Error when trying to create a new Dataset (NullPointerException) #10557
Comments
|
I have installed:
|
Yes, sure enough, getGlobalId must be null: @faborg thanks for the details about your installation. Is it a new installation? Did you upgrade from Dataverse 6.1? Which PID provider are you using (out of the box, DataCite is selected). |
|
We definitely haven't seen this in 6.2 yet. The only thing different in your setup is a later Payara - the Guides talk about 6.2023.8 and I've run 6.2023.11 without issue. Any details to help us reproduce this would be very welcome. |
|
It's a fresh install (locally in VMs) and I'm using a FAKE DOI Provider. |
|
Is your FAKE provider configured with the new multi-pid-provider JVM/micro-profile options or the original/legacy single-pid-provider database settings? |
|
How do we switch between multi and single? I encountered the same issue when trying to save dataset right after traditional installation following the guide? |
|
I think new installs by default should be using the settings described in https://guides.dataverse.org/en/latest/installation/config.html#dataverse-pid-type. If you can see those when you do a bin/asadmin list-jvm-options from your /usr/local/payara6 dir, you should be using the new multi-pid way. If you have http://localhost:8080/api/admin/settings/ :DoiProvider and related things set, you have the legacy settings. Those shouldn't be needed and should be ignored if you have the multi-pid setup in place (although they would be used if the multi-pid settings are invalid - do you see anything in the log about legacy settings when you start up?) I'm asking so I know how to try and reproduce your issue - the code for new and legacy settings are different so it's possible we have a problem with one and not the other. We could also have an issue in the build script for first time setups. |
|
2024/05/15 |
|
bin/asadmin list-jvm-options: Got this: {"status":"ERROR","message":"Setting :DoiProvider not found"} |
|
Before the log error you sent above there should be a line with "CommandException, when attempting to update the dataset: " with some message. Do you see that? Can you report what it says? |
|
Here is the log right after a restart. |
|
It also includes a try of saving a new dataset |
|
OK - some issue with finding the PidProvider which then leads to a null PID as in the first log you shared. Do you have -Ddataverse.pid.default-provider=fake set? If not, that is probably it and we can look for why that isn't getting set by the installer. If so, I'm not sure what is happening. Doing a bin/asadmin set-log-levels edu.harvard.iq.dataverse.pidproviders.PidProviderFactoryBean=FINE and restarting should add info the the log about whether the fake provider is getting created properly or not. |
|
I got these: |
|
OK - so you need to run |
|
Did it and now I can save new dataset. Thanks! |
|
-Ddataverse.pid.providers=fake |
|
Cool. I do see that we're missing that command from the non-docker install. Not sure how it got missed, but I'll make a PR to add it and add a check to make sure a missing default gets flagged. Thanks for reporting this! |
|
I thought the first line would be enough. Thanks for helping! |
|
In a multipid setup that first line would be a list of several providers, so rather than guessing which is the default (first one?), we have an explicit default setting. Probably could just assume when there's only one though. |
|
That explains it. Thanks for clarifying. |
This works for me, too. Thanks a lot! I'll close this Issue. |
|
I realize @faborg closed this issue (which is fine) and that @qqmyers has a pull request to fix it: I went ahead and reproduced this on Dataverse 6.2 on Rocky 8.9 by running the installer manually (thank you, @donsizemore for gdcc/dataverse-ansible#350 ). When you finish running the installer, you see something like this (I didn't bother to put in a proper hostname): It's quite easy to reproduce the bug. Just try to create a dataset:
On the backend, yes the error is Stacktrace[2024-05-16T16:14:40.577+0000] [Payara 6.2023.8] [WARNING] [AS-EJB-00056] [jakarta.enterprise.ejb.container] [tid: _ThreadID=90 _ThreadName=http-thread-pool::jk-connector(1)] [timeMillis: 1715876080577] [levelValue: 900] [[ A system exception occurred during an invocation on EJB EjbDataverseEngineInner, method: public java.lang.Object edu.harvard.iq.dataverse.EjbDataverseEngineInner.submit(edu.harvard.iq.dataverse.engine.command.Command,edu.harvard.iq.dataverse.engine.command.CommandContext) throws edu.harvard.iq.dataverse.engine.command.exception.CommandException]][2024-05-16T16:14:40.579+0000] [Payara 6.2023.8] [WARNING] [] [jakarta.enterprise.ejb.container] [tid: _ThreadID=90 _ThreadName=http-thread-pool::jk-connector(1)] [timeMillis: 1715876080579] [levelValue: 900] [[ jakarta.ejb.EJBException: Cannot invoke "edu.harvard.iq.dataverse.pidproviders.PidProvider.generatePid(edu.harvard.iq.dataverse.DvObject)" because "pidProvider" is null [2024-05-16T16:14:40.588+0000] [Payara 6.2023.8] [SEVERE] [] [edu.harvard.iq.dataverse.DatasetPage] [tid: _ThreadID=90 _ThreadName=http-thread-pool::jk-connector(1)] [timeMillis: 1715876080588] [levelValue: 1000] [[ [2024-05-16T16:14:40.590+0000] [Payara 6.2023.8] [WARNING] [] [jakarta.enterprise.resource.webcontainer.faces.lifecycle] [tid: _ThreadID=90 _ThreadName=http-thread-pool::jk-connector(1)] [timeMillis: 1715876080590] [levelValue: 900] [[ [2024-05-16T16:14:40.598+0000] [Payara 6.2023.8] [SEVERE] [] [jakarta.enterprise.resource.webcontainer.faces.context] [tid: _ThreadID=90 _ThreadName=http-thread-pool::jk-connector(1)] [timeMillis: 1715876080598] [levelValue: 1000] [[ Again the PR that we hope will fix this is this one: |
When I try to create a new Dataset, after clicking on "Save Dataset" the form is grayed out and no data is saved. Is this a bug? Help would be highly appreciated.
The payara-log contains the following exception:
The text was updated successfully, but these errors were encountered: