[Prometheus.HttpListener] Fix shutdown hang

[martincostello]

Changes

Avoid disposal race in PrometheusHttpListener that can cause shutdown to hang indefinitely and flaky tests (1, 2, 3).

Merge requirement checklist

• CONTRIBUTING guidelines followed (license requirements, nullable enabled, static analysis, etc.)
• Unit tests added/updated
• Appropriate CHANGELOG.md files updated for non-trivial changes
• Changes in public API reviewed (if applicable)

[opentelemetry-pr-dashboard]

Pull request dashboard status

Merged · refreshed 2026-08-04 21:27 UTC

Status above doesn't look right?

• Anything look wrong? Report it with what you expected; it helps us improve the dashboard.

[codecov]

Codecov Report

❌ Patch coverage is 69.23077% with 4 lines in your changes missing coverage. Please review.
✅ Project coverage is 90.49%. Comparing base (2ae52ae) to head (5d86411).
✅ All tests successful. No failed tests found.

Files with missing lines	Patch %	Lines
....Prometheus.HttpListener/PrometheusHttpListener.cs	69.23%	4 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #7587      +/-   ##
==========================================
- Coverage   90.51%   90.49%   -0.02%     
==========================================
  Files         290      290              
  Lines       16029    16039      +10     
==========================================
+ Hits        14509    14515       +6     
- Misses       1520     1524       +4     

Flag	Coverage Δ
unittests-Project-Experimental	90.49% <69.23%> (-0.07%)	⬇️
unittests-Project-Stable	90.53% <69.23%> (-0.08%)	⬇️
unittests-UnstableCoreLibraries-Experimental	51.35% <69.23%> (-0.01%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
....Prometheus.HttpListener/PrometheusHttpListener.cs	88.30% <69.23%> (-1.76%)	⬇️

... and 2 files with indirect coverage changes

[martincostello]

6 runs with no Prometheus related failures.

There was one failure here on attempt 5, but that has no specific cause logged but appears to be within OpenTelemetry.Tests.

[Copilot]

Pull request overview

This PR addresses a shutdown hang in the Prometheus HttpListener exporter by bounding shutdown waits and ensuring in-flight scrapes fail fast (HTTP 503) once disposal begins, with accompanying test hardening to avoid race-driven flakiness.

Changes:

• Add bounded timeouts to listener shutdown/drain logic and log when the processing loop fails to stop promptly.
• Ensure scrapes that are still collecting when disposal starts return HTTP 503 instead of completing with a successful response.
• Harden integration/unit tests by adding explicit scrape/promtool timeouts and avoiding thread-pool scheduling races during disposal.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 1 comment.

File	Description
test/OpenTelemetry.Exporter.Prometheus.HttpListener.Tests/PromToolFixture.cs	Adds scrape and overall promtool execution timeouts to prevent hangs in container-based validation.
test/OpenTelemetry.Exporter.Prometheus.HttpListener.Tests/PrometheusHttpListenerTests.cs	Makes the disposal-vs-scrape race test deterministic (thread-pool availability + dedicated disposal thread).
src/OpenTelemetry.Exporter.Prometheus.HttpListener/PrometheusHttpListener.cs	Implements bounded shutdown waits and returns 503 when disposal begins mid-collection.
src/OpenTelemetry.Exporter.Prometheus.HttpListener/CHANGELOG.md	Adds Unreleased entries describing the behavioral changes (needs PR link fixups).

Suppressed comments (1)

src/OpenTelemetry.Exporter.Prometheus.HttpListener/CHANGELOG.md:24

• This CHANGELOG entry still uses a placeholder PR reference (#TODO). Please update it to the real PR link before merging.

  ([#TODO](https://github.com/open-telemetry/opentelemetry-dotnet/pull/TODO))

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.