Suddenly some workflows stopped working, and I can not open them, saying they could not be found. Network tab shows 404 trying to get it /rest/workflows/XXXMhEXXXq9XXX, but I checked in the database, and all of them still exist.
What is the error message (if any)?
Could not find workflow
Could not load the workflow - you can only access workflows owned by you
Please share your workflow
If relevant, I can see this error in the console
2025-07-23T14:49:26.005600574Z There was a problem in âGmail Triggerâ node in workflow âXXXXXâ: âundefinedâ
2025-07-23T14:49:46.189595382Z Error while saving insights metadata and raw data
That workflow contains a simple Gmail Receive Email Trigger that used to long for a long time, and hasnât been touched for months. Today it suddenly broke, and I assume other workflows that are not working are the ones importing it.
Any idea how to fix it? Because I canât open the workflow to see it.
Shouldnât n8n grafcefully handle this case? Instead of breaking the entire workflow and showing a wrong error about ownership?
It appears being updated 1 day ago (even though I did not), but there is another user that has n8n member access to this dashboard (even though, they donât have access to any of the existing credentials, so they shouldât be able to modify them in any way).
If I try to click to open the credential, I get an error:
Problem loading credential
Credential with ID âXXXFilk3kepQbâ could not be found.*
How could a member take over the credentials of a project from the owner?
Where in the DB are the credentials â user relationships? I see in shared_credentials it is still assigned to the owner/same projectid:
credentialsId projectId role createdAt updatedAt
XXXX9Filk3kepQb XXXPFDKvE51TDu credential:owner 2025-05-02T14:40:00.096Z 2025-05-02T14:40:00.096Z
Ok, so the issue is that if I do
SELECT id FROM credentials_entity WHERE id ='XXXXm9Filk3kepQb'; - This doesnât return anything SELECT id FROM credentials_entity WHERE id LIKE '%XXXXm9Filk3kepQb%'; - This works
BUT, those values are identical, there are no extra white-space characters, and the hex encoding of those strings is the same.
This looks to be some Postgres error (?) If it matters, I also switched to pgvector (to be able to have vector store).