Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
31 changes: 26 additions & 5 deletions src/analysis/post_run/fetch.py
Original file line number Diff line number Diff line change
Expand Up @@ -15,7 +15,9 @@
DASHBOARD_URL,
SCRAPE_LABEL,
TIMEOUT_S,
UNCHANGED,
metric_dataset,
run_pods,
run_title,
)

Expand Down Expand Up @@ -69,15 +71,33 @@ def _query_response(rows: List[dict]) -> dict:
return {"data": {"result": [{"metric": m, "values": v} for m, v in series.values()]}}


def run_dataset(experiment: dict, name: str, url: str = DASHBOARD_URL) -> List[dict]:
"""A dataset's rows; a metric under UNCHANGED_RULE is one row per pod at the end of the run."""
metadata = experiment["metadata"]
unchanged = metadata.get(UNCHANGED, {}).get("metrics", {})
if name in unchanged:
end, same = metadata["results"]["complete"]["end"], unchanged[name]
return [
{"timestamp": end, "value": same["value"], "pod": pod, **same["labels"]}
for pod in run_pods(metadata["stack"])
]
if name not in {d["name"] for d in experiment.get("datasets", [])}:
raise LookupError(f"`{experiment['title']}` has no `{name}` dataset")
response = requests.get(
f"{url}/experiments/{experiment['id']}/datasets/{name}/data", timeout=TIMEOUT_S
)
if response.status_code == 404:
raise LookupError(f"`{experiment['title']}` has no stored `{name}` data")
response.raise_for_status()
return response.json()["data"]


def fetch_run_metrics(
run_name: str, dump: Path, url: str = DASHBOARD_URL, title: Optional[str] = None
) -> Path:
"""Write the run's stored datasets to `dump` with the scraper's own CSV writer."""
experiment = find_experiment(run_name, url, title)
queries = {d["name"]: d["query"] for d in experiment.get("datasets", [])}
response = requests.get(f"{url}/experiments/{experiment['id']}/datasets", timeout=TIMEOUT_S)
response.raise_for_status()
stored = {d["name"]: d["data"] for d in response.json()["datasets"]}

partial = dump.with_name(f"{dump.name}.partial")
shutil.rmtree(partial, ignore_errors=True)
Expand All @@ -91,9 +111,10 @@ def fetch_run_metrics(
query, current = queries[name], metric_dataset(metric, config)["query"]
if (query.get("expr"), query.get("step")) != (current["expr"], current["step"]):
raise ValueError(f"`{run_name}` stored {name} as {query}, not as today's {current}")
if not stored.get(name):
rows = run_dataset(experiment, name, url)
if not rows:
raise LookupError(f"`{run_name}` has no stored rows for {name}")
handler = DataRequestHandler(_query_response(stored[name]))
handler = DataRequestHandler(_query_response(rows))
handler.create_dataframe_from_request(
metric.extract_field, metric.container, metric.metrics_path
)
Expand Down
Loading
Loading