[SOLR-15570] JDBC table metadata should include fields declared in the schema (not dynamic) even if they are empty in the index

• Type: Improvement
• Status: Closed
• Priority: Minor
• Resolution: Fixed
• Fix Version/s: 9.0, 8.10
• Component/s: Parallel SQL
• Labels: RobustSQL
• Assignee: Timothy Potter
• Reporter: Timothy Potter
• Created: 28/Jul/21 21:15
• Updated: 28/Sep/21 20:27
• Resolved: 30/Jul/21 16:29

Description

Let's say I have field stringx declared in my schema but is empty (no docs in the index yet).

Currently, the SolrSchema uses Luke to generate the table metadata and since stringx is empty, it's not included in the table metadata.

select id, stringx from collection

will fail with:

Error while executing SQL "SELECT id, stringx FROM collection1": From line 1, column 12 to line 1, column 18: Column 'stringx' not found in any table
	at org.apache.solr.client.solrj.io.stream.JDBCStream.open(JDBCStream.java:286)
	at org.apache.solr.client.solrj.io.stream.ExceptionStream.open(ExceptionStream.java:52)

Of course, if an application uses JDBC and describe table then they won't see stringx so no harm no foul. However, some apps / users may know that stringx is a field in the schema, leading to some mismatch in expectations and results.

Issue Links

• links to: GitHub Pull Request #241
• links to: GitHub Pull Request #244
• links to: GitHub Pull Request #2539
• links to: GitHub Pull Request #2544