[Nikhil Navakiran (nikhilNava)]

Summary

• Affected PR: Reuse Azure Monitor offline storage for Agent365 #300 (failing benchmark job).
• Pin the Performance workflow to released pytest==9.1.1 and pytest-benchmark==5.3.0 in both candidate and baseline installation steps.
• Avoid pytest 9.2's removal of the private _nodeid attribute still accessed by pytest-benchmark 5.3.0. Without these pins, all four benchmarks fail during fixture setup before any performance measurements run.
• Keep runtime dependencies, SDK code, and benchmark behavior unchanged. No unmerged upstream code or monkey patches are used.

This is independent of #300 and is based directly on main.

Validation

• Ran all four existing performance benchmarks with the exact released pins: 4 passed; benchmark JSON generated successfully.
• Local runtime: Windows, Python 3.14. The Performance CI workflow also passed on Ubuntu/Python 3.12: candidate benchmarks, baseline benchmarks, and the performance comparison all succeeded.
• git diff --check passed.
• Existing OpenTelemetry deprecation warnings and localhost exporter retry messages remain unchanged.

[github-actions]

Performance comparison

Threshold: regressions >15.0% on gating scenarios fail the build. Higher ops/s is better; positive Δ means the PR is slower.

Scenario	Gating	Baseline (ops/s)	Candidate (ops/s)	Δ %	Status
azure_monitor_log	yes	21,928.9	21,934.2	-0.02%	✅
azure_monitor_span	yes	164,203.6	162,022.0	+1.35%	✅
otel_log	no	25,217.5	25,338.9	-0.48%	✅
otel_span	no	47,180.9	47,580.5	-0.84%	✅

[Copilot]

🟢 Approval recommended

No active review issues remain for the performance-workflow dependency pins, though tests were not run during review and Ubuntu/Python 3.12 validation remains for CI.

No open issues 🎉

What changed in this PR

Pins the performance workflow’s test dependencies to avoid the reported pytest compatibility failure, without changing SDK runtime dependencies or benchmark behavior.

Changes:

• Uses pytest==9.1.1 and pytest-benchmark==5.3.0 for both candidate and baseline runs.
• Keeps both sides of the benchmark comparison on the same test-tool versions.

No actionable findings. Tests were not run during this review; Ubuntu/Python 3.12 validation remains for CI.

File	Description
.github/​workflows/​performance.yml	Pins pytest and pytest-benchmark in both installation steps.

_{Review effort: Balanced}