S3 and dataset_logo_original · troubleshooting

Thanks for the info @Oliver Bertuch , could the feature flag disable-dataset-thumbnail-autoselect set to On solve this problem? I never configured a feature flag before, is asadmin create-jvm-options "-Ddataverse.feature.disable-dataset-thumbnail-autoselect=On" the correct way?

Oliver Bertuch (Nov 12 2025 at 08:49):

@César Ferreira I don't think so - the log messages you're seeing are triggered by the API endpoint /api/datasets/{id}/logo.

More details: the JSF UI will access this API endpoint from here, which is called anytime someone accesses a Dataset page in "view" mode. This API endpoint will always reach out the storage system, so there's no way for you to stop it with the feature flag. The flag is only influencing which thumbnail will be automatically selected if no explicit one has been configured by the dataset owner/admin.

César Ferreira (Nov 12 2025 at 09:47):

But if that feature flag is set to On a new dataset will use the generic thumbnail by default if no thumbnail is chosen by the dataset owner/admin, or am I understanding this wrong? This way that request would return null instead of an URL for the logo. I assume the option to use a generic thumbnail is configured with this feature flag because there is no option in the Thumbnail+Widgets page for that purpose.

[2025-11-01T16:09:06.292+0000] [Payara 6.2025.2] [WARNING] [] [edu.harvard.iq.dataverse.ThumbnailServiceWrapper] [tid: _ThreadID=6294 _ThreadName=http-thread-pool::http-listener-1(33)] [timeMillis: 1762013346292] [levelValue: 900] getDatasetCardImageAsUrl(): Failed to initialize dataset StorageIO for minio://10.82210/RZGI5S (S3AccessIO: Failed to cache auxilary object : dataset_logo_original)
[2025-11-07T09:37:01.756+0000] [Payara 6.2025.2] [WARNING] [] [edu.harvard.iq.dataverse.dataset.DatasetUtil] [tid: _ThreadID=6338 _ThreadName=http-thread-pool::http-listener-1(77)] [timeMillis: 1762508221756] [levelValue: 900] getThumbnail(): Failed to initialize dataset StorageIO for null (getDataAccessObject: null or invalid datafile.)

The first log matches the enpoint you shared. Regarding the second log, I noticed that in the dataset table the column storagedriver is set to null for every entry, wasn't this supposed to have the storage driver id I configured? I followed this part of the documentation to setup S3 and the uploads are working.

Philip Durbin 🚀 (Nov 12 2025 at 15:43):

César Ferreira (Nov 13 2025 at 09:33):

César Ferreira (Nov 17 2025 at 17:49):

Hi, an update regarding this issue. After configuring the feature flag I tried to create new datasets and the column usegenericthumbnail was set to True. I also had to set that value manually for existing datasets that didn't have a thumbnail. Until now it seems to be working as expected, the requests to S3 for an missing file stopped after these changes.
Now I think I found another bug. When creating a new dataset I can't upload files, the progress bar doesn't move, it stays like in the image bellow, and there is feedback for the operation (no logs in payara, s3 or browser console). If I save the new dataset without uploading files, after that I am able to upload files.
image.png

César Ferreira (Nov 17 2025 at 17:53):

Philip Durbin 🚀 (Nov 17 2025 at 18:14):

César Ferreira (Nov 18 2025 at 09:36):

This only happens during the dataset creation process, after saving the dataset the upload works. I only found this endpoint to create a dataset, but I don't see an option to also upload a file to test this situation.

Philip Durbin 🚀 (Nov 18 2025 at 13:49):

Oh, I see. This "create the dataset and upload files all at the same time" feature is going away with the SPA, by the way. See "Changes from the original Dataverse JSF application" at https://github.com/IQSS/dataverse-frontend

Stream: troubleshooting

Topic: S3 and dataset_logo_original

César Ferreira (Nov 11 2025 at 11:43):