mirror of
https://github.com/CopilotKit/CopilotKit.git
synced 2026-09-14 16:26:20 +08:00
Jordan Ritter
0522b7fa41
The monitoring/alerting service is a test harness (probes, assertions, alerting), not an operations service. Rename the directory, package name (@copilotkit/showcase-ops → @copilotkit/showcase-harness), all internal references (Dockerfile, Prometheus metric prefix showcase_ops_ → showcase_harness_, orchestrator log messages, probe YAML nameExcludes, test fixtures), and regenerate pnpm-lock.yaml. Wire protocol names (X-Ops-* headers) and shell-dashboard internal API naming (OPS_BASE_URL, ops-api.ts) are intentionally unchanged — they are stable contracts between sender and receiver.
40 lines
1.7 KiB
YAML
40 lines
1.7 KiB
YAML
id: smoke-red-fleet
|
|
name: "Fleet-wide smoke-red escalation"
|
|
owner: "@oss"
|
|
|
|
# Cross-service aggregation (plan Item 4). When N >= 3 smoke-red signals land
|
|
# within a 2-minute window — covers one 90s probe tick plus slack — the engine
|
|
# emits ONE composite channel ping instead of N per-service red-ticks. The
|
|
# per-service smoke-red-tick YAML still fires its own alert (routed to the
|
|
# same channel via the rate-limit / dedupe gates); the fleet rule exists
|
|
# specifically to page on-call when the problem is fleet-wide rather than
|
|
# per-starter.
|
|
signal:
|
|
dimension: smoke
|
|
|
|
# Fleet aggregation is for red-only signals. `red_to_green` recoveries do NOT
|
|
# ingest (see alert-engine.ts aggregation ingress — A1); declaring it here
|
|
# would be decorative but misleading.
|
|
triggers:
|
|
- green_to_red
|
|
- sustained_red
|
|
|
|
# A7: groupBy omitted — the rule's `signal.dimension: smoke` already partitions
|
|
# traffic to smoke-only events, so there's no finer partition to apply. The
|
|
# bucket key collapses to `rule.id` alone (see buildBucketKey in aggregation.ts).
|
|
aggregation:
|
|
# windowMs: 2 min — one probe tick (90s) plus slack for signal arrival jitter.
|
|
windowMs: 120000
|
|
minMatches: 3
|
|
# The channel-ping token below is intentional and load-bearing: per-service
|
|
# red-tick YAMLs no longer page channel; this rule is the single pager.
|
|
template: |
|
|
<!channel> smoke red across fleet — {{count}} services: {{services}}
|
|
|
|
# rate_limit.window spans multiple aggregation windows so back-to-back 2-minute
|
|
# buckets with the same groupValues collapse to one dispatch. Composite dedupe
|
|
# key is stable across windows (see buildCompositeDedupeKey in aggregation.ts).
|
|
conditions:
|
|
rate_limit:
|
|
window: 10m
|