Whose wait it was: the forge's or the app's (#10858)

An app's production speed card says the slowest wait the forge relayed to it in the last 10 minutes, and whether the forge or the app spent it. When the forge's own event-loop lag (forge-lag) is more than half the time to the app's headers (forge-ttfb), the wait is the forge's (forgeWaited, src/perf/loop_worst.ts), and the card says so instead of calling the app slow.

What held it (#11000)

When the wait was the forge's, the forge names what blocked its loop in the lag's 10 s window: the stalls the loop watch (src/perf/loop.ts) noted, each by the route patterns that ran across it, else the background jobs as job:<name>. Longest stall first, each name once, at most three. A name is only ever a pattern — a path pattern such as /:owner/:name, job:<name> or (unmatched) — never a path's ids or query (isStallPattern, src/apps/relay_wait.ts); a stall nothing named ran across is left out. The card and terminal line add , held by <name> (<ms>)…; no held by when the wait was the app's own.

A signed-in forge admin (active, not an agent) also gets the names on the relayed answer itself, beside forge-lag: Server-Timing: forge-stall;desc="job:sweep-runs";dur=600.0, one entry per name. That copy is sent cache-control: private, no-store (and without etag), so no shared cache keeps the forge's route names; visitors and everyone else never get forge-stall.

Production only; previews are not counted. Held in the forge's memory (at most 2,000 apps), so a restart forgets it.

Try it: hold a production app's answer behind a busy forge (a long synchronous task on the forge, then load the app), then open its repository's Settings → App within 10 minutes. Signed in as a forge admin, the app's own response carries forge-stall too.

↑↓ navigate ↵ open ⌘↵ its app esc close

Keyboard shortcuts

Docs