issues with saving metadata, previewing files, publishing

80 views
Skip to first unread message

Deirdre Kirmis

unread,
Apr 29, 2021, 12:33:43 PM4/29/21
to dataverse...@googlegroups.com

We have recently started having issues in our Dataverse installation with some of the datasets … errors when trying to save metadata (just “unable to update metadata” error), and unable to preview certain files (mainly .txt files) … gives this error:

 

“Unable to retrieve file.Please try again. The most common issue is that your login has timed out. If the problem persists, please contact the support team of this data repository. Please include any status code included at the end of this message: error". However, other files will preview just fine.

 

We have recently upgraded to v5.3, and since then have also had an issue with publishing a dataset that had a guestbook enabled (using the “Publish Current Version" Option). Both datasets that we are having issues with have a guestbook configured.

 

Also, the one with the metadata save and preview issues currently has just had about 17TB of data uploaded to it … zip files that are 155-365G … and is unpublished .. but it has a guestbook enabled. The author took 3-4 months to upload all the files, during which time we upgraded to 5.2 and 5.3. This dataset has had numerous changes made to the files, including uploaded about 5000 and deleting them, as well as the large file uploads using direct upload. Wondering if something is corrupted somewhere with all of those unpublished changes, or if we just need to upgrade to 5.4? It doesn’t seem like we had these problems before upgrading to v5.3, so wondering if something went wrong somewhere with the payara upgrade, or permissions?

 

Any help is greatly appreciated … we don’t want to try to publish the “big” dataset until we can get the metadata saved and the preview issues resolved.

 

Night Owl

danny...@g.harvard.edu

unread,
May 3, 2021, 2:23:19 PM5/3/21
to Dataverse Users Community
Hi Deirdre, 

Hopefully others will have input here, but two quick thoughts/questions:

- There was an update script in 5.4 related to the update current version command, under the "notes for admins" section. https://github.com/IQSS/dataverse/releases/tag/v5.4. Taking a closer look at the release note, I don't think this is related to what you're seeing, but since you mentioned "Publish Current Version" I thought I'd at least highlight it. 
- Have y'all done anything with custom roles in your Dataverse installation? 

Thanks!

Deirdre Kirmis

unread,
May 4, 2021, 2:35:25 AM5/4/21
to dataverse...@googlegroups.com

Hi Danny,

Thanks so much for the response. I do think part of the issue that we had originally is with the “update current version” command .. when we tried to save the dataset using that option (which was available as superuser), it gave an error that it couldn’t publish and would never save. We ended up just deleting the draft and changing the description directly in the database, as the user did not want the version/release to change. We are preparing to upgrade to 5.4.1 which hopefully will help!

 

We do have some global custom roles defined .. is that related?

 

For the issue with saving metadata, I had a separate conversation with Jim Myers, who explained that sometimes when editing a dataset, adding files or updating the metadata can trigger a reindex. If another change is made quickly, a conflict can occur which may have caused the update error that we experienced (and the “OptimisticLockException” errors that we are seeing in the logs). The person editing the dataset made a LOT of changes, and when he tried to save the metadata at the end, he got the “unable to update metadata” error and then lost all of his changes. =(  After Jim’s info, he went back through and made the changes again, just a few at a time, but saving periodically as he went, and was able to save them all. I guess we need to tell our users to make edits slowly! =D

 

For the previewer issue, we also found (thanks to Jim!) that we needed to change the CORS policy on our S3 bucket as it was only allowing our site URL and so some of the previewer domains that are called were not getting through. Once we did that, we could preview all of the .txt files again. Jim also suggested increasing the connection-pool-size for our default s3 store, which seems to have helped with some of the other S3 timeout errors that we were seeing.

 

So sorry, I should have updated this thread earlier with Jim’s notes in case anyone else runs into these things! And, thanks to Jim for all of the help … it is much appreciated!!

 

dk

--
You received this message because you are subscribed to the Google Groups "Dataverse Users Community" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dataverse-commu...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/dataverse-community/57df46f6-39c1-4cfd-9c0b-bdbc03f706b3n%40googlegroups.com.

Reply all
Reply to author
Forward
0 new messages