e2ealerting helps measure the end to end alerting latency between Prometheus scrapes and Alertmanager notification delivery. We achieve that by creating a metric that exposes the current time as its value, then creating an alert definition that uses that value as an annotation, and finally a built-in Alertmanager webhook receiver observes the duration between the value of the annotation and webhook reception time as part of a Histogram.
A client for synchorinizing Prometheus rules definition and Alertmanager configuration allows us to experiment with different values directly within the tool without having to turn knobs configuration in different places. Keep in mind, that these are meant to only support the Cortex configuration API for the Ruler and Alertmanager.
Refer to the diagram below for a brief explanation on how this is all connected.
┌─────────────────────────────────────┐
┌─────────────┐ │ ┌─────────────┐ +-+-+-+-+-+-+│
│ Prometheus │ │ │ │ |C|o|r|t|e|x|│
┌──────│ or │──────Remote Write─────────┼─▶│ Remote Write│ +-+-+-+-+-+-+│
│ │ Agent │ │ │ Endpoint │ │
│ └─────────────┘ │ └─────────────┘ │
Scrape │ │
│ │ │
│ ┌──────────Alert Notification Webhook──┼─────────┐ │
│ │ │ │ │
│ ▼ │ │ │
│ ┌─────────────┐ │ ┌─────────────┐ │
│ │ │ │ │ │ │
└─▶│ e2ealerting │──────Alertmanager Config──────┼─▶│ Alertmanager│ │
│ │ │ │ │ │
└─────────────┘ │ └─────────────┘ │
│ │ ┌─────────────┐ │
│ │ │ │ │
└──────────────Rule Group Config───────┼─▶│ Ruler │ │
│ │ │ │
│ └─────────────┘ │
└─────────────────────────────────────┘
You can compile the binary directly by cloning this repo and running make cmd/e2ealerting/e2ealerting. The binary will be located in cmd/e2ealerting/e2ealerting.
Docker images are part of grafana/e2ealerting in DockerHub.
An example e2ealerting configuration would look like:
-configs.key=<your-api-key>
-configs.alertmanager-file=/data/alertmanger_config.yaml
-configs.alertmanager-id=<alertmanager-id>
-configs.alertmanager-url=https://alertmanager-us-central1.grafana.net
-configs.rulegroup-file=/data/ruler_config.yaml
-configs.ruler-id=<prometheus-id>
-configs.ruler-url=https://prometheus-us-central1.grafana.net
-configs.sync-interval=2h
Please note the both the alertmanager and ruler configuration are optional. If not defined as part of the config, the syncing process is disabled.
The Alertmanager configuration file alertmanger_config.yaml can be defined as:
global:
receivers:
- name: e2e-alerting
webhook_configs:
- url: <e2ealerting-url>/api/v1/receiver
route:
group_interval: 1s
group_wait: 1s
receiver: e2e-alerting
repeat_interval: 1sFinally, the rule configuration file ruler_config.yaml:
name: e2ealerting
rules:
- alert: E2EAlertingAlwaysFiring
annotations:
time: '{{ $value }}'
expr: e2ealerting_now_in_seconds > 0
for: 10s