{ "Schema": "syrf.project-statistics.phase2c-staging-proof/v1", "_comment": "Template for one FEAT-024 Phase 2C staging-proof run. Copy to phase2c-staging__.json and replace every TODO. Store raw response bodies, not summaries of them: a verdict without its watermarks and scopeStates cannot be re-interpreted later. Never record reviewer identities, study titles or other project content - ids, counts and response envelopes are enough. Runbook: ../../phase2c-staging-proof-runbook.md", "RunLabel": "TODO", "DateUtc": "TODO 2026-00-00", "Environment": "staging", "Operator": "TODO who ran it", "ProgrammeOwnerApproval": { "Granted": false, "By": "TODO", "AtUtc": "TODO", "Reference": "TODO link to the approval" }, "Gates": { "Syrf3196Merged": false, "Syrf3232Merged": false, "Syrf3185Closed": false, "PilotProjectIdRecorded": false, "Notes": "TODO anything that changed since the runbook was written" }, "Deployment": { "ClusterGitOpsPullRequest": "TODO camaradesuk/cluster-gitops#807", "ClusterGitOpsMergeSha": "TODO", "ApiImageDigest": "TODO from kubectl, not from chartTag", "ProjectManagementImageDigest": "TODO", "ObservedFlags": { "_comment": "Read from the running pods, not from the values file. Both hosts must agree.", "api": { "materializedProjectStatisticsWrites": "TODO", "materializedProjectStatisticsServing": "TODO", "materializedProjectStatisticsScreening": "TODO", "materializedProjectStatisticsMembershipScreening": "TODO", "materializedProjectStatisticsAnnotation": "TODO", "materializedProjectStatisticsMembershipAnnotation": "TODO", "materializedProjectStatisticsQuestionAnswers": "TODO", "materializedProjectStatisticsSearchPopulation": "TODO", "materializedProjectStatisticsDerivedSummaries": "TODO", "materializedProjectStatisticsPages": "TODO", "materializedProjectStatisticsSignalR": "TODO", "materializedProjectStatisticsExports": "TODO" }, "projectManagement": { "_comment": "Same twelve keys; record them, do not assume they match the API." } }, "ObservedAllowlist": { "api": "TODO value of SYRF__ProjectStatistics__ProjectAllowlist as rendered", "projectManagement": "TODO" }, "EffectiveSerilogMinimumLevel": "TODO confirm Debug lines actually appear in kubectl logs" }, "PilotProject": { "ProjectId": "TODO GUID", "SelectionRationale": "TODO real screening activity; small enough to backfill in minutes; not production-critical and not somebody's live work", "ConfirmedSafeToUseBy": "TODO", "StudyCount": null, "ReviewerCount": null, "ScreeningDecisionCount": null, "AgreementThresholdStable": null }, "LegacyBaseline": { "_comment": "Step 2. The authoritative screening statistics captured through the ordinary surface BEFORE anything was enabled. This is what 'prove legacy response equivalence' is measured against.", "CapturedAtUtc": "TODO", "Response": {} }, "Backfill": { "_comment": "Step 3. POST api/admin/project-statistics/{projectId}/backfill. Synchronous despite the 202.", "RequestedAtUtc": "TODO", "DurationSeconds": null, "ClientTimeoutSeconds": null, "HttpStatus": null, "ResponseBody": {}, "Interpretation": "TODO 202 with scopesStale 0 and checkpointStatus Succeeded, or the failureReason and what was done about it" }, "RebuildsPerformed": [ { "_comment": "Step 3/5. POST .../rebuild (forced). Record each one, including any run to clear a ControlVersionMismatch.", "Reason": "TODO", "RequestedAtUtc": "TODO", "HttpStatus": null, "ResponseBody": {} } ], "ParityAudits": [ { "_comment": "Step 4 and step 5. Record EVERY attempt, including inconclusive ones. Store the whole report: inParity, isInconclusive, sourceAdvancedDuringAudit, materializedAvailable, readSource, fallbackReason, capacityFailure, checkpoint, watermarks, scopeStates, metrics, configurationMatchesCurrentSettings.", "Label": "TODO e.g. post-backfill", "RequestedAtUtc": "TODO", "HttpStatus": null, "ResponseBody": {}, "Verdict": "TODO in parity / out of parity / inconclusive / unaudited", "Interpretation": "TODO. Remember: a bundle that refuses to serve is not in parity, it is unaudited - read fallbackReason. configurationMatchesCurrentSettings null never means it passed." } ], "ScreeningDecisionsExercised": [ { "_comment": "Step 5. A real decision through the ordinary UI, not a database write.", "AtUtc": "TODO", "StudyId": "TODO", "ReviewerId": "TODO", "Decision": "TODO include | exclude", "CommittedProjectionRevisionBefore": null, "CommittedProjectionRevisionAfter": null, "ExpectedCounterMovement": "TODO which metric keys should move and by how much", "ObservedCounterMovement": "TODO", "ParityAuditLabelAfter": "TODO cross-reference into ParityAudits" } ], "PathsExercised": { "_comment": "The Phase 2 staging proof names these explicitly: force Stale/Rebuilding/Incompatible and the kill-switch paths.", "KillSwitchOffAndBack": { "Exercised": false, "Observation": "TODO reads fell back; parity returned 409 writes-disabled; projection intact and in parity after re-enable + rebuild" }, "StaleObserved": { "Exercised": false, "Observation": "TODO" }, "RebuildingObserved": { "Exercised": false, "Observation": "TODO" }, "IncompatibleObserved": { "Exercised": false, "Observation": "TODO" }, "ForcedRebuild": { "Exercised": false, "Observation": "TODO new publicationGeneration, still in parity" } }, "FallbackReasonsObserved": { "_comment": "Record the SET, not only the counts. A reason nobody expected is a finding whether or not parity held. Adapter reasons (ProjectScreeningStatisticsFallbackReason) and reader reasons (ProjectStatisticsFallbackReason) appear in one log line separated by '/'.", "Source": "TODO pod stdout substring query, and/or the parity report's fallbackReason field", "Observed": [], "UnexpectedReasons": [], "ParityDivergenceWarnings": { "_comment": "The Warning line 'FEAT-024 declined to substitute the materialized screening section'. Any occurrence is a finding.", "Count": 0, "Details": [] } }, "Soak": { "_comment": "Phase 6 acceptance: at least seven ACTUAL days, 10,000 reads, 1,000 relevant mutations, 100% exact counter parity, zero stale materialized serves, 100% injected fallback success, no unresolved rebuild failures, bounded delta-ledger and storage growth. There is no read or mutation counter in the code; see the runbook.", "StartedAtUtc": "TODO", "EndedAtUtc": "TODO", "ElapsedDays": null, "ReadCount": null, "ReadCountMethod": "TODO log line count, or a driver counting its own calls. State WHICH reads are being counted: with every consumer flag off, ordinary page reads never touch the projection.", "MutationCount": null, "MutationCountMethod": "TODO difference between two observations of watermarks.committedProjectionRevision", "ParitySamples": [], "StaleMaterializedServes": null, "UnresolvedRebuildFailures": null, "DeltaLedgerGrowth": "TODO", "StorageGrowth": "TODO" }, "Rollback": { "_comment": "Step 8. Flags false on both hosts, allowlist back to the placeholder, through cluster-gitops only.", "Performed": false, "AtUtc": "TODO", "ClusterGitOpsRevertRef": "TODO", "VerifiedReadsAuthoritative": false, "StateLeftBehind": "pmProjectStatistics* documents, control row, delta ledger and the backfill-observed checkpoint remain and are never read. Re-enabling requires a fresh backfill or forced rebuild: the rows are intact but no longer current, and the reader correctly refuses them as Stale." }, "Outcome": { "ProofPassed": null, "Findings": [], "FollowUpIssues": [], "Notes": "TODO" } }