With the advent of version 2 of OpsGenie Heartbeats, this proxy is obsolete. If you want to send heartbeats without using this proxy, you need to migrate existing v1 hearbeat to v2 which is quite easy. See instructions in the Hearbeat settings page.
Use a receiver definition like this (inspired by this comment):
- name: opsgenie
webhook_configs:
- url: 'https://api.opsgenie.com/v2/heartbeats/<the-heartbeat-name>/ping'
send_resolved: false
http_config:
basic_auth:
password: 123e4567-e89b-12d3-a456-426655440000
Proxy Prometheus Alertmanager webhooks to OpsGenie Heartbeat API to work around issue discussed here prometheus/alertmanager#444
Docker container can be pulled here traumfewo/opsgenie-heartbeat-proxy.
Provide OpsGenie API Key and Heartbeat name as env variable.
docker run -e OPSGENIE_API_KEY='<API-KEY>' -e HEARTBEAT_NAME='<HEARTBEAT_NAME>' -d -p 8080:8080 traumfewo/opsgenie-heartbeat-proxy:latest
For debbuging purpose you can use optional DEBUG env.
Setup an always firing prometheus alert
ALERT OpsGenieHeartBeat
IF vector(1)
LABELS { app = "opsgenie" }
ANNOTATIONS {
description = "Ping OpsGenie heartbeat api.",
}
which invokes a webhook to trigger proxy script running in docker container
Alertmanager route:
- match:
app: opsgenie
receiver: opsgenieproxy
continue: false
Alertmanager receiver:
- name: opsgenieproxy
webhook_configs:
- send_resolved: false
url: http://<container-ip>:8080/proxy
If you want to use opsgenie heartbeat proxy for a prometheus setup on kubernetes you can use the yaml file from examples to deploy the proxy to your running kubernetes cluster. If you run prometheus inside your kubernetes cluster the service type "ClusterIP" should be sufficient. Otherwise, you can change the kubernetes service type to "LoadBalancer" to request a public routable ip.