Karl Magnus Nilsen (Nov 06 2025 at 15:16): Karl Magnus Nilsen (Nov 06 2025 at 15:16):We recently upgraded https://dataverse.no from v6.3 to v6.6. After the upgrade we are not able to publish datasets that were created before the upgrade. The error message says "Your dataset named ... could not be published due to a failure to register, or update the Global Identifier for the dataset or one of the files in it." From the logs I only find something similar "Failed to publicize the identifier doi:10.83125/BILMLY, or to publicize a file in the dataset; notifying the user(s), unlocking the dataset|#]"
Datasets that are created after the upgrade works as expected and are published without error. Is there something that should have been done to these datasets during upgrade? Maybe they are missing some new fields? Or could it be something else?
Philip Durbin 🚀 (Nov 06 2025 at 17:23):@Karl Magnus Nilsen hi! Nothing comes immediately to mind but I'm asking on our internal Slack.
Don Sizemore (Nov 06 2025 at 18:57):@Karl Magnus Nilsen is there any more helpful stacktrace in the Payara log?
Philipp Conzett (Nov 07 2025 at 06:50):@Karl Magnus Nilsen has access to log. Maybe the reason is that the DataCite metadata is not up-to-date. So, we're currently running a modifyRegistrationMetadata.
Juan (Nov 07 2025 at 10:37):@Karl Magnus Nilsen . We are having the same problem sometimes with our v6.5 version. In our case, is associated with the ROR in the CVocConf . After removing problematic ROR (writing the name of the institution), we can publish the dataset.
Philipp Conzett (Nov 07 2025 at 11:08):Thanks, @Juan . Do you mean you remove the value in the Affiliation metadata field and add a new value? Or is there an issue with the ROR configuration?
Juan (Nov 07 2025 at 11:19):I don't investigate the problem yet. But removing the Affiliation value provided by ROR and writing a new value without ROR is working for us.
It only fails with some institutions, so we write the name of the institution without ROR in these cases.
Philipp Conzett (Nov 07 2025 at 11:19):Thanks! I'll try it out on our datasets that failed to be published.
Philipp Conzett (Nov 07 2025 at 11:56):Adding the affiliation (and author name) manually, not fetching from ROR (or ORCID) didn't help in our case.
Juan (Nov 07 2025 at 11:58):I'm sorry, we don't have problems with ORCID, only with some ROR values
Philip Durbin 🚀 (Nov 07 2025 at 12:00):Can you each please start a new thread at https://groups.google.com/g/dataverse-community ? Jim isn't in Zulip but I bet he might have some ideas for us.
Philipp Conzett (Nov 07 2025 at 12:02):I've just sent an email to Jim. :-)
Juan (Nov 07 2025 at 12:31):Philip Durbin 🚀 said:
start
@Philip Durbin 🚀 . Thank you, I will investigate a little more our problem and I will start the thread.
Philip Durbin 🚀 (Nov 07 2025 at 12:40):@Juan if it helps, Jim just posted this in Slack:
Probably #11941 - they did not upgrade their ROR config when ROR switched to using v2 as the default and now have a null value cached triggering a bug in creating the DataCite export (#11782 that missed getting into 6.8). The fix is to update the ROR script to explicitly use v1 (as in the repo) and remove the bad values from the externalvocabularyvalue table (truncating the table is fine).
Juan (Nov 07 2025 at 12:51):Thanks @Philip Durbin 🚀 , I will investigate it.
Juan (Nov 10 2025 at 10:38):Thanks @Philip Durbin 🚀 (and Jim). I have updated the ror.js and people.js files, and also the files modified by #11782.
I checked some ROR that failed before, and now it is working.
I do not remember where I found the other .js files.
Philipp Conzett (Nov 11 2025 at 05:22):We try to turn on granular logging of PID-related issues by running this command in the Docker container:
asadmin --user=${ADMIN_USER} --passwordfile=${AS_PASSWORD_FILE} set-log-levels edu.harvard.iq.dataverse.pidproviders.AbstractPidProvider=FINE
but get this error message:
remote failure: Could not set logger levels for server.
Command set-log-levels failed.
Any idea what is going wrong here? @Philip Durbin 🚀 @Don Sizemore @Oliver Bertuch
Oliver Bertuch (Nov 11 2025 at 07:14):@Philipp Conzett please try using "edu.harvard.iq.dataverse.pidproviders=FINE" - this is usually group (package) based.
Oliver Bertuch (Nov 11 2025 at 07:16):That way you will also see debugging output from the concrete (non-abstract) PID provider.
Philipp Conzett (Nov 11 2025 at 07:52):Thanks, @Oliver Bertuch ! I'll pass this on.
Last updated: Jan 09 2026 at 14:18 UTC