fix(OMN-15978): reconcile runner fleet config to live 88-runner fleet

[jonahgabriel]

Summary

The .201 self-hosted runner host (omninode-pc.tail75df5e.ts.net) was scaled 72 -> 88 by the fleet controller on operator authorization; all 88 runners are registered with GitHub org OmniNode-ai and working. The repo's authoritative fleet config (config/runner_fleet.yaml) and compose file (docker/docker-compose.runners.yml) were left at 72.

Danger this PR removes: scripts/deploy-runners.sh rsyncs the repo's compose file over the host and runs docker compose ... up -d --remove-orphans. With the repo at 72, the next deploy would have deleted the live runners 73-88 out from under the fleet.

Ordering: live-first, reconcile-second. The host is already correct (88 running, registered, healthy); this PR only brings the repo's declared source of truth into agreement with it, so the deploy script stops being a latent fleet-shrinking hazard.

Changes

• config/runner_fleet.yaml: expected_count/burst_count 72 -> 88, plus the adjacent doc comments that describe current fleet architecture (not dated historical-measurement snapshots, which are intentionally left untouched — e.g. the 2026-08-14 "MEASURED STATE: 72/72 runners warm" tool-cache note still says 72 because that's what was literally measured then).
• docker/docker-compose.runners.yml: appended omninode-runner-73 .. omninode-runner-88 service blocks and their runner-N-creds named volumes, matching omninode-runner-72's block shape exactly (5 volume mounts, container_name, RUNNER_NAME, the repo's !!merge <<: *runner-base / *runner-env anchor convention). Also updated the resource-allocation and egress-concurrency doc comments that state "72 runners" as current architecture.
  • Note on live-host byte-shape: the live host's ad-hoc 73-88 additions use plain <<: *runner-base (no explicit !!merge tag), while runners 1-72 on both the host and the repo use !!merge <<: *runner-base. Both forms are semantically identical YAML merge keys — docker compose config renders them the same. I matched the established repo/host convention (!!merge) for internal file consistency rather than the inconsistent ad-hoc host edit.
• tests/unit/observability/runner_health/test_runner_fleet_config.py: the two tests that hardcoded the fleet size as 72 are updated to 88, including renaming test_runner_compose_reconciled_to_saturation_scale_72_fleet -> ..._88_fleet.

Verification

• docker compose -f docker/docker-compose.runners.yml config --quiet validates cleanly with dummy env vars (RUNNER_TOKEN, DEPLOY_RUNNER_TOKEN, DEPLOY_RUNNER_OMNI_HOME, DEPLOY_RUNNER_OPERATOR_ENV_FILE, OMNI_HOME) — 89 services (88 runners + omninode-deploy-runner), 88 runner-N-creds volumes.
• Targeted tests all green: tests/unit/observability/runner_health/test_runner_fleet_config.py, tests/integration/observability/test_runner_fleet_config_contract.py, tests/ci/test_check_runner_fleet_image_drift.py, tests/scripts/test_deploy_runners_sync_paths.py, tests/unit/observability/runner_health/test_cli_runner_health.py — 48/48 passed.
• pre-commit run --files <changed files> passes end to end (ruff format/check, yamlfmt, ONEX architecture/contract/pattern/naming validators, SPDX, deploy-scope DoD, etc.).
• Live host confirmed via ssh jonah@192.168.86.201: docker-compose.runners.yml on the host has 89 container_name: entries and runners omninode-runner-73..88 present and registered, matching what this PR now declares.

Ticket

OMN-15978 is the prior saturation-scale ticket (48 -> 64 -> 72). No fresher Linear ticket exists specifically for the 72 -> 88 reconciliation (searched "runner fleet scale", "88 runner", "reconcile runner fleet" — nothing newer matched), so it's reused here per the OMN-15978 pattern of prior fleet-config reconciliations (OMN-12582 did the same for the 48-runner fleet).

dod_evidence

Deploy-scoped surface (docker/docker-compose.runners.yml) touched; verification evidence is the test run + docker compose config output + pre-commit pass cited above. This is a pure config-reconciliation change (no new behavior) bringing the repo in line with an already-live, already-approved host state.

Evidence-Ticket: OMN-15978
Evidence-Source: OCC#6698

Summary by CodeRabbit

• Infrastructure
  • Expanded the runner fleet from 72 to 88 steady-state instances.
  • Increased burst capacity to support up to 88 runners.
  • Added persistent credentials and services for the additional runners.
• Documentation
  • Updated runner fleet capacity, resource, and egress references to reflect 88 runners.
• Tests
  • Updated fleet validation checks to verify all 88 services and credential volumes.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: e3d6fa86-c116-4af4-8696-72be9027db62

📥 Commits

Reviewing files that changed from the base of the PR and between b99e1a9 and 6415a7e.

📒 Files selected for processing (3)

• config/runner_fleet.yaml
• docker/docker-compose.runners.yml
• tests/unit/observability/runner_health/test_runner_fleet_config.py

Included review availability: 0 reviews are currently available. Your included PR review attempts over the past 7 days set your current allowance at 1 review per hour.

---

📝 Walkthrough

Walkthrough

The runner fleet increases from 72 to 88 steady-state runners. Configuration, Docker Compose services, credential volumes, capacity documentation, and reconciliation tests now use the 88-runner topology.

Changes

Runner fleet expansion

Layer / File(s)	Summary
Fleet capacity contract config/runner_fleet.yaml, docker/docker-compose.runners.yml, tests/unit/observability/runner_health/test_runner_fleet_config.py	Expected and burst capacity change to 88 runners. Related resource, egress, mirror, and test documentation now references 88 runners.
Runner Compose topology docker/docker-compose.runners.yml	Runner services omninode-runner-73 through omninode-runner-88 and their dedicated credential volumes are added.
Fleet reconciliation validation tests/unit/observability/runner_health/test_runner_fleet_config.py	The reconciliation test validates 88 contiguous runner services and credential volumes with the existing health-check and credential requirements.

Estimated code review effort: 2 (Simple) | ~10 minutes

Merge Risk: ⚪ Minimal · up to 6415a

This change reconciles the declared runner fleet from 72 to 88 and adds the matching runner services and tests; no actionable merge-blocking risk remains after normal checks and review.

Possibly related PRs

• OmniNode-ai/omnibase_infra#2735: Modifies the same runner fleet configuration, Docker Compose topology, and reconciliation tests.

Suggested labels: ci:ready

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 66.67% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly and concisely describes reconciling the runner fleet configuration with the live 88-runner fleet.

✨ Finishing Touches 💡 1

📝 Generate docstrings 💡

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch jonah/omn-15978-reconcile-fleet-72-to-88

---

_{Comment @coderabbitai help to get the list of available commands.}

[github-actions]

⚠️ Hostile Reviewer — DEGRADED (informational)

Blocking findings (critical): 0
Total findings: 0
Models succeeded: none

Note: All reviewer models failed or were unavailable. Degraded results are informational during the pilot phase (OMN-8468/OMN-8524) and do not block merge. Error: cli_review exited non-zero (fewer than 2 models succeeded — partial/total outage — or binary error)

---

Gate semantics (pilot phase)

Verdict	Meaning	Blocks merge?
passed	No critical findings	No
blocked	CRITICAL findings found	Yes
degraded	All models unavailable (infra)	No (pilot)

Powered by omniintelligence.review_pairing.cli_review — node-based adversarial review via HandlerLlmCliSubprocess (OMN-8468/OMN-8524)