:hedgehog: PostHog is the leading platform for building self-driving products. Our developer tools – AI observability, analytics, session replay, flags, experiments, error tracking, logs, and more – capture all the context agents need to diagnose problems, uncover opportunities, and ship fixes. Steer it all from Slack, web, desktop, or the MCP.
Python ab-testingai-analyticsanalyticscdpdata-warehouseexperimentsfeature-flagsjavascriptproduct-analyticspythonreactsession-replaysurveystypescriptweb-analytics
feat(warehouse): add depot job metrics, artifacts and failure diagnosis tables (#119245)
Tom Owersf38aa935a9e2
diff --git a/products/warehouse_sources/backend/temporal/data_imports/sources/COVERAGE_GAPS_APPENDIX.md b/products/warehouse_sources/backend/temporal/data_imports/sources/COVERAGE_GAPS_APPENDIX.mdindex 61b96721af18..900437b9bef8 100644--- a/products/warehouse_sources/backend/temporal/data_imports/sources/COVERAGE_GAPS_APPENDIX.md+++ b/products/warehouse_sources/backend/temporal/data_imports/sources/COVERAGE_GAPS_APPENDIX.md@@ -3683,14 +3683,14 @@ Note: The v2 Deploy API is tiny — 15 GET operations. Everything queryable as a## Depot — gaps-Today (1): `job_attempts`+Today (4): `artifacts`, `failure_diagnoses`, `job_attempts`, `job_metrics`Diffed against: <https://depot.dev/openapi/depot-ci/v1.json>-- [ ] `Job metrics (POST /depot.ci.v1.CIService/GetJobMetrics)` — measure CPU and memory use across attempts to identify inefficient jobs and improve runner sizing (high)+- [x] `Job metrics (POST /depot.ci.v1.CIService/GetJobMetrics)` — measure CPU and memory use across attempts to identify inefficient jobs and improve runner sizing (high)- [ ] `Job attempt logs (POST /depot.ci.v1.CIService/GetJobAttemptLogs)` — analyze failure messages and workflow step activity across imported attempts (medium)-- [ ] `Failure diagnosis (POST /depot.ci.v1.CIService/GetFailureDiagnosis)` — group recurring failures by fingerprint and compare causes across runs, workflows, and attempts (medium)-- [ ] `Artifacts (POST /depot.ci.v1.CIService/ListArtifacts)` — track artifact counts, sizes, and production times by workflow and job (medium)+- [x] `Failure diagnosis (POST /depot.ci.v1.CIService/GetFailureDiagnosis)` — group recurring failures by fingerprint and compare causes across runs, workflows, and attempts (medium)+- [x] `Artifacts (POST /depot.ci.v1.CIService/ListArtifacts)` — track artifact counts, sizes, and production times by workflow and job (medium)Note: Parsed the official CI OpenAPI document and read the operation descriptions. Depot uses POST for read RPCs; job_attempts already includes run, workflow, job, and attempt metadata. Missing metrics, logs, diagnoses, and artifact collections need parent-ID fan-out; GetJobAttemptMetrics offers an alternative for individual attempts. Secrets, variables, downloads, and execution commands were excluded; the audit covers Depot CI.diff --git a/products/warehouse_sources/backend/temporal/data_imports/sources/depot/canonical_descriptions.py b/products/warehouse_sources/backend/temporal/data_imports/sources/depot/canonical_descriptions.pyindex ec5b79b8ec93..9c5f4d20f5fb 100644--- a/products/warehouse_sources/backend/temporal/data_imports/sources/depot/canonical_descriptions.py+++ b/products/warehouse_sources/backend/temporal/data_imports/sources/depot/canonical_descriptions.py@@ -42,4 +42,80 @@ CANONICAL_DESCRIPTIONS: CanonicalDescriptions = {"sandbox_id": "Identifier of the sandbox that ran the attempt.",},},+ "job_metrics": {+ "description": (+ "One row per attempt of a Depot CI job, with CPU and memory statistics from the sandbox that ran it. "+ "Jobs that never ran have no rows."+ ),+ "docs_url": "https://depot.dev/docs/api/ci/reference",+ "columns": {+ "run_id": "Unique identifier of the Depot CI run.",+ "run_created_at": "Time the run was created, as an RFC 3339 timestamp.",+ "workflow_id": "Unique identifier of the workflow within the run.",+ "workflow_name": "Name of the workflow.",+ "workflow_path": "Path of the workflow file in the repository.",+ "job_id": "Unique identifier of the job.",+ "job_key": "Key of the job from the workflow.",+ "job_status": "Status of the job.",+ "job_conclusion": "Conclusion of the job: success, failure, cancelled, skipped or timed_out.",+ "snapshot_at": "Time Depot computed the metrics, as an RFC 3339 timestamp.",+ "attempt_id": "Unique identifier of the job attempt.",+ "attempt": "Attempt number of the job.",+ "attempt_status": "Status of the attempt.",+ "attempt_conclusion": "Conclusion of the attempt: success, failure, cancelled, skipped or timed_out.",+ "attempt_started_at": "Time the attempt started, as an RFC 3339 timestamp.",+ "attempt_finished_at": "Time the attempt finished, as an RFC 3339 timestamp.",+ "sandbox_id": "Identifier of the sandbox that ran the attempt.",+ "availability": "Why metrics are present or not, for example available, no_sandbox, no_time_range or no_samples.",+ "sample_count": "Number of CPU and memory samples Depot observed for the attempt.",+ "peak_cpu_utilization": "Peak CPU utilization, as a fraction from 0 to 1.",+ "average_cpu_utilization": "Average CPU utilization, as a fraction from 0 to 1.",+ "peak_memory_utilization": "Peak memory utilization, as a fraction from 0 to 1.",+ "average_memory_utilization": "Average memory utilization, as a fraction from 0 to 1.",+ "peak_memory_usage_bytes": "Peak memory usage in bytes.",+ "average_memory_usage_bytes": "Average memory usage in bytes.",+ "observed_started_at": "Time of the first sample, as an RFC 3339 timestamp.",+ "observed_finished_at": "Time of the last sample, as an RFC 3339 timestamp.",+ },+ },+ "artifacts": {+ "description": "One row per artifact that a Depot CI job attempt uploaded. Download URLs are not synced.",+ "docs_url": "https://depot.dev/docs/api/ci/reference",+ "columns": {+ "artifact_id": "Unique identifier of the artifact.",+ "run_id": "Unique identifier of the Depot CI run.",+ "run_created_at": "Time the run was created, as an RFC 3339 timestamp.",+ "workflow_id": "Unique identifier of the workflow that produced the artifact.",+ "workflow_path": "Path of the workflow file in the repository.",+ "job_id": "Unique identifier of the job that produced the artifact.",+ "job_key": "Key of the job from the workflow.",+ "attempt_id": "Unique identifier of the job attempt that produced the artifact.",+ "attempt": "Attempt number that produced the artifact.",+ "name": "Artifact name from the workflow.",+ "size_bytes": "Size of the artifact in bytes.",+ "created_at": "Time the artifact was created, as an RFC 3339 timestamp.",+ },+ },+ "failure_diagnoses": {+ "description": (+ "One row per failure group in a failed or cancelled Depot CI run. Depot groups failures whose errors "+ "match after it removes variable details such as timestamps and IDs. A run with too many failed jobs "+ "to diagnose one by one has no rows."+ ),+ "docs_url": "https://depot.dev/docs/api/ci/reference",+ "columns": {+ "run_id": "Unique identifier of the Depot CI run.",+ "run_created_at": "Time the run was created, as an RFC 3339 timestamp.",+ "run_status": "Terminal status of the run: failed or cancelled.",+ "fingerprint": "Identifier of the failure group, computed from the normalized error text.",+ "source": "Origin of the error text: attempt_error, job_error, workflow_error or status.",+ "count": "Number of attempts in the failure group.",+ "error_message": "Representative error message of the group, possibly truncated.",+ "error_message_truncated": "True when Depot truncated the error message.",+ "diagnosis": "AI-generated summary of the failure. Empty when not available.",+ "possible_fix": "AI-generated suggested fix. Empty when not available.",+ "representatives": "Sample attempts of the failure, with their workflow, job, error, and relevant log lines.",+ "omitted_representative_count": "Number of attempts in the group that are not in representatives.",+ },+ },}diff --git a/products/warehouse_sources/backend/temporal/data_imports/sources/depot/depot.py b/products/warehouse_sources/backend/temporal/data_imports/sources/depot/depot.pyindex f5a9576d524e..623eb895fc46 100644--- a/products/warehouse_sources/backend/temporal/data_imports/sources/depot/depot.py+++ b/products/warehouse_sources/backend/temporal/data_imports/sources/depot/depot.py@@ -12,8 +12,11 @@ from products.warehouse_sources.backend.temporal.data_imports.sources.common.htt)from products.warehouse_sources.backend.temporal.data_imports.sources.common.typings import SourceResponsefrom products.warehouse_sources.backend.temporal.data_imports.sources.depot.settings import (+ ARTIFACTS,+ FAILURE_DIAGNOSES,JOB_ATTEMPTS,- PRIMARY_KEY,+ JOB_METRICS,+ PRIMARY_KEYS,RUN_CREATED_AT,)@@ -26,6 +29,9 @@ REQUEST_TIMEOUT_SECONDS = 60# until Depot fixes its cursor. Depot caps pages at 100.LIST_RUNS_PAGE_SIZES = (100, 57)TERMINAL_STATUSES = ["finished", "failed", "cancelled"]+# A finished run has no failed or cancelled job left to diagnose.+DIAGNOSED_RUN_STATUSES = frozenset({"failed", "cancelled"})+LIST_ARTIFACTS_PAGE_SIZE = 500# Connect sends every RPC as a POST. Every RPC this source calls is a read, so a retry is as safe as# a retried GET._DEPOT_RETRY = DEFAULT_RETRY.new(allowed_methods=frozenset(DEFAULT_RETRY.allowed_methods or ()) | {"POST"})@@ -148,13 +154,145 @@ def _run_attempt_rows(session: Session, run: JSONObject) -> list[JSONObject]:]+def _int_or_none(value: int | str | None) -> int | None:+ # Connect encodes int64 fields as JSON strings.+ return int(value) if value is not None and value != "" else None+++def _run_job_metrics_batches(session: Session, run: JSONObject) -> Iterator[list[JSONObject]]:+ # GetRunMetrics fails with ResourceExhausted for a run with many attempts, so metrics are read per job.+ workflows = _call(session, "GetRunStatus", {"runId": run["runId"]}).get("workflows", [])+ for workflow in workflows:+ for job in workflow.get("jobs", []):+ if not job.get("attempts"):+ continue+ metrics = _call(session, "GetJobMetrics", {"jobId": job["jobId"]})+ workflow_context = metrics.get("workflow") or {}+ job_context = metrics.get("job") or {}+ shared_columns = {+ "run_id": run["runId"],+ "run_created_at": run["createdAt"],+ "workflow_id": workflow["workflowId"],+ "workflow_name": workflow_context.get("name"),+ "workflow_path": workflow_context.get("workflowPath"),+ "job_id": job["jobId"],+ "job_key": job_context.get("jobKey"),+ "job_status": job_context.get("status"),+ "job_conclusion": job_context.get("conclusion"),+ "snapshot_at": metrics.get("snapshotAt"),+ }+ rows: list[JSONObject] = []+ for summary in metrics.get("attempts", []):+ attempt = summary.get("attempt") or {}+ availability = summary.get("availability") or {}+ stats = summary.get("stats") or {}+ rows.append(+ {+ **shared_columns,+ "attempt_id": attempt["attemptId"],+ "attempt": attempt.get("attempt"),+ "attempt_status": attempt.get("status"),+ "attempt_conclusion": attempt.get("conclusion"),+ "attempt_started_at": attempt.get("startedAt"),+ "attempt_finished_at": attempt.get("finishedAt"),+ "sandbox_id": attempt.get("sandboxId"),+ "availability": availability.get("reason"),+ "sample_count": stats.get("sampleCount"),+ "peak_cpu_utilization": stats.get("peakCpuUtilization"),+ "average_cpu_utilization": stats.get("averageCpuUtilization"),+ "peak_memory_utilization": stats.get("peakMemoryUtilization"),+ "average_memory_utilization": stats.get("averageMemoryUtilization"),+ "peak_memory_usage_bytes": _int_or_none(stats.get("peakMemoryUsageBytes")),+ "average_memory_usage_bytes": stats.get("averageMemoryUsageBytes"),+ "observed_started_at": stats.get("observedStartedAt"),+ "observed_finished_at": stats.get("observedFinishedAt"),+ }+ )+ if rows:+ yield rows+++def _run_artifact_batches(session: Session, run: JSONObject) -> Iterator[list[JSONObject]]:+ body: JSONObject = {"runId": run["runId"], "pageSize": LIST_ARTIFACTS_PAGE_SIZE}+ while True:+ page = _call(session, "ListArtifacts", body)+ rows = [+ {+ "artifact_id": artifact["artifactId"],+ "run_id": run["runId"],+ "run_created_at": run["createdAt"],+ "workflow_id": artifact.get("workflowId"),+ "workflow_path": artifact.get("workflowPath"),+ "job_id": artifact.get("jobId"),+ "job_key": artifact.get("jobKey"),+ "attempt_id": artifact.get("attemptId"),+ "attempt": artifact.get("attempt"),+ "name": artifact.get("name"),+ "size_bytes": _int_or_none(artifact.get("sizeBytes")),+ "created_at": artifact.get("createdAt"),+ }+ for artifact in page.get("artifacts", [])+ ]+ if rows:+ yield rows+ next_page_token = page.get("nextPageToken")+ if not next_page_token:+ return+ body = {**body, "pageToken": next_page_token}+++def _run_failure_diagnosis_batches(session: Session, run: JSONObject) -> Iterator[list[JSONObject]]:+ if run.get("status") not in DIAGNOSED_RUN_STATUSES:+ return+ diagnosis = _call(+ session,+ "GetFailureDiagnosis",+ {"targetId": run["runId"], "targetType": "FAILURE_DIAGNOSIS_TARGET_TYPE_RUN"},+ )+ rows = [+ {+ "run_id": run["runId"],+ "run_created_at": run["createdAt"],+ "run_status": run.get("status"),+ "fingerprint": group["fingerprint"],+ "source": group.get("source"),+ "count": group.get("count"),+ "error_message": group.get("errorMessage"),+ "error_message_truncated": group.get("errorMessageTruncated"),+ "diagnosis": group.get("diagnosis"),+ "possible_fix": group.get("possibleFix"),+ "representatives": group.get("representatives", []),+ "omitted_representative_count": group.get("omittedRepresentativeCount"),+ }+ for group in diagnosis.get("failureGroups", [])+ ]+ if rows:+ yield rows+++def _run_job_attempt_batches(session: Session, run: JSONObject) -> Iterator[list[JSONObject]]:+ rows = _run_attempt_rows(session, run)+ if rows:+ yield rows+++_RUN_BATCHES: dict[str, Callable[[Session, JSONObject], Iterator[list[JSONObject]]]] = {+ JOB_ATTEMPTS: _run_job_attempt_batches,+ JOB_METRICS: _run_job_metrics_batches,+ ARTIFACTS: _run_artifact_batches,+ FAILURE_DIAGNOSES: _run_failure_diagnosis_batches,+}++def depot_source(api_token: str,repository: str,created_after: dt.datetime | str | None,logger: FilteringBoundLogger,- on_complete: Callable[[], None] | None = None,+ on_complete: Callable[[str | None], None] | None = None,+ endpoint: str = JOB_ATTEMPTS,) -> SourceResponse:+ run_batches = _RUN_BATCHES[endpoint]lower_bound = _parse_timestamp(created_after) if created_after is not None else Nonedef items() -> Iterator[list[JSONObject]]:@@ -163,20 +301,19 @@ def depot_source(runs = [run for _, run in reversed(_list_runs(session, repository, TERMINAL_STATUSES, lower_bound))]logger.info("depot_ci.runs_to_sync",+ endpoint=endpoint,run_count=len(runs),created_after=lower_bound.isoformat() if lower_bound else None,)for run in runs:- rows = _run_attempt_rows(session, run)- if rows:- yield rows+ yield from run_batches(session, run)if on_complete is not None:- on_complete()+ on_complete(runs[0]["createdAt"] if runs else None)return SourceResponse(- name=JOB_ATTEMPTS,+ name=endpoint,items=items,- primary_keys=[PRIMARY_KEY],+ primary_keys=PRIMARY_KEYS[endpoint],partition_mode="datetime",partition_format="week",partition_keys=[RUN_CREATED_AT],diff --git a/products/warehouse_sources/backend/temporal/data_imports/sources/depot/settings.py b/products/warehouse_sources/backend/temporal/data_imports/sources/depot/settings.pyindex c40fcba8f953..81a47727a6f7 100644--- a/products/warehouse_sources/backend/temporal/data_imports/sources/depot/settings.py+++ b/products/warehouse_sources/backend/temporal/data_imports/sources/depot/settings.py@@ -1,21 +1,33 @@from products.warehouse_sources.backend.types import IncrementalField, IncrementalFieldTypeJOB_ATTEMPTS = "job_attempts"+JOB_METRICS = "job_metrics"+ARTIFACTS = "artifacts"+FAILURE_DIAGNOSES = "failure_diagnoses"-ENDPOINTS = (JOB_ATTEMPTS,)+ENDPOINTS = (JOB_ATTEMPTS, JOB_METRICS, ARTIFACTS, FAILURE_DIAGNOSES)-PRIMARY_KEY = "attempt_id"+PRIMARY_KEYS: dict[str, list[str]] = {+ JOB_ATTEMPTS: ["attempt_id"],+ JOB_METRICS: ["attempt_id"],+ ARTIFACTS: ["artifact_id"],+ # Depot computes the fingerprint per run, so the same failure in two runs has two rows.+ FAILURE_DIAGNOSES: ["run_id", "fingerprint"],+}# Creation time is stable for partitioning. The source replays runs to pick up late retries.RUN_CREATED_AT = "run_created_at"+# Failure diagnoses sync as a full refresh: a retry that passes removes a run's failure groups, and a+# merge never deletes the rows it no longer receives.INCREMENTAL_FIELDS: dict[str, list[IncrementalField]] = {- JOB_ATTEMPTS: [+ endpoint: [{"label": RUN_CREATED_AT,"type": IncrementalFieldType.DateTime,"field": RUN_CREATED_AT,"field_type": IncrementalFieldType.DateTime,},- ],+ ]+ for endpoint in (JOB_ATTEMPTS, JOB_METRICS, ARTIFACTS)}diff --git a/products/warehouse_sources/backend/temporal/data_imports/sources/depot/source.py b/products/warehouse_sources/backend/temporal/data_imports/sources/depot/source.pyindex bfa9dc8fbae5..d76d749ec674 100644--- a/products/warehouse_sources/backend/temporal/data_imports/sources/depot/source.py+++ b/products/warehouse_sources/backend/temporal/data_imports/sources/depot/source.py@@ -49,6 +49,8 @@ _APPEND_UNSUPPORTED = (class DepotReconciliationCursor:cursor_kind: ClassVar[str] = "depot_reconciliation"reconciled_at: str+ # The newest run a completed sync scanned. It stands in for the watermark of a table whose runs yield no rows.+ scanned_through: str | None = None@SourceRegistry.register@@ -72,7 +74,7 @@ class DepotSource(SimpleSource[DepotSourceConfig], CursorSource[DepotReconciliatcategory=DataWarehouseSourceCategory.ENGINEERING___MONITORING,label="Depot",keywords=["depot.dev", "depot ci"],- caption="Sync Depot CI job attempts for one GitHub repository. Create an organization API token in your Depot organization settings, then enter the repository as `owner/name`.",+ caption="Sync Depot CI job attempts, job metrics, artifacts, and failure diagnoses for one GitHub repository. Create an organization API token in your Depot organization settings, then enter the repository as `owner/name`.",releaseStatus=ReleaseStatus.ALPHA,docsUrl="https://posthog.com/docs/cdp/sources/depot",iconPath="/static/services/depot.png",@@ -146,7 +148,9 @@ class DepotSource(SimpleSource[DepotSourceConfig], CursorSource[DepotReconciliatnow = datetime.datetime.now(datetime.UTC)reconciled_at = parse_datetime_value(cursor.reconciled_at) if cursor is not None else Nonewatermark = (- parse_datetime_value(inputs.db_incremental_field_last_value)+ parse_datetime_value(+ inputs.db_incremental_field_last_value or (cursor.scanned_through if cursor is not None else None)+ )if inputs.should_use_incremental_fieldelse None)@@ -159,12 +163,21 @@ class DepotSource(SimpleSource[DepotSourceConfig], CursorSource[DepotReconciliatif inputs.history_start is not None:lower_bound = max(lower_bound, inputs.history_start)+ previous = cursor or DepotReconciliationCursor(reconciled_at=now.isoformat())++ def on_complete(scanned_through: str | None) -> None:+ manager.stage(+ DepotReconciliationCursor(+ reconciled_at=now.isoformat() if reconcile else previous.reconciled_at,+ scanned_through=scanned_through or previous.scanned_through,+ )+ )+return depot_source(api_token=config.api_token,repository=config.repository,created_after=lower_bound,logger=inputs.logger,- on_complete=(lambda: manager.stage(DepotReconciliationCursor(reconciled_at=now.isoformat())))- if reconcile- else None,+ endpoint=inputs.schema_name,+ on_complete=on_complete,)diff --git a/products/warehouse_sources/backend/temporal/data_imports/sources/depot/tests/test_depot.py b/products/warehouse_sources/backend/temporal/data_imports/sources/depot/tests/test_depot.pyindex 80acde19233a..727c916ddc56 100644--- a/products/warehouse_sources/backend/temporal/data_imports/sources/depot/tests/test_depot.py+++ b/products/warehouse_sources/backend/temporal/data_imports/sources/depot/tests/test_depot.py@@ -21,6 +21,12 @@ from products.warehouse_sources.backend.temporal.data_imports.sources.depot.depodepot_source,validate_credentials,)+from products.warehouse_sources.backend.temporal.data_imports.sources.depot.settings import (+ ARTIFACTS,+ FAILURE_DIAGNOSES,+ JOB_ATTEMPTS,+ JOB_METRICS,+)from products.warehouse_sources.backend.temporal.data_imports.sources.depot.source import (DepotReconciliationCursor,DepotSource,@@ -83,8 +89,31 @@ def _fake_session(next_page_token = str(end) if end < len(terminal_runs) else ""return _response(200, {"runs": page, "nextPageToken": next_page_token}, method)if method == "GetRunStatus":- run_workflows = workflows.get(json["runId"], [])- return _response(200, {"workflows": [{"workflowId": w["workflowId"]} for w in run_workflows]}, method)+ return _response(200, {"workflows": workflows.get(json["runId"], [])}, method)+ if method == "GetJobMetrics":+ attempts = next(+ job["attempts"]+ for workflow in workflows_by_id.values()+ for job in workflow["jobs"]+ if job["jobId"] == json["jobId"]+ )+ summaries = [{"attempt": attempt, "stats": {"peakMemoryUsageBytes": "1024"}} for attempt in attempts]+ return _response(200, {"job": {"jobKey": "build"}, "attempts": summaries}, method)+ if method == "ListArtifacts":+ artifacts_page: dict[str, Any]+ if "pageToken" not in json:+ artifacts_page = {+ "artifacts": [{"artifactId": f"{json['runId']}-a1", "sizeBytes": "2048"}],+ "nextPageToken": "p2",+ }+ else:+ artifacts_page = {"artifacts": [{"artifactId": f"{json['runId']}-a2", "sizeBytes": "4096"}]}+ return _response(200, artifacts_page, method)+ if method == "GetFailureDiagnosis":+ groups = [{"fingerprint": "fp-1", "count": 2, "errorMessage": "exit status 1"}]+ return _response(+ 200, {"state": "FAILURE_DIAGNOSIS_STATE_GROUPED_FAILURES", "failureGroups": groups}, method+ )return _response(200, workflows_by_id[json["workflowId"]], method)session = mock.MagicMock()@@ -96,9 +125,11 @@ def _batches(response: SourceResponse) -> list[list[dict[str, Any]]]:return list(cast(Iterable[list[dict[str, Any]]], response.items()))-def _synced_rows(session: mock.MagicMock, created_after: dt.datetime | str | None) -> list[dict[str, Any]]:+def _synced_rows(+ session: mock.MagicMock, created_after: dt.datetime | str | None, endpoint: str = JOB_ATTEMPTS+) -> list[dict[str, Any]]:with mock.patch(f"{MODULE}.make_tracked_session", return_value=session):- batches = _batches(depot_source(API_TOKEN, REPOSITORY, created_after, mock.MagicMock()))+ batches = _batches(depot_source(API_TOKEN, REPOSITORY, created_after, mock.MagicMock(), endpoint=endpoint))return [row for batch in batches for row in batch]@@ -147,17 +178,75 @@ class TestDepotSource:assert [row["run_id"] for row in rows] == expected_run_ids+ @pytest.mark.parametrize(+ "endpoint, columns, expected_rows, expected_calls",+ [+ (+ JOB_METRICS,+ ("run_id", "attempt_id", "peak_memory_usage_bytes"),+ [("failed", "failed-attempt", 1024), ("ok", "ok-attempt", 1024)],+ [("GetJobMetrics", {"jobId": "failed-job"}), ("GetJobMetrics", {"jobId": "ok-job"})],+ ),+ (+ ARTIFACTS,+ ("run_id", "artifact_id", "size_bytes"),+ [+ ("failed", "failed-a1", 2048),+ ("failed", "failed-a2", 4096),+ ("ok", "ok-a1", 2048),+ ("ok", "ok-a2", 4096),+ ],+ [+ ("ListArtifacts", {"runId": "failed", "pageSize": 500}),+ ("ListArtifacts", {"runId": "failed", "pageSize": 500, "pageToken": "p2"}),+ ("ListArtifacts", {"runId": "ok", "pageSize": 500}),+ ("ListArtifacts", {"runId": "ok", "pageSize": 500, "pageToken": "p2"}),+ ],+ ),+ (+ FAILURE_DIAGNOSES,+ ("run_id", "fingerprint", "count"),+ [("failed", "fp-1", 2)],+ [("GetFailureDiagnosis", {"targetId": "failed", "targetType": "FAILURE_DIAGNOSIS_TARGET_TYPE_RUN"})],+ ),+ ],+ )+ def test_run_child_tables(+ self,+ endpoint: str,+ columns: tuple[str, ...],+ expected_rows: list[tuple[Any, ...]],+ expected_calls: list[tuple[str, dict[str, Any]]],+ ) -> None:+ runs = [{**_run("failed", dt.timedelta(hours=1)), "status": "failed"}, _run("ok", dt.timedelta(hours=2))]+ workflows = {run["runId"]: [_single_attempt_workflow(run)] for run in runs}+ workflows["ok"][0]["jobs"].append({"jobId": "ok-skipped-job", "attempts": []})+ session = _fake_session(runs, workflows_by_run=workflows)++ rows = _synced_rows(session, None, endpoint)++ assert [tuple(row[column] for column in columns) for row in rows] == expected_rows+ assert all(row["run_created_at"] for row in rows)+ called_method = expected_calls[0][0]+ calls = [+ (call.args[0].rsplit("/", 1)[-1], call.kwargs["json"])+ for call in session.post.call_args_list+ if call.args[0].endswith(f"/{called_method}")+ ]+ assert calls == expected_calls+class TestDepotReconciliation:@pytest.mark.parametrize(- "reconciled_days_ago, history_days, has_watermark, expected_runs, expected_reconciled",+ "reconciled_days_ago, history_days, has_watermark, scanned_without_rows, expected_runs, expected_reconciled",[- (None, 30, True, ["recent", "old"], True),- (1, 30, True, ["recent"], False),- (7, 30, True, ["recent", "old"], True),- (1, 30, False, ["recent", "old"], True),- (None, None, True, ["recent", "old", "outside"], True),- (1, 1, True, [], False),+ (None, 30, True, False, ["recent", "old"], True),+ (1, 30, True, False, ["recent"], False),+ (7, 30, True, False, ["recent", "old"], True),+ (1, 30, False, False, ["recent", "old"], True),+ (1, 30, False, True, ["recent"], False),+ (None, None, True, False, ["recent", "old", "outside"], True),+ (1, 1, True, False, [], False),],)def test_replays_retries_and_reconciles_retained_history(@@ -165,12 +254,16 @@ class TestDepotReconciliation:reconciled_days_ago: int | None,history_days: int | None,has_watermark: bool,+ scanned_without_rows: bool,expected_runs: list[str],expected_reconciled: bool,) -> None:source = DepotSource()cursor = (- DepotReconciliationCursor(reconciled_at=_iso(NOW - dt.timedelta(days=reconciled_days_ago)))+ DepotReconciliationCursor(+ reconciled_at=_iso(NOW - dt.timedelta(days=reconciled_days_ago)),+ scanned_through=_iso(NOW - dt.timedelta(minutes=10)) if scanned_without_rows else None,+ )if reconciled_days_ago is not Noneelse None)@@ -180,6 +273,7 @@ class TestDepotReconciliation:db_incremental_field_last_value=_iso(NOW - dt.timedelta(minutes=10)) if has_watermark else None,history_start=NOW - dt.timedelta(days=history_days) if history_days is not None else None,source_cursor=manager,+ schema_name=JOB_ATTEMPTS,)runs = [_run("recent", dt.timedelta(days=2)),@@ -198,12 +292,19 @@ class TestDepotReconciliation:rows = [row for batch in _batches(response) for row in batch]assert [row["run_id"] for row in rows if row["attempt"] == 2] == expected_runs- assert (manager.staged is not None) == expected_reconciled+ # mypy keeps the narrowing from the assertion before the sync ran.+ staged = cast(DepotReconciliationCursor | None, manager.staged)+ assert staged is not None+ assert (staged.reconciled_at == NOW.isoformat()) == expected_reconciled+ if expected_runs:+ assert staged.scanned_through == runs[0]["createdAt"]def test_incomplete_reconciliation_does_not_advance_the_cursor(self) -> None:source = DepotSource()manager = SourceCursorManager(source.cursor_class(), None, source)- inputs = mock.MagicMock(should_use_incremental_field=False, history_start=None, source_cursor=manager)+ inputs = mock.MagicMock(+ should_use_incremental_field=False, history_start=None, source_cursor=manager, schema_name=JOB_ATTEMPTS+ )runs = [_run("recent", dt.timedelta(days=2)), _run("old", dt.timedelta(days=14))]session = _fake_session(runs)post = session.post.side_effectdiff --git a/products/warehouse_sources/backend/temporal/data_imports/sources/depot/tests/test_depot_source.py b/products/warehouse_sources/backend/temporal/data_imports/sources/depot/tests/test_depot_source.pyindex cce130c51b2f..6df69b2ed5cf 100644--- a/products/warehouse_sources/backend/temporal/data_imports/sources/depot/tests/test_depot_source.py+++ b/products/warehouse_sources/backend/temporal/data_imports/sources/depot/tests/test_depot_source.py@@ -66,7 +66,7 @@ class TestDepotSource:source = DepotSource()session = mock.MagicMock()session.post.return_value = _response(status, reason, body)- inputs = mock.MagicMock(should_use_incremental_field=False, history_start=None)+ inputs = mock.MagicMock(should_use_incremental_field=False, history_start=None, schema_name="job_attempts")with mock.patch(f"{MODULE}.make_tracked_session", return_value=session):response = source.source_for_pipeline(DepotSourceConfig(api_token="token", repository="a/b"), inputs)