Page on-call when flaky failures spike into a suite meltdown

Distinguishes ordinary background flakiness from a sudden systemic failure — a bad dependency bump, a broken shared fixture, or infra degradation — that masquerades as a flood of "flaky" tests. When the quarantine rate and overall failure volume both spike together, it escalates instead of silently quarantining everything.

Use this as a safety net on top of an automated quarantine program, so the system never quietly hides a real outage by quarantining dozens of specs at once.

1. 1A webhook fires each time a spec is quarantined, carrying the running counts.
2. 2The flow queries Postgres for the number of newly quarantined specs and total CI failures in the trailing time window.
3. 3A logic gate checks whether both the quarantine rate and failure volume exceed their meltdown thresholds simultaneously.
4. 4If the gate trips, it triggers a PagerDuty incident for the on-call engineer.
5. 5It also posts a meltdown alert to Slack summarizing the spike and the affected specs so the team can converge immediately.