely. In the second interview, after page 3, no further access to the questionnaire was possible, so unfortunately I had to cancel it completely.

What could be the reason for this? If you say it is due to the experiment - what could be the issue here, if it has been working previously?

Grateful for any support!

Comments

Please see below or above, depending on your browser: https://support.soscisurvey.de/?qa=66257/sosci-survey-questionnaire-crashes-blank-screen-random-points&show=66264#a66264

Answer 2

What could be the reason for this? If you say it is due to the experiment - what could be the issue here, if it has been working previously?

There are different possible causes. One may be related to our server moving to new hardware on 04.10.2025, which also went along with some configuration changes. In the beginning, we had some issues with 502 "Bad Gateway" errors, though that seems solved by now. I see a few such errors in the error logs for 01.12. around 11:00 UTC, but nothing scary.

I also see some "500" server errors in the log, such as described above. These are usually related to "Fatal Errors" in SoSci Survey itself or in the PHP code used in the questionnaires. However, those fatal errors I observe in the error logs are caused from invalid PHP code, for example:

[03-Dec-2025 15:47:20 Europe/Berlin] PHP Fatal error: Cannot use isset() on the result of an expression (you can use "null !== expression" instead) in /s2data/s2survey/html/s2survey/phpcode/PHPCodeExecutor.php(568) : eval()'d code

If you have an isset() on one of your questionnaire pages, give it a closer look. Such problems should however occur when one creates a questionnaire, not months later.

Another possible reason are updates of the underlying PHP version. The current version 8.4 (will be 8.5 in near future) is stricter than the previous 8.3, which may cause some code to fail that worked before. Again, one would expect something like to go along with error messages on the questionnaire page.

If I reload this URL, I continue to get the same error.

Please feel free to send such an URL to info@soscisurvey.de, so we can track down the issue. If a problem can be replicated, it's usually easy to solve.

Answer 3

Thank you for sending the link and the description how to reproduce the error.

According to our error logs, your PHP code causes trouble:

Error: Warning: An error occured during PHP-code evaluation: 1: Allowed memory size of 67108864 bytes exhausted (tried to allocate 20480 bytes)

So, your code requires more than 64 MB of memory. However, I do not immediately see the reason in the code. What I see is that a large number of internal variables is written. Where does this variable $dim come from and how much does it contain?

Comments

The experiment is quite long, but it’s strange that this issue has never occurred before. If it’s a memory space problem, shouldn’t it have happened earlier?

Regarding the variable $dim, it is defined as Dimensions (DM01) in the list, but I don’t believe it occupies significant space. If this is a memory problem, I think it might lie with the large array called "blocks" on page 11. When testing the questionnaire in debug mode, I received the following warning message:

Warnung (Seite 11)
"Sie speichern Zusatz-Daten (%blocks%) mit mehr als 2 KB in diesem Interview. Bitte verwenden Sie für umfangreiche Platzhalter den Befehl show() anstatt replace()."

This warning has always appeared, but I’ve never had any issues related to it before.