fix(testing): cancel draining after fatal silo errors

[ReubenBond]

Fatal silo errors represent process loss. Graceful draining in the test host can wait on unreachable peers and delay cleanup.

Stop the affected host with a pre-canceled drain budget while preserving in-process cleanup. Log single and aggregated cancellation at Debug and actual stop failures at Error. The regression probe verifies canceled drain tokens, one host stop for repeated fatal notifications, and log classification for single, multiple, and nested stop failures.

Extracted from commit 48302ff632bbc7f99eadc0ce366b08c2ca4d9b88 in #10236 as a standalone prerequisite for that PR. Addresses the fatal-drain policy implicated in #11286; end-to-end reproduction of that streaming shutdown remains a separate verification gap.

[Copilot]

Copilot review overview

🟡 Changes recommended

Cancellation aggregates may be logged at Error, and the regression test does not verify the required logging behavior.

Get a fresh assessment by requesting another Copilot review.

Review effort: Lite
Findings: 1

Open (1)

• Handle hosted-service cancellation aggregates as expected · New

What changed in this PR

Updates test-host fatal-silo handling to cancel graceful draining promptly while preserving cleanup.

Changes:

• Uses a pre-canceled token for fatal-error shutdown.
• Adds regression coverage for cancellation and duplicate stop scheduling.

File	Review summary
test/​TestInfrastructure/​Orleans.TestingHost.Tests/​TestClusterFatalErrorHandlerTests.cs	Tests cancellation and single-stop behavior; logging assertions are still needed (nit, 1 vote).
src/​Orleans.TestingHost/​TestClusterFatalErrorHandler.cs	Needs aggregate cancellation handling to preserve Debug logging while retaining Error logging for real failures (moderate, 2 votes).

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[github-actions]

Code coverage

Metric	Pull request	Current main	Variance
Lines	82.10% (112,247 / 136,719)	82.12% (112,272 / 136,709)	-0.0243 pp
Branches	71.29% (32,298 / 45,305)	71.32% (32,306 / 45,295)	-0.0334 pp

Report-only conclusion: regressed.

The current-main baseline is commit 2e40fa8a3b and uses the same reviewed coverage matrix.

Coverage combines every CI test matrix job, including providers, CodeGen, .NET 8/10, Linux, Windows, and macOS, using canonical physical source and branch identities.

The comparison remains report-only while normal line and branch variance is calibrated.

Coverage details

• Pull request commit: e0517181dd
• View the pull request test run and coverage artifacts
• View the trusted coverage reporting run

[Copilot]

Copilot review overview

🔵 Needs a closer look

The nested aggregate test case has an assertion mismatch that must be corrected before approval.

Review effort: Lite
Findings: None

Resolved since last review (1)

• Handle hosted-service cancellation aggregates as expected