Subject Group endpoint (staging) dislikes specific accounts

[shaunanoordin]

Functionality Issue

Affects: /grouped endpoint, on Panoptes staging
Notes: user-specific ; confirmed only on staging ; potentially temporary?

Here's a puzzler for you. The /grouped endpoint on Panoptes staging (as used on the Survos Testing 2020 project) will give an error that reads "Number of subject_ids does not match the group_size", but only if you're logged in as darkeshard. (That's me! Staging user ID is 1325361)

Testing

Replication:

• Login as darkeshard and go to (the Survos Testing 2020 project) on staging
  • Expected: the Classify page should should the Subject Group Viewer
  • Actual: nothing is displayed. Inspecting the Network tab, we see that the /grouped endpoint has returned an error message.
• Copy darkeshard's bearer token. It's time to go postal(man).

Replication (Postman 1):

• Using Postman, submit a GET request (https://panoptes-staging.zooniverse.org/api/subjects/grouped?num_columns=5&num_rows=5&workflow_id=3412&http_cache=true) with...
  • Accept: application/vnd.api+json; version=1
  • Content-Type: application/json
  • Authorization: (darkeshard's bearer token)
• Submit.
  • Expected: 200 { "subjects": [ ... ] }
  • ❌ Actual: 422 { errors: [ "message": "Number of subject_ids does not match the group_size" } ]

Replication (Postman 2):

• Using the same Postman GET request, make the following changes:
  • Remove Authorization (i.e. make a GET request as an anonymous user)
• Submit.
  • ✔️ Results: 200 { "subjects": [ ... ] }
• Using the same Postman GET request, make the following changes:
  • add Authorization: (zootester1's bearer token) (with zootester1 being one of our many random tester accounts.)
• Submit.
  • ✔️ Results: 200 { "subjects": [ ... ] }

PS: if you're curious, submitting an invalid bearer token to Panoptes causes the /grouped endpoint to just shrug and go, "eh, that's fine," and still give valid results

Analysis

• This problem seems to have occurred only on Tue 20 Jan 2026. I was actually testing the /grouped endpoint quite fine on 19 Jan 2026.
• I was with @Tooyosi when I encountered this error, and he was around when I consistently replicated it, so I'm assured I'm not insane even if this problem magically resolves itself after a while.
• This issue was tested across two different computers (Shaun's and Toyosi's); using different Zooniverse accounts; using Chrome (Incognito) and Postman; and the only common factor is the darkeshard account.
• Using Postman discounts issues with the Chrome browser, or browser-related cookies.

Status

On its own, this problem is pretty funny because it just shows how cursed I am, specifically, when it comes to weird bugs.

But in context, this problem seems related to an issue I've been investigating for the past two weeks (Freshdesk ticket 27111 / FEM issue 7176), where a specific volunteer (citsci-rancho) consistently sees problems that no other volunteer does. In that investigation, I've narrowed the issue down to either the Project resource getting cached (which is its own sub-investigation) or that user specifically getting wonky responses from Panoptes. (Their User Project Preferences or their Project Roles, most likely.)

If this is the case, then... I'm not sure. Replication will be a challenge if it's only constrained to specific users, and I don't have enough information on Panoptes to even hypothesise what might be the cause.

I'm passing the baton to @Tooyosi to see what the next investigation steps should be, if there are any.