[bigquery][jdbc] DatabaseMetaData.getTables() fails for Analytics Hub linked datasets

[abdusalam2015]

[bigquery][jdbc] DatabaseMetaData.getTables() fails for Analytics Hub linked datasets

Thanks for stopping by to let us know something could be better!

---

Environment details

1. API: BigQuery (JDBC / metadata discovery)
2. OS type and version: macOS 14.x (Apple Silicon)
3. Java version: OpenJDK 17
4. Version(s):
   • BigQuery JDBC driver (Simba-based, JDBC 4.2, Google-distributed)
   • Tested with the latest available driver at time of writing

Steps to reproduce

1. Create or subscribe to a dataset using BigQuery Analytics Hub (linked dataset).
2. Connect to BigQuery using the BigQuery JDBC driver.
3. Call JDBC metadata discovery:

   DatabaseMetaData meta = connection.getMetaData();
   meta.getTables(null, "<linked_dataset>", null, null);

Code example

Connection conn = DriverManager.getConnection(jdbcUrl);
DatabaseMetaData meta = conn.getMetaData();
meta.getTables(null, "<linked_dataset>", null, null);

Stack trace

 Error fetching metadata.
Cause: java.lang.NullPointerException
Cannot invoke "java.lang.Boolean.booleanValue()" because
the return value of
"com.google.api.services.bigquery.model.TableList$Tables$View.getUseLegacySql()"
is null

at com.simba.googlebigquery.jdbc42.BQS42DatabaseMetaData.createMetaDataResult(...)

External references such as API reference guides

BigQuery JDBC driver documentation
https://cloud.google.com/bigquery/docs/reference/odbc-jdbc-drivers

BigQuery INFORMATION_SCHEMA (works correctly in this scenario)
https://cloud.google.com/bigquery/docs/information-schema-tables

Related issue (general JDBC metadata discussion):
https://github.com/googleapis/google-cloud-java/issues/2477

Any additional information below

This issue is not related to permissions or IAM.

Using the same service account:

SQL queries against tables/views work

Queries against project.dataset.INFORMATION_SCHEMA.TABLES work

The failure only occurs when using JDBC metadata discovery (getTables).

The same code works when using a normal (non-linked) BigQuery dataset.

This impacts tools that rely on JDBC metadata discovery (e.g. visual schema browsers), while SQL-based access works correctly.

[abdusalam2015]

Filed a corresponding BigQuery bug in Google Issue Tracker - 476947755 and linked it there for backend triage.
Happy to provide additional details or test with a specific JDBC driver version if needed.

[lqiu96]

@logachev @keshavdandeva Would this be something you guys would be able to help look into?

[keshavdandeva]

Hi @abdusalam2015

Thank you for reporting this issue! I tried to reproduce the NullPointerException following the steps provided, using an Analytics Hub linked dataset containing a view, but unfortunately, I wasn't able to trigger the error on my end.

Could you please provide a bit more information on how to reproduce this? Is there anything specific about the views or datasets you are using that might be missing from the original steps (e.g., Authorized Views, specific permissions, etc.)?

Additionally, could you please capture and share the JDBC driver logs with LogLevel=6 when reproducing the issue? This will help us pinpoint exactly what the driver is receiving from the BigQuery API.

[keshavdandeva]

Also, what application are you using? (SQuirreL, DBeaver, Tableau, etc.)
If the intention is to use legacy SQL, then QueryDialect=BIG_QUERY should be added to the connection settings.

[logachev]

@abdusalam2015 thanks for sharing the logs. I removed your comment with logs attached because they contain your project id\dataset name.

[abdusalam2015]

@abdusalam2015 thanks for sharing the logs. I removed your comment with logs attached because they contain your project id\dataset name.

Thanks for taking care of that.
The project ID and dataset names in the logs were already anonymized. So they are not actual production identifiers. No problem though, and appreciate the cleanup.

[abdusalam2015]

Also, what application are you using? (SQuirreL, DBeaver, Tableau, etc.) If the intention is to use legacy SQL, then QueryDialect=BIG_QUERY should be added to the connection settings.

Thanks for the follow-up.

The behavior is reproduced without using any external application such as SQuirreL, DBeaver, or Tableau. I am connecting directly to the BigQuery JDBC driver using a small Python script to isolate the JDBC behavior, confirming that it occurs at the JDBC driver level.

We are also not using legacy SQL, and the issue occurs before any query execution. The exception is triggered during JDBC metadata discovery when calling DatabaseMetaData.getTables().

The key point is that when trying to retrieve the list of tables from an Analytics Hub linked dataset, the JDBC metadata call does not return the tables. However, when using the same script with a normal (non-linked) dataset, the tables are listed correctly.

This suggests the behavior is specific to JDBC metadata enumeration for Analytics Hub linked datasets, rather than related to query dialect or client tooling.

If you run the Python script I shared against a normal dataset, you should see the difference and be able to retrieve the table list successfully.

Please let me know if any additional information would help.

Thanks.

[logachev]

@abdusalam2015 do you see this issue with multiple datasets or a specific dataset? If you create a new dataset with similar configuration, does it have the same failure?

We tried to repro by creating a new dataset, but it worked as expected. I found similar issue report for some dataset created via Terraform, but it was resolved as no-repro.

We asked internally if useLegacySql=null is expected behavior or not, but that team is in a different timezone, so answers are delayed.

Also, if you could share real projectid/dataset with us, that might help Analytics hub team to investigate. You can share it via email bigquery-drivers-feedback@google.com

[logachev]

@abdusalam2015 we shared Simba's private build with you over the email, but didn't hear back. I'll close this bug as inactive.

If you get a chance to try private build and see if issue is still there, please feel free to reopen