Exercise mixed application capacity and recovery

[valthon]

Summary

The capacity fixture previously measured only HTTP reads and writes. It now verifies tenant-scoped SSE delivery and transactional durable jobs alongside those requests, with bounded normal-load, stress, drain, and recovery phases. Job payload identities and individual event revisions must match acknowledged writes; malformed, missing, duplicated, foreign, or incomplete work fails the report.

A paired raw-report walkthrough records a real configuration bottleneck: a serial claim batch of two took roughly ten seconds to drain 40 jobs, versus roughly 0.4 seconds with batch 128 on the same shared Debug host. This is evidence about polling/batching, not a production capacity claim. The backlog explicitly defers response caching.

Documentation & examples sync

• Canonical testing docs, generated skill reference, changelog fragment, and backlog updated.
• Runnable capacity fixture and negative/live tests updated; raw reports and reproduction commands retained.
• Public-site generation, validation, build, doctor, static/generated checks, and browser smoke passed.
• N/A: public application APIs, SDKs, screenshots, and the three example applications.

Verification

• Mixed-capacity live suite: 12 passed, 16 subtests.
• Default-load short-duration run and paired 120-request-per-phase experiment passed all request, SSE, job, and isolation checks.
• Independent source review; malformed SSE and equal-count incorrect job controls added in response.
• Clean-checkout chapter gate: enabled unit suite plus live capacity/workbench and documentation checks.

Limits remain explicit: local SQLite, one server, closed-loop Python load, synthetic CPU jobs, and warm data. Drain completes before recovery traffic starts. A controlled small-machine saturation report remains open.

[Copilot]

Copilot review overview

🟡 Changes recommended

The runner has a critical missing import and unresolved validation gaps.

Get a fresh assessment by requesting another Copilot review.

Review effort: Lite
Findings: 1

Open (1)

• Import urllib.parse before constructing RealtimeObserver · New

What changed in this PR

Extends the application-capacity fixture with mixed HTTP, tenant-scoped SSE, and durable-job workloads across recovery phases.

Changes:

• Adds mixed workload validation and isolation checks.
• Adds tests, fixture updates, diagnostics, and documentation.
• Updates backlog and changelog.

File	Summary
tools/​application_capacity.py	Mixed workload runner and validation.
tests/​tools/​test_application_capacity.py	Unit and live coverage.
skills/​zigbase-app-genesis/​references/​testing.md	Testing guidance updates.
fixtures/​application-capacity/​main.zig	Durable job fixture configuration.
docs/​testing.md	Capacity testing documentation.
diagnostics/​application-capacity/​batch-sizing/​README.md	Batch-sizing methodology.
diagnostics/​application-capacity/​batch-sizing/​batch-2.json	Batch-2 raw results.
diagnostics/​application-capacity/​batch-sizing/​batch-128.json	Batch-128 raw results.
changelog.d/​mixed-application-capacity.md	Changelog entry.
BACKLOG.md	Capacity backlog updates.

---

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

[Copilot]

Copilot review overview

🟡 Changes recommended

Critical SSE duplicate detection and durable-job kind validation issues remain unresolved.

Get a fresh assessment by requesting another Copilot review.

Review effort: Lite
Findings: 2

Open (2)

• Filter durable-job verification by job kind · New
• Allow settling time to detect late duplicate SSE deliveries · New

Resolved since last review (1)

• Import urllib.parse before constructing RealtimeObserver

[Copilot]

Copilot review overview

🟢 Approval recommended

No unresolved blocking issues were identified.

Review effort: Lite
Findings: None

Resolved since last review (2)

• Allow settling time to detect late duplicate SSE deliveries
• Filter durable-job verification by job kind