Rationale

Reconcile and consolidate multiple index handling code paths into a single method that creates/drops indices on provisioned domains. And, especially, stop trying to guess an index's name when dropping it.

In the related PR, AbstractAuditTypeProvider.updateIndices() (which was already inspecting index meta data to determine indices to add/drop based on their characteristics) was taught to pass existing index names on to StorageProvisioner for dropping. This PR pushes the audit implementation into StorageProvisioner and switches callers of the redundant methods to use it.

The PR includes a bonus side quest that fixes domain URIs and domain URI resolution in a bunch of audit domains to ensure that we can use domain.getDomainKind() with confidence.

Related Pull Requests

• Dropping an index requires callers to provide its name #6633

Tasks 📍

• Move AbstractAuditTypeProvider.updateIndices() code into StorageProvisioner
• Fix Domains getting tagged with the wrong DomainKind
• Junit tests that flag audit DomainKinds with overlapping prefixes
• Switch all callers to use new StorageProvisioner.ensureTableIndices() method
• Remove old code
• Manual Testing / Verify Fix @labkey-chrisj
  • Ideally on a long-running database that's seen multiple upgrades, Before pulling down the feature branch and upgrading:
    • Review the "Admin Console - Audit Log" events for LDAP sync events, Study Security Escalations, Logged query events, Query export events, and EHR Security Escalations (if you have EHR built locally...)
    • Review in the Query Schema Browser queries in the "auditLog" schema corresponding to the above - PremiumLdapAuditEvent, StudySecurityEscalationEvent, LoggedQuery, QueryExportAuditEvent, and EHRSecurityEscalationEvent - Use Folder Filter -> All Folders to show all rows like the above do automatically
  • Sync FBs, build and start the server to run the exp upgrade
    • Re-review all "Admin Console - Audit Log" events and "auditLog" queries from above and verify the same row counts appear after upgrade. And no errors, of course.
  • On some lists and datasets, for some columns, check the Advanced checkbox that requires unique values. (Of course the values need to be unique or the table needs to be empty.) Check and save. Uncheck and save. Check and save. This will create and drop indices.
• Needs Automation: No