Skip to content

Keep SQL usage endpoints whose names differ only in case - #6153

Open
warwickschroeder wants to merge 1 commit into
masterfrom
fix-licensing-case-only-endpoints
Open

warwickschroeder wants to merge 1 commit into
masterfrom
fix-licensing-case-only-endpoints

Conversation

@warwickschroeder

@warwickschroeder warwickschroeder commented Nov 18, 2026 •

Copy link
Copy Markdown
Contributor

What this fixes

Throughput is stored per source: the broker, audit and monitoring each save their own licensing endpoint row under the name they see.

Audit and monitoring report the endpoint name as configured, such as Sales.Billing, while some brokers report the queue in lower case, sales.billing, so one endpoint can arrive in two spellings.

On SQL Server and PostgreSQL the throughput lookup matched those spellings case-insensitively and gave all the throughput to the first one, so the other read as empty, and reading the usage data (the ServicePulse Usage page, the usage report) deleted that row and all its throughput history. The SQL store now returns each spelling its own throughput, as RavenDB does, so nothing reads as empty and nothing is deleted.

#6146 moves that deletion into a daily background job that reads through the same lookup, so it needs this fix too.

  • Each endpoint's throughput is returned under its own sanitized name instead of under the first spelling that matches case-insensitively.
  • Two endpoints that share a sanitized name each get their own entry, which also removes a crash when both have a count for the same day.
  • Docs: none needed; no page under docs/ describes how throughput is matched by queue name (docs/README.md, throughput-collection.md and usage-report.md reviewed).

Tests

  • ServiceControl.Persistence.Tests/Throughput/EndpointsTests.cs: three new shared store tests, run on SQL Server, PostgreSQL, RavenDB and InMemory, pin per-spelling results, an unrequested spelling left out, and one entry per endpoint.
  • ServiceControl.Persistence.Tests/UsageSummaryTests.cs (new): reading the usage summary keeps both case-only spellings and their counts, on SQL Server, PostgreSQL and RavenDB.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants