M26-5 Lazy lifecycle metrics + SSE events #592

New issue

Closed

opened 2026-04-30 19:31:52 +00:00 by claude-desktop · 0 comments

claude-desktop commented

2026-04-30 19:31:52 +00:00

Collaborator

User story

As a service operator, I want lazy starts, stops, and start failures to emit observable events on the SSE bus, so that the dashboard can show cold-start latency, idle stops, and failures without me tailing logs.

Acceptance criteria

Events

New SSE envelope types on /events:
- container.lazy_started { instance, latency_ms } — emitted after readiness probe succeeds.
- container.lazy_stopped { instance, idle_seconds } — emitted after docker stop returns.
- container.lazy_start_failed { instance, error } — readiness probe failed or docker start errored.
Wired into the existing event log so the dashboard's Agents page surfaces them without schema work.
Events also written to the SQLite event_log so they survive service restart for post-hoc audit.

Dashboard surface

Agents page Activity timeline includes the three event types with appropriate icons and human-readable text ("reviewer-default started in 287 ms" / "designer-default stopped after 12 min idle").
No new aggregate widgets in this ticket — single-event rendering is enough; rollups can come later if needed.

Tests

Lifecycle module emits each event at the right transition (test via injected event sink).
Latency value is monotonic (Date.now() based, not random clock skew).
Failure path: docker start rejects → lazy_start_failed emitted, no lazy_started, dispatch surfaces error.

Out of scope

Lifecycle module (M26-1).
Reconcile + watchdog (M26-2).
Pool selection (M26-3).
Config + dashboard CRUD (M26-4).
Aggregate widgets (cold-start histograms, idle-time distribution) — follow-up if telemetry justifies.

References

specs/container-lazy-lifecycle.md §Metrics + events.
apps/server/src/infrastructure/event-log.ts — existing event log.
apps/web/ Agents page activity timeline.

## User story As a service operator, I want lazy starts, stops, and start failures to emit observable events on the SSE bus, so that the dashboard can show cold-start latency, idle stops, and failures without me tailing logs. ## Acceptance criteria ### Events - [ ] New SSE envelope types on `/events`: - `container.lazy_started` `{ instance, latency_ms }` — emitted after readiness probe succeeds. - `container.lazy_stopped` `{ instance, idle_seconds }` — emitted after `docker stop` returns. - `container.lazy_start_failed` `{ instance, error }` — readiness probe failed or `docker start` errored. - [ ] Wired into the existing event log so the dashboard's Agents page surfaces them without schema work. - [ ] Events also written to the SQLite `event_log` so they survive service restart for post-hoc audit. ### Dashboard surface - [ ] Agents page Activity timeline includes the three event types with appropriate icons and human-readable text (`"reviewer-default started in 287 ms"` / `"designer-default stopped after 12 min idle"`). - [ ] No new aggregate widgets in this ticket — single-event rendering is enough; rollups can come later if needed. ### Tests - [ ] Lifecycle module emits each event at the right transition (test via injected event sink). - [ ] Latency value is monotonic (`Date.now()` based, not random clock skew). - [ ] Failure path: `docker start` rejects → `lazy_start_failed` emitted, no `lazy_started`, dispatch surfaces error. ## Out of scope - Lifecycle module (M26-1). - Reconcile + watchdog (M26-2). - Pool selection (M26-3). - Config + dashboard CRUD (M26-4). - Aggregate widgets (cold-start histograms, idle-time distribution) — follow-up if telemetry justifies. ## References - `specs/container-lazy-lifecycle.md` §Metrics + events. - `apps/server/src/infrastructure/event-log.ts` — existing event log. - `apps/web/` Agents page activity timeline.

claude-desktop added the

area:agents

type:user-story

labels

2026-04-30 19:32:12 +00:00

claude-desktop added this to the M26 — Container lazy lifecycle milestone

2026-04-30 19:32:20 +00:00