Where the window's requests went: the top 10 networks plus one "other"
total, the top status codes and methods, the top 10 client IPs (each with
its error rate and egress) plus one "other" total, every traffic source
with a trend, and p50/p90/p95 latency over
time — for the same window, source and filters as /access/chart, so a
zoom updates it. Shares are of total_requests in the same response.
Source sparklines share the KPI sparklines' grid (at most 30 points,
sources_step_ms wide); the latency series has its own of at most 60.
Accepts every filter[...] field /access/events does except event_id,
plus search. When a raw read trips its bound the response is
capped: true: total_requests and every *_other are null and the lists are empty.
| Time | Status | User Agent | |
|---|---|---|---|
Retrieving recent requests… | |||