Lotje suggested I post this here after emailing support.

Static study assets on jatos.mindprobe.eu are served uncompressed. For my study the main HTML file is 124 KB and goes over the wire as 124 KB.

Anyone can check this from a running study:

const e = performance.getEntriesByType('resource').find(x => x.name.endsWith('index.html'));

console.log(e.nextHopProtocol, e.encodedBodySize, e.decodedBodySize);

// h2 123760 123760 <- encoded == decoded, no compression

The same file compresses locally to 36 KB with gzip -6 (-71%) and 29 KB with brotli (-76%).

Why this matters beyond bandwidth: some of my participants connect from regions with unstable routes to European servers. I have one confirmed case where the HTML download stalled part-way. The browser had parsed the title but never received the body, so the page stayed blank, the tab kept spinning, and the participant could not take part. Reloading did not help. A smaller transfer means fewer opportunities for the connection to break, and the benefit is largest exactly for the participants who are hardest to reach.

This affects every study on the instance, not just mine: jspsych.js, plugin files and CSS are all text and all served uncompressed.

Would it be possible to enable compression for text types under /study_assets? In nginx that is roughly:

gzip on;

gzip_types text/html text/css application/javascript application/json image/svg+xml;

gzip_min_length 1024;

Happy to verify from my side once it is in.

Thanks,

Margarita (study 28199)

I’m currently trying to access my task on OSWeb via the JATOS server, but since yesterday I’ve been getting this kind of message when I access the task through the JATOS link. Nothing happens when I click OK or Cancel; the only result is that the background turns black.

Could this be related to the issues that MindProbe/JATOS is currently experiencing?

Thank you for your help.

Ayrton

I am running a behavioural experiment on a JATOS server, with the experiment implemented in PsychoPy/PsychoJS.

For each participant, I generated a unique JATOS link. When generating each link, I entered the participant's participant code and name in the Comment field so that I could identify which result belonged to which participant.

For most participants, the data appears normally in the bulk results export. However, for some participants who have definitely completed the task, their results are not appearing in the combined/bulk exported data.

I have checked the individual results on the JATOS server, and the situation is interesting:

• The individual result is present on the JATOS server.
• When I select the individual result and download/extract the result as a TXT file, the task data is clearly present in the TXT file.
• However, when I convert this TXT data to CSV, the resulting CSV does not contain the expected data.
• I have also noticed that some fields/variables may be missing or formatted differently in these problematic files, although the actual trial data appears to be present.
• Additionally, the Participant Code that was entered in the JATOS link/comment is not appearing in the final exported raw dataset, making it difficult to associate these results with participants.

I would therefore like to understand:

1. Is the Comment field associated with each generated JATOS link stored in the result metadata, and is there a way to include/export this Comment field together with the corresponding result data?
2. Why might a result that is visible and contains data when downloaded individually as TXT not appear correctly in the bulk CSV export?
3. Is there a recommended way to recover/export these individual results while preserving all the trial-level data and metadata?
4. Could differences in the structure or data types of individual data.txt files cause the CSV export/conversion to omit the data?
5. Is there a recommended JATOS export format or procedure for recovering these results without losing any trial data?

I am particularly concerned because the individual result files appear to contain the data, so I would like to make sure that the data are not being lost simply during the export/conversion process.

I can provide a sample of the problematic .txt result file and/or the JATOS results export if that would help diagnose the issue.

Thank you!

Since this evening (10 September 2026, CEST), the MindProbe JATOS server has been returning the following error:

502 Bad Gateway

nginx/1.18.0 (Ubuntu)

I cannot access the MindProbe server in any browser. I checked jatos.mindprobe.eu using a website availability service with servers in multiple countries, and it returned the same 502 error from every location. Therefore, this seems to be a server-side outage affecting the entire MindProbe JATOS service.

Unfortunately, I had just launched a study and was about to begin active data collection, so I have had to pause recruitment.

Does anyone know:

• whether this outage is already known or related to maintenance;
• whether a similar MindProbe outage has happened before;
• how long these outages usually take to resolve;
• and whether researchers need to take any action after the server becomes available again?

Any information or help would be greatly appreciated, as the outage is currently delaying active data collection.

Thank you!

I’m currently trying to access my task on OSWeb via the JATOS server, but since yesterday I’ve been getting this kind of message when I access the task through the JATOS link. Nothing happens when I click OK or Cancel; the only result is that the background turns black.

Could this be related to the issues that MindProbe/JATOS is currently experiencing?

Thank you for your help.

Ayrton

I am using mind probe, and I have several studies in the server that I am unable to delete.

When I try, I get the following error message:

Does anyone know how to fix this issue and remove the study from the server?

Thank you

Hi, 

I am running a visual world eye tracking experiment using OpenSesame 4.1.9 and OSWeb 2.2.9 on MindProbe. The camera opens correctly during the experiment but WebGazer never produces any gaze predictions and my gaze_json variable is always empty.

The browser console shows these errors:

• Failed to load resource: face_mesh_solution_simd_wasm_bin.js (404)
• Failed to load resource: face_mesh_solution_p...assets_loader.js (404)
• Failed to load resource: mediapipe/face_mesh/face_mesh.binaryb (404)
• No stream
• Uncaught TypeError: t is not a function (face_mesh.js:53)

I am attaching a screenshot of the console. It looks like the TFFacemesh model files are missing from MindProbe's server. Is this a known issue and how can I fix it?

Thank you

I've just downloaded and installed the Jatos Bundle (version 3.10.4) on Windows 11. All tests on my local instance are "OK", however creating a new study or importing any jzip from the Example Studies cause the following error: "Couldn't generate path to file or directory"

I would appreciate any help!

I created a lexical decision task in OpenSesame. However, whenever I try to share the link with participants, they receive the following message in their web browser: web page not available net::ERR_Connection_refused

Does anyone know hot to fix this issue?

Do I need to install JATOS on every participant's own laptop or computer?

Kind regards,

Yoran

I created an experiment using PsychoPy and adapted it for online use with JATOS. It was working normally until yesterday around 17:00 (GMT+2). Since then, when I try to run it, I get the following error message during loading:

“Unable to download resource: abcde_training_block - NOFEEDBACK.csv”

It also fails to download another resource file.

At the same time, my colleague, who created her experiment using OpenSesame and was also using JATOS without issues, is now experiencing a freeze with the message: “Building experiment structure.”

Could anyone help us understand what might be causing this issue?

Thank you very much!!

We have been running an online experiment using OSWeb v2.2.9 and JATOS mindprobe servers for a while now. Recently, we received reports from a few participants stating that the experiment freezes, crashes, or occasionally fails to display certain images. We launched the study after piloting, during which none of these issues occurred. We have already reached a sample size of 30, and only recently 2-3 participants reported such problems.

I was able to replicate the issues myself. Interestingly, different browsers behave differently. Chrome does not even allow the experiment to start, and the console throws the following error message: "Failed to load resource: the server responded with a status of 404 ()", specifically for the following image files: T_NC_toy%20car.jpg, T_NC_toy%20cell%20phone.jpg, T_NC_baby%20doll.jpg, and O_NC_pencil%20case.jpg.

Firefox allows the experiment to run initially, but then it freezes, I assume when one of the problematic images is about to be presented. Opera also allows the experiment to run, but the "broken" images are simply not displayed (we can tell because the text descriptions presented below the images still appear on those trials).

I have not included the experiment script, as I suspect the issue may be related to the JATOS server or OSWeb itself. However, I can share relevant parts of the script if necessary. We are not sure what we might be missing and would greatly appreciate any help or suggestions regarding this issue.

Thank you!

P.S.: Just noticed a recent discussion noting that the latest JATOS update is stricter with filenames so spaces are not allowed. I guess this explains everything.

Suddenly, I cannot access my study results on Mindprobe/JATOS. I was able to download the results on Friday, but now the Results page freezes completely.

When I open the Results page, the “Working...” window appears and the page freezes completely. I cannot click anything. Export also does not work. And that window never disappears. I've been waiting for so long on different browsers.

The browser console shows this error:

Uncaught RangeError: Invalid time value

  at Date.toISOString (<anonymous>)

  at getLocalTimeDataTables (2fae453d15a226314ed01950e7e5ff94-helpers.js:66:55)

  at Object._render (a7508584d686f953d98b51ddb380c2bb-datatables.min.js:16:12162)

  at Object.fnGetData (a7508584d686f953d98b51ddb380c2bb-datatables.min.js:16:14890)

  at G (a7508584d686f953d98b51ddb380c2bb-datatables.min.js:16:19257)

  at Nt (a7508584d686f953d98b51ddb380c2bb-datatables.min.js:16:31711)

  at s (a7508584d686f953d98b51ddb380c2bb-datatables.min.js:16:26079)

  at a7508584d686f953d98b51ddb380c2bb-datatables.min.js:16:34084

  at Object.e [as success] (a7508584d686f953d98b51ddb380c2bb-datatables.min.js:16:30354)

  at c (2c872dbe60f4ba70fb85356113d8b35e-jquery-3.7.1.min.js:2:25304)

Is there a simple way to:

• export the data without opening the Results page, or
• remove the broken result?

I am not very familiar with JATOS technical details, so I would really appreciate simple instructions.

Study UUID: 9e0a6666-56b8-478e-86da-bc6a7cd9c113

Thank you very much.

I was encountering the same frozen results page problem, using Jatos through MindProbe, as reported in a previous discussion. But I accidentally deleted my study while I was trying to find a solution. Then when I imported again, I got this error message and the importing was not sucessful:

My jzip file is complete and workable as I used the same file and can sucessfully import it last time. I am wondering what should I do now? The error message seems a problem on the online server side which I don't have any access to visit. Any thoughts or suggestions are much appreciated. Thank you very much!

I'm trying to use JATOS during my lab sessions where my students conduct experiments. I'm using JATOS 3.8.2 and Lab.JS, along with “General Multiple” links. There are about thirty 3KB images in the experiment.

The first time the students ran the experiment, I had no problems.

However, the second time (in the same room and on the same computers), the students started the experiment, but it froze very quickly and they were unable to continue. On the other hand, students using their own computers (who had not run the experiment the first time) were able to complete it.

Would you have any idea what the problem might be?

I'm trying to integrate my experiment on Jatos into Sona. I followed all the steps described here https://osdoc.cogsci.nl/4.1/manual/osweb/sonasystems/ except the last one, because I do not want to automatically grant the credits. I generated the study link and added "?SONA_ID=%SURVEY_ID%" at the end of the URL. The issues I'm getting are: batch session data is not getting the condition from "pending" and putting into "started" or "finished", so all participants did the default condition so far; I'm not getting the sona_participant_id (it's always -1). Apparently, there is something wrong with this integration.

This is my "sona" code on OpenSesame (version 3.3.11):

This is my "jatos_init_condition" code:

This is my "jatos_finish_condition" code:

This is how my experiment is configured on Jatos:

Example of how my Batch session data is structured:

I would really appreciate if you could help with that. I'm running out of time with this experiment.

Thanks in advance!

We're currently testing JATOS v3.9.8 rollout (in combination with OpenSesame 4.1.6), and noticed that the usual code for picking up query parameters is no longer working, e.g:

E.g. if (window.jatos && jatos.urlQueryParameters.SONA_ID) {
    console.log('Sona information is available')
    var sona_participant_id = jatos.urlQueryParameters.SONA_ID
} else {
    console.log('Sona information is not available (setting value to -1)')
    var sona_participant_id = -1
}

Specifically, window.jatos now returns -1, which obviously breaks the logic. I think I understand the reasoning from No data received on JATOS since update (on mindprobe) — Forum, and either window.jatos = jatos or checking for window.jatos_uuid instead could be a future solution. .

However rolling out with this change would impact many running experiments that would all need to be updated to prevent them from breaking. Has anyone else encountered this issue, and if so, found a good workaround?

Kind regards,

Kerwin

As it seems that it is not possible to manage the components directly from MindProbe, but only locally (to access the study assets folder), I installed JATOS Win Java on my PC and used the localhost9000 JATOS server. 

For now, I have created three fake tasks to test the components and check how they link to each other.

However, while component 1 jumps to component 2, component 2 never jumps to component 3. 

When muting task 2, however, it works well from task 1 to task 3 and the studies are displayed. This is not related to the tasks, as I also tested random task numbers; the problem lies with the serial position of 2 in the component order.

I followed the procedure indicated here: https://www.jatos.org/Combine-two-pre-written-studies-into-one.html and the changes in the html files here:

https://www.jatos.org/Write-your-own-Study-Basics-and-Beyond.html

I created the tasks in Opensesame and exported them as separate JZIP file. From the JATOS local GUI, I imported the studies separately to create three different studies, each with one component. 

This created three folders in my Jatos local folder. From my study assets folder, I copied and pasted:

1) only the HTML files for tasks 2–3, from their study asset folder, into the task 1 folder;

2) the full repertoire for both tasks into the task 1 folder. Neither of these worked, and sometimes studies 2–3 don’t even launch, staying on the 'OSWeb Loading studies' page.

and 3) I directly exported the html files from OpenSesame "Save as autonomous html file". On Jatos, this method worked better as the tasks could be run.

Is there an easy way for me to create this multi-component study ? 

I have been trying to make it work for days, and I even asked a colleague, but now I am stuck. 

Thank you very much for your help with this.

We are using JATOS for multiplayer experiments for quite some time now - and we absolutely love it!

In the past few days it seems that (perhaps because mindprobe has updated its JATOS to 3.9.8) we are experiencing a strange bug - where all our participants get a message saying "Study run is already finished", whereas JATOS status if failed for all of them and we get a message saying "Inactive group member was forced to leave its group". I am not sure how to read into this exactly, so any help would be appreciated!

best,

Isaac

1. It asks the participant to fill in a subject number, which should not happen.
2. It does not redirect the participant to Qualtrics at the end, but gets frozen on a black screen.

These problems do not occur in Firefox or Safari. I am up-to-date with the most recent versions of Chrome and OS Web. Below are screenshots of with the console log at the time it requests a subject number, and at the end of the experiment when it does not redirect to Qualtrics. Attached is the experiment file. Would welcome any suggestions for making the experiment run!

AB_Flow.osexp

10.65 KB

since the Mindprobe update to JATOS 3.9.8, participant data appears to no longer be transferred to or stored on the server. Participants are able to start and complete the study, but in the results "no data" is shown (see below). It also does not appear to be an isse for one specific study.

Is anyone else experiencing this problem and is there a quick fix?

Thanks!

I'm making my first ever OpenSesame experiment, and I'm pretty sure I haven't structured things as it's intended. Essentially, I have 12 studies on Mindprobe/JATOS that are all slightly different variants of the same experiment. I did this just to get 12 different links that I could give to people. Each of these 12 studies contain only one component, which is the entire experiment. When I run the studies from my view on Mindprobe, there's no problem. However, whenever I use a link (General Multiple) two things can happen:

1. If I have disabled "Allow reload" for my single component, the experiment freezes at the data transfer by the end of the run. On the result page it will say "It is not allowed to reload this component". Sometimes it shows errors in the browser console as well.
2. If I have enabled "Allow reload", the experiment finishes properly but the results get messy with several component results featuring duplicate data.

So far, I cannot understand what is causing these component reloads. I would of course like to avoid them. Again, I'm rather new to this and my effort has mostly gone to figuring out OpenSesame and OSWeb. I have likely misunderstood a lot about JATOS. Any help is appreciated.

Thanks!

We are going to conduct an online experiment using JATOS. However, it can only record 5M of data size, how can we extend this limit?

Thanks and best regards,

Lawliet

Of course, I can just set up a new experiment. No data has been lost. However, I think it would be important to figure out what happened here. Has anyone had similar experiences?

semi-frequently when I test run my components whether locally or on mindprobe it asks whether cookies are allowed in the browser (firefox) with an error in the console saying jatos cookies id 1, 2 or 6 are outdated, is this a local issue for me or is it likely to happen to other participants running the same experiment?

hopefully someone can help,

thank you,

Adrien

all my JATOS links of every experiment suddenly return 404 Not Found across all my studies and batches. Even Run → Preview inside JATOS results in 404. I didn’t change anything on my side (studies, batches, or components are Active).

Examples:

Full link (fails): https://jatos-server.web.rug.nl/publix/ycNvEauEVVE

Could you please help me with this? This is blocking live data collection...

Thank you!

Best

Runyu

I have tried to use the API for a recent study which requires frequent downloading of data and subsequent updating of study asset files. It works fine on the cortex test server.

When I tried it on mindprobe, I run into the problem that the server always returns a (the same) HTML file, regardless of which API endpoint I call. It seems to be the Login page (or redirect to one).

I found this to be a known problem with web servers, which redirect all requests to the Login page when a user is not logged in (probably checked via a cookie). Instead, calls to the API are authorized with the tokens, so there should be no redirect. Perhaps web server needs to be configured such that it does not redirect calls to URLs with "/api/" in it?

Is it me, or could it be a server issue?

Thank you for looking into it!

Christoph

However, now it looks as in the attached picture (even though the main study page does indicate we have 100+ results):

The reason why I noticed this is that I tried setting up a new study and was running into the same issue. Any help is much appreciated!

I have installed JATOS on my institutional server, and I can officially access it via web browser. It brings up a login page asking for username and password, as it expected from my experience on a local server. I can type in the boxes, but the buttons for sign in and such seem inactive, and the page is unresponsive. Anyone have any experience with this?

Thanks!