ci(regression): expose local scenario groups and owned restart checks

.github/workflows/daily-endurance.yml

@@ -5,6 +5,15 @@ on:
     # 03:23 Asia/Shanghai; GitHub schedules use UTC and may start late.
     - cron: '23 19 * * *'
   workflow_dispatch:
+    inputs:
+      scenario_groups:
+        description: 'Optional local groups, space separated; empty runs only base cases'
+        type: string
+        default: 'faults resources lifecycle'
+      restart_services:
+        description: 'Check owned Core and Runtime restarts once before resource sampling'
+        type: boolean
+        default: true
 
 permissions:
   contents: read
@@ -42,9 +51,16 @@ jobs:
           chmod +x lbctl_linux_amd64
           docker pull python:3.12-alpine
       - name: Run repeated scenarios and recovery checks
+        env:
+          SCENARIO_GROUPS: ${{ github.event_name == 'schedule' && 'faults resources lifecycle' || inputs.scenario_groups }}
+          RESTART_SERVICES: ${{ github.event_name == 'schedule' || inputs.restart_services }}
         run: |
+          read -r -a groups <<< "$SCENARIO_GROUPS"
+          args=()
+          if [[ "$RESTART_SERVICES" == true ]]; then args+=(--restart-services); fi
           uv run --no-sync python tools/regression_ci.py \
-            --suite daily --lbctl .endurance-bin/lbctl_linux_amd64 --output .qa/daily
+            --suite daily --scenario-groups "${groups[@]}" "${args[@]}" \
+            --lbctl .endurance-bin/lbctl_linux_amd64 --output .qa/daily
       - name: Publish observations
         if: always()
         run: |
@@ -68,6 +84,12 @@ jobs:
               out.write(f"Daily regression: {outcome}\n\nCore: `{os.environ['GITHUB_SHA']}`.\n\n")
               out.write('One 1800-second scenario/resource budget, including 120 seconds reserved for drain/cleanup and 60 seconds for cooldown. The outer process guard is 1860 seconds; deployment and plugin installation precede the scenario clock.\n\n')
               out.write('Real Core/Runtime/published plugins/Box; mock generation and embedding. Throughput/P95 are observations; business, health and resource failures still fail this non-required job.\n\n')
+              out.write(f"Enabled local groups: {gate.get('scenario_groups', 'not reported')}.\n\n")
+              restart = gate.get('restart_checks')
+              if restart:
+                  out.write(f"Owned-service restart checks: {restart.get('outcome', 'not reported')}; {restart.get('elapsed_seconds', 0):.2f}s, deducted from the shared budget. Resource baselines begin after restarts.\n\n")
+                  for error in restart.get('errors', []):
+                      out.write(f"Restart check error: {error}\n")
               for plugin in manifest.get('plugins', []):
                   out.write(f"Published plugin: `{plugin['id']}@{plugin['version']}`; installation readback: `{plugin.get('installed', 'not confirmed')}`.\n\n")
               out.write(f"Required first pass complete: {suite.get('first_pass_complete', 'not reported')}; cycles: {len(suite.get('cycles', []))}; requests: {suite.get('requests', 'not reported')}.\n\n")
@@ -118,6 +140,13 @@ jobs:
             .qa/daily/gate/suite/coverage.json
             .qa/daily/gate/suite/coverage.md
             .qa/daily/plugins/manifest.json
+            .qa/daily/restart-checks/profile.json
+            .qa/daily/restart-checks/cases.json
+            .qa/daily/restart-checks/summary.json
+            .qa/daily/restart-checks/coverage.json
+            .qa/daily/restart-checks/coverage.md
+            .qa/daily/restart-checks/cleanup-checks.json
+            .qa/daily/restart-checks/owned.json
             .qa/daily/*.log
           if-no-files-found: warn
           retention-days: 30