Publisher run duration distribution (percentiles)
How long my apps’ runs take: p50 / p90 / p95 / p99 / max of
finished_at - started_at in milliseconds over a bounded range,
optionally per app or per app version (slowest p95 first). Same
ownership, range and filter semantics as GET /publisher/usage; the
sample is every finished run that has a start time (failed and
cancelled runs included — the distribution describes the runs, not
only the successful ones), and every percentile is null when there
is none. Percentiles need run details, so both bounds are required
and may be at most 92 days apart (400 range-too-wide).
Authorizations
DataHub API Key
Query Parameters
optional single dimension: data_app (per app) or version (per app version); default = range totals only
restrict to these apps (<username>/<app_name>, repeatable); default = all my apps
restrict to one calling channel (same value as the runs list filter)
Response
Successful Response
Publisher-side duration distribution over a bounded range, optionally broken down by app or by app version (slowest p95 first).