SLO Manager

Service: slo-manager · Port: 8100 · Database schema: slo

slo-manager tracks service-level objectives (SLOs): the reliability targets you promise for your services. It evaluates each SLO against live Prometheus data, tells you how much error budget is left and how fast it's burning, and alerts you before you run out.

Key ideas

Term	Meaning	Example
SLI (indicator)	A measurement between 0 and 1 of good behaviour.	Share of requests that succeed.
SLO target	The minimum SLI you commit to over a rolling window.	99.9% over 30 days.
Error budget	How much badness the target allows: 1 − target.	0.1% of requests may fail.
Burn rate	How fast the budget is being used: (1 − SLI) / (1 − target).	2.0 = the budget runs out in half the window.

A burn rate of 1 uses the budget exactly over the window. Higher means you'll run out early.

Define an SLO

In Monitoring → SLOs → New SLO, choose a service and a type, and KubeOpera writes the query for you from your app's standard HTTP metrics. Or use the API:

{
  "name": "payments-api availability",
  "service": "payments-api",
  "namespace": "production",
  "type": "availability",
  "target": 0.999,
  "window_days": 30
}

Type	Measures
availability	Share of requests without a 5xx error.
latency	Share of requests faster than a threshold (threshold_ms).
error_rate	Share of requests without any error.
throughput	Whether request rate stays above a minimum.
custom	Any PromQL expression you supply in prom_query.

Burn-rate alerts

slo-manager alerts on multi-window burn rates, the approach recommended by the Google SRE workbook — fast enough to catch sudden outages, calm enough to ignore brief blips:

Alert	Condition	Meaning
Page	Burn rate > 14.4 over both 1h and 5m	2% of a 30-day budget gone in an hour.
Page	Burn rate > 6 over both 6h and 30m	5% gone in six hours.
Ticket	Burn rate > 1 over both 3d and 6h	Steadily trending toward running out.

Alerts go to the incident manager, which opens an incident and notifies your channels.

REST API

Method	Path	Description
GET · POST	/api/v1/slos	List or create SLOs.
GET · PUT · DELETE	/api/v1/slos/{id}	SLO definition.
GET	/api/v1/slos/{id}/status	Live SLI, error budget remaining, burn rates and breach state.
GET	/api/v1/slos/{id}/history	Budget over time (?from=&to=).
GET	/api/v1/slos/summary	Totals: healthy, at risk and breaching.
GET	/healthz	Health check.

Configuration

Variable	Default	Description
DATABASE_URL	—	PostgreSQL connection.
PROMETHEUS_URL	http://kube-prometheus-stack-prometheus.monitoring:9090	Prometheus.
EVALUATION_INTERVAL	1m	How often SLOs are evaluated for alerting.
RABBITMQ_URL	—	Publishes burn-rate alerts.
AUTH_JWT_ACCESS_SECRET	—	Validates tokens.
PORT	8100	HTTP port.