@fx-data/ai-diagnostics 0.1.20261026070913 → 0.1.20261107142011

package/content/knowledge/update/diagnostic/observability/lookups/analysis-update-child-release.json

@@ -1,146 +0,0 @@
-{
-  "lookup_id": "analysis-update-child-release",
-  "logical_event": "child-frontier-released",
-  "lifecycle_stage": "control",
-  "graph_kind": "execution",
-  "signal_status": "source-confirmed",
-  "signals": {
-    "span_names": [
-      "dispatch_callback_children"
-    ],
-    "attribute_keys": [],
-    "structured_fields": [],
-    "log_signatures": [
-      "dispatch callback children, nodeTaskId {}, executeId {}, chainId {}, children size {},"
-    ],
-    "metric_names": []
-  },
-  "service_scope": [
-    "main-server"
-  ],
-  "produced_when": "DispatchExecuteHandler wraps ordinary child callback in dispatch_callback_children; ChildrenTaskDispatcher emits the stable template after deleting parent details, evaluating children, and saving the new child frontier.",
-  "correlations": [
-    {
-      "edge_id": "edge-primary-result-to-child-frontier",
-      "from_unit": "node-primary-result",
-      "to_unit": "execution-detail",
-      "correlation_keys": [
-        "parent node table ID",
-        "child node table ID",
-        "chain or merged-chain context"
-      ],
-      "semantics": "On an ordinary path that reaches child callback, a primary node result already exists before current dependency checks create the next child frontier; post-result listener phases intervene, so this is a coarse prerequisite edge rather than direct adjacency.",
-      "asynchronous": false,
-      "correlation_status": "source-confirmed-ordered-coarse",
-      "durability": "frontier-lifetime",
-      "coverage_limit": "This edge does not bypass or time listener callbacks, does not apply when StopWork is rethrown before child callback, and the new detail is not a persisted copy of the logical edge."
-    },
-    {
-      "edge_id": "edge-listener-complete-to-child-release",
-      "from_unit": "listener-invocation",
-      "to_unit": "execution-detail",
-      "correlation_keys": [
-        "update chain or merged-chain context",
-        "node table ID",
-        "final onComplete listener invocation in the ordered same-call sequence"
-      ],
-      "semantics": "When nodeTaskExec returns normally after ordinary success or failure, the ordered onComplete invocation set has returned or isolated its exceptions before callbackChildren begins; when the set is non-empty, this edge is anchored by its final completed onComplete invocation rather than every invocation.",
-      "asynchronous": false,
-      "correlation_status": "source-confirmed-in-process-conditional",
-      "durability": "execution-scope",
-      "coverage_limit": "StopWork is rethrown after the finally callback and skips child callback; an empty listener set has no listener invocation, and earlier onComplete invocations are followed by later listeners, so the coarse primary-result-to-child edge remains necessary.",
-      "callback_phases": [
-        "onComplete"
-      ]
-    },
-    {
-      "edge_id": "edge-logical-parent-to-child",
-      "from_unit": "dag-node",
-      "to_unit": "dag-node",
-      "correlation_keys": [
-        "parent table ID",
-        "child table ID",
-        "chain or merged-chain scope"
-      ],
-      "semantics": "A current dependency-provider relation allows a parent result to gate and eventually release a child.",
-      "asynchronous": false,
-      "correlation_status": "source-confirmed-logical",
-      "durability": "not-snapshotted",
-      "coverage_limit": "Later current lineage cannot prove the earlier historical edge."
-    }
-  ],
-  "counting_unit": "execution-detail",
-  "input_units": [
-    "node-primary-result",
-    "logical-dag-edge",
-    "execution-detail"
-  ],
-  "output_units": [
-    "execution-detail"
-  ],
-  "state_owner": "Child-task dispatcher and execution-detail repository.",
-  "reads": [
-    "current direct children, parents, and ancestor map",
-    "unfinished ancestor details across current or mergeable chains",
-    "merge round and limit"
-  ],
-  "writes": [
-    "deletion of consumed parent frontier details",
-    "new READY or INIT child frontier details"
-  ],
-  "success_condition": "The parent frontier is replaced transactionally by the next eligible child frontier, or deleted when no child remains.",
-  "failure_recovery": [
-    {
-      "branch": "other ancestors remain unfinished",
-      "outcome": "The child is omitted or stored INIT according to merge policy.",
-      "recovery": "Another parent release or INIT inspection may later make it READY."
-    },
-    {
-      "branch": "child transaction fails",
-      "outcome": "The source-confirmed enabled test retains remaining parent detail at the observed failure point.",
-      "recovery": "No stronger eventual recovery is claimed from that test alone."
-    }
-  ],
-  "source_evidence_grade": "source-confirmed-with-enabled-test",
-  "edge_refs": [
-    "edge-primary-result-to-child-frontier",
-    "edge-listener-complete-to-child-release",
-    "edge-logical-parent-to-child"
-  ],
-  "field_presence_preflight": [
-    {
-      "field": "dispatch_callback_children",
-      "signal_type": "span-name",
-      "required": true,
-      "failure_effect": "needs_revalidation"
-    },
-    {
-      "field": "dispatch callback children, nodeTaskId {}, executeId {}, chainId {}, children size {},",
-      "signal_type": "log-signature",
-      "required": false,
-      "failure_effect": "needs_revalidation"
-    }
-  ],
-  "coverage_limits": [
-    "Child frontiers incrementally materialize execution order; they are not logical DAG edges or historical attempts.",
-    "The template reports only a child count and the parent payload identity; it does not expose child execution-detail IDs or publication acknowledgements.",
-    "No-child deletion and transaction or publication failures require separate interpretation."
-  ],
-  "invalidators": [
-    {
-      "kind": "revision-change",
-      "event": "The frozen main-server revision or build anchor changes.",
-      "effect": "needs_revalidation"
-    },
-    {
-      "kind": "signal-drift",
-      "event": "A named span, attribute, structured field, log template, metric, or its production condition changes.",
-      "effect": "needs_revalidation"
-    },
-    {
-      "kind": "scope-change",
-      "event": "The dispatch, update, queue, listener, or engine-handoff service boundary changes.",
-      "effect": "needs_revalidation"
-    }
-  ]
-}

package/content/knowledge/update/diagnostic/observability/lookups/analysis-update-dlx-recovery.json

@@ -1,100 +0,0 @@
-{
-  "lookup_id": "analysis-update-dlx-recovery",
-  "logical_event": "infrastructure-stopped-attempt-recovered",
-  "lifecycle_stage": "queue",
-  "graph_kind": "execution",
-  "signal_status": "source-confirmed",
-  "signals": {
-    "span_names": [],
-    "attribute_keys": [],
-    "structured_fields": [],
-    "log_signatures": [
-      "receive msg at dlx queue, msgId = {}",
-      "send execute info to fast, msgId = {}"
-    ],
-    "metric_names": []
-  },
-  "service_scope": [
-    "main-server"
-  ],
-  "produced_when": "RabbitDlxNodeTaskMQListener emits the receive template for each DLX delivery and the send-to-fast template after the execution checker succeeds and producer.sendNodeTask returns.",
-  "correlations": [
-    {
-      "edge_id": "edge-dlx-recovery-to-attempt",
-      "from_unit": "requeue",
-      "to_unit": "execution-attempt",
-      "correlation_keys": [
-        "execution detail ID"
-      ],
-      "semantics": "Dead-letter recovery republishes the same detail to the main queue after execution availability returns.",
-      "asynchronous": true,
-      "correlation_status": "source-confirmed-payload",
-      "durability": "message-lifetime",
-      "coverage_limit": "The recovery cycles and later attempt have no independent IDs."
-    }
-  ],
-  "counting_unit": "requeue",
-  "input_units": [
-    "execution-detail",
-    "dead-letter-delivery"
-  ],
-  "output_units": [
-    "queue-publication"
-  ],
-  "state_owner": "Dead-letter consumer and dispatch execution checker.",
-  "reads": [
-    "same execution detail payload",
-    "current server execution availability"
-  ],
-  "writes": [
-    "main-queue republication when execution becomes available"
-  ],
-  "success_condition": "Infrastructure-stopped work is returned to the main queue only after the execution checker succeeds.",
-  "failure_recovery": [
-    {
-      "branch": "execution remains unavailable",
-      "outcome": "The dead-letter consumer waits and checks again.",
-      "recovery": "No business attempt is counted until a later main-queue delivery."
-    }
-  ],
-  "source_evidence_grade": "source-confirmed",
-  "edge_refs": [
-    "edge-dlx-recovery-to-attempt"
-  ],
-  "field_presence_preflight": [
-    {
-      "field": "receive msg at dlx queue, msgId = {}",
-      "signal_type": "log-signature",
-      "required": true,
-      "failure_effect": "needs_revalidation"
-    },
-    {
-      "field": "send execute info to fast, msgId = {}",
-      "signal_type": "log-signature",
-      "required": true,
-      "failure_effect": "needs_revalidation"
-    }
-  ],
-  "coverage_limits": [
-    "This recovery is infrastructure availability handling, not proof of an application-level retry contract for ordinary failures.",
-    "msgId is the reused execution-detail ID, not a DLX attempt or recovery ID; repeated retries cannot be uniquely numbered.",
-    "The templates do not prove broker acceptance or the later main-consumer attempt."
-  ],
-  "invalidators": [
-    {
-      "kind": "revision-change",
-      "event": "The frozen main-server revision or build anchor changes.",
-      "effect": "needs_revalidation"
-    },
-    {
-      "kind": "signal-drift",
-      "event": "A named span, attribute, structured field, log template, metric, or its production condition changes.",
-      "effect": "needs_revalidation"
-    },
-    {
-      "kind": "scope-change",
-      "event": "The dispatch, update, queue, listener, or engine-handoff service boundary changes.",
-      "effect": "needs_revalidation"
-    }
-  ]
-}

package/content/knowledge/update/diagnostic/observability/lookups/analysis-update-effective-root-selection.json

@@ -1,111 +0,0 @@
-{
-  "lookup_id": "analysis-update-effective-root-selection",
-  "logical_event": "effective-root-selected",
-  "lifecycle_stage": "control",
-  "graph_kind": "logical",
-  "signal_status": "source-confirmed",
-  "signals": {
-    "span_names": [
-      "dispatch_gen_roots"
-    ],
-    "attribute_keys": [],
-    "structured_fields": [],
-    "log_signatures": [],
-    "metric_names": []
-  },
-  "service_scope": [
-    "main-server"
-  ],
-  "produced_when": "DispatchDependencyManager wraps each root-reduction call in the dispatch_gen_roots span.",
-  "correlations": [
-    {
-      "edge_id": "edge-candidate-to-effective-root",
-      "from_unit": "candidate-root",
-      "to_unit": "effective-root",
-      "correlation_keys": [
-        "table ID within the same dispatch invocation"
-      ],
-      "semantics": "Entry eligibility and ancestor reduction determine whether a candidate remains an effective root.",
-      "asynchronous": false,
-      "correlation_status": "source-confirmed-in-process",
-      "durability": "not-durable",
-      "coverage_limit": "Filtered candidates are not stored as a durable ledger."
-    },
-    {
-      "edge_id": "edge-effective-root-to-chain",
-      "from_unit": "effective-root",
-      "to_unit": "update-chain",
-      "correlation_keys": [
-        "update chain ID",
-        "table ID"
-      ],
-      "semantics": "The selected roots are accepted under the chain ID already present in the serialized context.",
-      "asynchronous": false,
-      "correlation_status": "source-confirmed-durable",
-      "durability": "frontier-lifetime",
-      "coverage_limit": "A returned chain can exist without any detail when no effective root is persisted."
-    }
-  ],
-  "counting_unit": "effective-root",
-  "input_units": [
-    "eligible-candidate-root"
-  ],
-  "output_units": [
-    "effective-root"
-  ],
-  "state_owner": "Dispatch dependency manager and submit handler.",
-  "reads": [
-    "candidate roots",
-    "current ancestor relations from lineage and linked-dependency providers"
-  ],
-  "writes": [
-    "in-memory effective-root set"
-  ],
-  "success_condition": "A candidate is an effective root when no ancestor from the same candidate set supersedes it.",
-  "failure_recovery": [
-    {
-      "branch": "dependency lookup fails",
-      "outcome": "Submission fails with dependency-unavailable behavior.",
-      "recovery": "No persisted frontier is claimed."
-    },
-    {
-      "branch": "non-empty candidates yield no roots",
-      "outcome": "The submit handler logs the inconsistency and falls back to all candidates as roots.",
-      "recovery": "Frontier persistence continues with the fallback roots."
-    }
-  ],
-  "source_evidence_grade": "source-confirmed",
-  "edge_refs": [
-    "edge-candidate-to-effective-root",
-    "edge-effective-root-to-chain"
-  ],
-  "field_presence_preflight": [
-    {
-      "field": "dispatch_gen_roots",
-      "signal_type": "span-name",
-      "required": true,
-      "failure_effect": "needs_revalidation"
-    }
-  ],
-  "coverage_limits": [
-    "Root selection proves neither a complete DAG snapshot nor eventual node execution.",
-    "The span confirms that root reduction ran but carries neither candidate IDs nor selected root IDs, so it cannot count effective roots."
-  ],
-  "invalidators": [
-    {
-      "kind": "revision-change",
-      "event": "The frozen main-server revision or build anchor changes.",
-      "effect": "needs_revalidation"
-    },
-    {
-      "kind": "signal-drift",
-      "event": "A named span, attribute, structured field, log template, metric, or its production condition changes.",
-      "effect": "needs_revalidation"
-    },
-    {
-      "kind": "scope-change",
-      "event": "The dispatch, update, queue, listener, or engine-handoff service boundary changes.",
-      "effect": "needs_revalidation"
-    }
-  ]
-}

package/content/knowledge/update/diagnostic/observability/lookups/analysis-update-engine-handoff.json

@@ -1,171 +0,0 @@
-{
-  "lookup_id": "analysis-update-engine-handoff",
-  "logical_event": "server-engine-task-handed-off",
-  "lifecycle_stage": "engine-handoff",
-  "graph_kind": "boundary",
-  "signal_status": "source-confirmed",
-  "signals": {
-    "span_names": [
-      "DataFlowExecutor",
-      "execute"
-    ],
-    "attribute_keys": [
-      "write_metric",
-      "metric",
-      "taskName",
-      "compute.retry.pipeline_parallelism_reduced"
-    ],
-    "structured_fields": [
-      "taskId"
-    ],
-    "log_signatures": [
-      "table {} load fail!",
-      "table {} load success."
-    ],
-    "metric_names": [
-      "XS03"
-    ]
-  },
-  "service_scope": [
-    "main-server"
-  ],
-  "produced_when": "BlockUpdateEngineHelper emits the terminal engine templates and engine-span attributes around write completion; UpdateMetricListener copies the final WriterMetricInfo taskId into the XS03 point on ordinary node completion.",
-  "correlations": [
-    {
-      "edge_id": "edge-load-operation-to-engine-task",
-      "from_unit": "load-operation",
-      "to_unit": "engine-task",
-      "correlation_keys": [
-        "server call scope",
-        "engine task ID returned in writer metric on success"
-      ],
-      "semantics": "A load operation creates an engine task only when it enters a Polars write executor.",
-      "asynchronous": false,
-      "correlation_status": "source-confirmed-conditional",
-      "durability": "result-dependent",
-      "coverage_limit": "Failed and retried engine attempts are not fully exposed as a server-side durable lineage."
-    },
-    {
-      "edge_id": "edge-engine-task-to-primary-result",
-      "from_unit": "engine-task",
-      "to_unit": "node-primary-result",
-      "correlation_keys": [
-        "engine task ID when returned",
-        "server execution scope"
-      ],
-      "semantics": "The final returned writer result or engine failure contributes to the server node primary result.",
-      "asynchronous": false,
-      "correlation_status": "source-confirmed-conditional",
-      "durability": "result-dependent",
-      "coverage_limit": "A missing returned task ID cannot be replaced by table or time proximity."
-    }
-  ],
-  "counting_unit": "engine-task",
-  "input_units": [
-    "load-operation"
-  ],
-  "output_units": [
-    "engine-task"
-  ],
-  "state_owner": "Server-side Polars write executor until the engine transaction accepts and completes the task.",
-  "reads": [
-    "write task, route decision, update context, source relation, and write configuration"
-  ],
-  "writes": [
-    "new engine task ID",
-    "engine ETL task and transaction commit or rollback",
-    "returned writer metric containing the successful task ID when available"
-  ],
-  "success_condition": "A Polars write task receives a task ID, is handed to an engine transaction, and returns a writer result after commit.",
-  "failure_recovery": [
-    {
-      "branch": "retryable engine execution failure",
-      "outcome": "The server executor may retry and generate another engine task ID for the same load operation.",
-      "recovery": "The successful result exposes only the final returned task ID."
-    },
-    {
-      "branch": "engine write fails",
-      "outcome": "The engine transaction rolls back and the server load fails.",
-      "recovery": "Further attribution requires a concrete engine task or request identity; otherwise stop at the server boundary."
-    }
-  ],
-  "source_evidence_grade": "source-confirmed",
-  "edge_refs": [
-    "edge-load-operation-to-engine-task",
-    "edge-engine-task-to-primary-result"
-  ],
-  "field_presence_preflight": [
-    {
-      "field": "write_metric",
-      "signal_type": "attribute",
-      "required": false,
-      "failure_effect": "needs_revalidation"
-    },
-    {
-      "field": "metric",
-      "signal_type": "attribute",
-      "required": false,
-      "failure_effect": "needs_revalidation"
-    },
-    {
-      "field": "taskId",
-      "signal_type": "structured-field",
-      "required": true,
-      "failure_effect": "needs_revalidation"
-    },
-    {
-      "field": "table {} load fail!",
-      "signal_type": "log-signature",
-      "required": false,
-      "failure_effect": "needs_revalidation"
-    },
-    {
-      "field": "table {} load success.",
-      "signal_type": "log-signature",
-      "required": false,
-      "failure_effect": "needs_revalidation"
-    },
-    {
-      "field": "XS03",
-      "signal_type": "metric",
-      "required": false,
-      "failure_effect": "needs_revalidation"
-    },
-    {
-      "field": "taskName",
-      "signal_type": "attribute",
-      "required": false,
-      "failure_effect": "needs_revalidation"
-    },
-    {
-      "field": "compute.retry.pipeline_parallelism_reduced",
-      "signal_type": "attribute",
-      "required": false,
-      "failure_effect": "needs_revalidation"
-    }
-  ],
-  "coverage_limits": [
-    "Engine internals beyond task construction, transaction write, commit or rollback are outside this main-server Phase 2 model.",
-    "No engine task exists for pass-through, inactive, needLoad=false, compute-model no-op, or other non-executed results.",
-    "taskId is empty on no-engine paths and only the final successful returned engine task is exposed to the server listener; failed or retried task lineage is incomplete.",
-    "XS03 construction is source-confirmed, but ingestion, indexing, and retention were not verified.",
-    "For a datasource-lake MPP write with an actual task MemoryLimitExceeded, DataFlowExecutor allows one retry and overrides pipeline parallelism to 1. The compute.retry.pipeline_parallelism_reduced attribute is conditional on this retry decision. This is not a general retry policy for every OOM, query or table type; failed and successful task IDs remain separate attempts."
-  ],
-  "invalidators": [
-    {
-      "kind": "revision-change",
-      "event": "The frozen main-server revision or build anchor changes.",
-      "effect": "needs_revalidation"
-    },
-    {
-      "kind": "signal-drift",
-      "event": "A named span, attribute, structured field, log template, metric, or its production condition changes.",
-      "effect": "needs_revalidation"
-    },
-    {
-      "kind": "scope-change",
-      "event": "The dispatch, update, queue, listener, or engine-handoff service boundary changes.",
-      "effect": "needs_revalidation"
-    }
-  ]
-}

package/content/knowledge/update/diagnostic/observability/lookups/analysis-update-execution-frontier-persisted.json

@@ -1,115 +0,0 @@
-{
-  "lookup_id": "analysis-update-execution-frontier-persisted",
-  "logical_event": "execution-frontier-persisted",
-  "lifecycle_stage": "control",
-  "graph_kind": "execution",
-  "signal_status": "missing",
-  "signals": {
-    "span_names": [],
-    "attribute_keys": [],
-    "structured_fields": [],
-    "log_signatures": [],
-    "metric_names": []
-  },
-  "service_scope": [
-    "main-server"
-  ],
-  "produced_when": "No dedicated signal is emitted immediately after saveBatch proves which root execution-detail rows were persisted.",
-  "correlations": [
-    {
-      "edge_id": "edge-effective-root-to-chain",
-      "from_unit": "effective-root",
-      "to_unit": "update-chain",
-      "correlation_keys": [
-        "update chain ID",
-        "table ID"
-      ],
-      "semantics": "The selected roots are accepted under the chain ID already present in the serialized context.",
-      "asynchronous": false,
-      "correlation_status": "source-confirmed-durable",
-      "durability": "frontier-lifetime",
-      "coverage_limit": "A returned chain can exist without any detail when no effective root is persisted."
-    },
-    {
-      "edge_id": "edge-chain-node-to-execution-detail",
-      "from_unit": "dag-node",
-      "to_unit": "execution-detail",
-      "correlation_keys": [
-        "update chain ID",
-        "node table ID",
-        "execution detail ID"
-      ],
-      "semantics": "Each effective root is materialized as one transient frontier detail for its chain context.",
-      "asynchronous": false,
-      "correlation_status": "source-confirmed-durable",
-      "durability": "frontier-lifetime",
-      "coverage_limit": "Details are later deleted and do not form a permanent node history."
-    },
-    {
-      "edge_id": "edge-execution-detail-to-publication",
-      "from_unit": "execution-detail",
-      "to_unit": "queue-publication",
-      "correlation_keys": [
-        "execution detail ID"
-      ],
-      "semantics": "A message containing the detail ID, node ID, and context is published after detail visibility.",
-      "asynchronous": true,
-      "correlation_status": "source-confirmed-payload",
-      "durability": "detail-and-message-lifetime",
-      "coverage_limit": "No separate publication ID is created."
-    }
-  ],
-  "counting_unit": "execution-detail",
-  "input_units": [
-    "effective-root",
-    "update-chain"
-  ],
-  "output_units": [
-    "execution-detail"
-  ],
-  "state_owner": "Dispatch execution-detail repository.",
-  "reads": [
-    "effective root IDs",
-    "serialized chain context"
-  ],
-  "writes": [
-    "frontier detail ID, node ID, serialized context, chain ID, merge mode, status, send flag, tenant, and submit time"
-  ],
-  "success_condition": "The complete batch of root frontier details is committed before a consumer must read it.",
-  "failure_recovery": [
-    {
-      "branch": "detail transaction fails",
-      "outcome": "No durable frontier is established.",
-      "recovery": "The caller may submit a separate later request; no queue publication is claimed."
-    }
-  ],
-  "source_evidence_grade": "source-confirmed-with-enabled-test",
-  "edge_refs": [
-    "edge-effective-root-to-chain",
-    "edge-chain-node-to-execution-detail",
-    "edge-execution-detail-to-publication"
-  ],
-  "field_presence_preflight": [],
-  "coverage_limits": [
-    "A detail is a transient frontier record and is deleted after progression; it is not an attempt history row.",
-    "dispatch_submit and its completion log span persistence, listener startup, and publication scheduling and also occur for an empty root list; they are not a row-persistence signal.",
-    "Repository state is a transient frontier rather than durable execution history."
-  ],
-  "invalidators": [
-    {
-      "kind": "revision-change",
-      "event": "The frozen main-server revision or build anchor changes.",
-      "effect": "needs_revalidation"
-    },
-    {
-      "kind": "signal-drift",
-      "event": "A named span, attribute, structured field, log template, metric, or its production condition changes.",
-      "effect": "needs_revalidation"
-    },
-    {
-      "kind": "scope-change",
-      "event": "The dispatch, update, queue, listener, or engine-handoff service boundary changes.",
-      "effect": "needs_revalidation"
-    }
-  ]
-}