schema_version: 1 case_id: single-node-cpu-core-scope-compare-in-premium-cluster-during-incident-window ground_truth: root_cause: >- Correctly answering this ad-hoc case requires isolating the exact TiKV pod `the runtime-selected target TiKV node`, deriving its CPU spec cores from TiKV quota evidence, and returning the average and peak CPU cores for the three fixed 30-minute windows without mixing in `the runtime-selected target TiKV node0/91/...` or cluster-wide aggregates. judge: semantic_strategy: undecided contract: checks: - id: cluster_scope_match required: true description: Actual diagnosis cluster scope matches the expected cluster id. match: all_of rules: - source: trace_fact matcher: bool_true path: cluster_scope_match workflow: checks: - id: final_answer_emitted weight: 0 required: true description: Final answer is emitted for the ad-hoc query. match: all_of rules: - source: workflow_fact matcher: bool_true path: final_answer_emitted - id: collect_evidence_not_used weight: 0 required: true description: This ad-hoc case must not invoke collect_evidence.py. match: all_of rules: - source: workflow_fact matcher: bool_false path: collect_evidence_invoked execution_contract: checks: - id: required_workflow_present required: true description: The minimal ad-hoc workflow is present. match: all_of rules: - source: workflow_check matcher: subset_observed paths: - final_answer_emitted - id: collect_evidence_not_used required: true description: This ad-hoc case must not invoke collect_evidence.py. match: all_of rules: - source: workflow_check matcher: subset_observed paths: - collect_evidence_not_used - id: required_report_present required: true description: Workspace diagnosis report is present. match: all_of rules: - source: artifact_fact matcher: subset_observed paths: - diagnosis_report_present - id: query_or_metric_channel_observed required: true description: At least one query or metric evidence channel is observed. match: all_of rules: - source: channel_fact matcher: count_gte paths: - query_or_metric threshold: 1 evidence_collection: weight: 6 checks: - id: required_report_present weight: 2 required: true description: Workspace diagnosis report is present and readable. match: all_of rules: - source: artifact_fact matcher: subset_observed paths: - diagnosis_report_present - id: query_or_metric_channel_observed weight: 2 required: true description: Query or metric evidence is observed for this ad-hoc request. match: all_of rules: - source: channel_fact matcher: count_gte paths: - query_or_metric threshold: 1 critical_evidence: weight: 79 checks: - id: cpu_metric_evidence_present weight: 5 required: true description: Evidence includes TiKV CPU core and quota metrics for this request. match: all_of rules: - source: artifact_json artifacts: - metrics - workspace_report matcher: json_text_contains_any keywords: - process_cpu_seconds_total - source: artifact_json artifacts: - metrics - workspace_report matcher: json_text_contains_any keywords: - tikv_server_cpu_cores_quota - id: cpu_1m_rate_window_evidence_present weight: 4 required: true description: Evidence shows the CPU query used the 1-minute rate window rather than a wider window. match: all_of rules: - source: artifact_json artifacts: - workspace_report matcher: json_text_contains_any keywords: - rate(process_cpu_seconds_total - source: artifact_json artifacts: - workspace_report matcher: json_text_contains_any keywords: - '[1m]' - id: output_cluster_id_matches_reference weight: 5 required: true description: Output JSON cluster_id matches the runtime reference. match: all_of rules: - source: output_json matcher: json_path_equals_reference path: cluster_id reference_output_id: reference_answer reference_path: cluster_id - id: output_node_matches_reference weight: 8 required: true description: Output JSON node matches the exact runtime reference node. match: all_of rules: - source: output_json matcher: json_path_equals_reference path: node reference_output_id: reference_answer reference_path: node - id: output_cpu_spec_matches_reference weight: 12 required: true description: Output JSON cpu_spec_cores matches the runtime reference. match: all_of rules: - source: output_json matcher: json_path_equals_reference path: cpu_spec_cores reference_output_id: reference_answer reference_path: cpu_spec_cores - id: evidence_sufficient_matches_reference weight: 4 required: true description: Output JSON evidence_sufficient matches the runtime reference. match: all_of rules: - source: output_json matcher: json_path_equals_reference path: evidence_sufficient reference_output_id: reference_answer reference_path: evidence_sufficient - id: windows_identity_matches_reference weight: 20 required: true description: Output JSON windows ordering and identity fields match the runtime reference. match: all_of rules: - source: output_json matcher: json_array_matches_reference path: windows reference_output_id: reference_answer reference_path: windows key_fields: - start - end - id: windows_numeric_fields_match_reference weight: 25 required: true description: Output JSON windows numeric fields match the runtime reference. match: all_of rules: - source: output_json matcher: json_array_matches_reference path: windows reference_output_id: reference_answer reference_path: windows key_fields: - start - end numeric_fields: - avg_cpu_cores - peak_cpu_cores threshold: 0.0001 evidence_routing: weight: 0 checks: - id: analysis_patch_valid_when_present weight: 0 required: true description: Analysis patch is only required to be valid when it exists. match: any_of rules: - source: routing_fact matcher: bool_false path: analysis_patch_present - source: routing_fact matcher: bool_true path: analysis_patch_valid_json critical: - id: output_node_matches_reference description: Output JSON node matches the exact runtime reference node. - id: output_cpu_spec_matches_reference description: Output JSON cpu_spec_cores matches the runtime reference. - id: output_windows_match_reference description: Output JSON windows match the runtime reference values and ordering. - id: cpu_1m_rate_window_evidence_present description: CPU evidence uses the 1-minute rate window, not a wider one.