Expand description
Aggregate delivery latency and outcome metrics.
These answer “is push slow?” and “is push reliable?” for all traffic without
tracking individual messages. Latency is recorded in Prometheus histograms
(served from /metrics) so quantiles can be aggregated across pods and time
windows. Terminal outcomes are statsd counters (notification.outcome).
Structs§
- Delivery
Metrics - The process wide Prometheus registry and the histograms it holds.
Enums§
- Cause
- Who is responsible for a non-delivery.
- Delivery
Path - The destination router of a notification.
- Delivery
Route - Whether the notification was read from storage before delivery.
- Outcome
- The terminal status of a notification.
Constants§
- CONTENT_
TYPE - Content type for the
/metricsresponse. - DELIVERY_
BUCKETS - Latency histogram bucket boundaries, in seconds.
Functions§
- emit_
outcome - Increment
notification.outcome. - emit_
outcomes - Add
counttonotification.outcome.