go get

Operator for monitoring liveness of Routes and URL's on OpenShift

by openshift

v0.0.0-20211117122737-fed903f1b3fe (see all)License:Apache-2.0
go get


What does this do?

Automatically enables blackbox probes for routes on OpenShift clusters to be consumed by the Cluster Monitoring Operator or any vanilla Prometheus Operator.

How does this work?


The operator is making sure that there is one deployment + service of the blackbox exporter. If it does not exist in openshift-monitoring, it creates one.


The probes are effectively configured via ServiceMonitors, see more details in Prometheus Operator troubleshooting docs. openshift-route-monitor-operator creates ServiceMonitors based on the defined RouteMonitors.


The operator watches all namespaces for routeMonitors. They are used to define what route to probe. RouteMonitors are namespace scoped and can reference Routes from other namespaces.


The operator watches all namespaces for ClusterUrlMonitors.

They are used to define what URL to probe, based on the cluster domain to allow monitoring of URLs of applications deployed to the cluster, which do not make use of a Route (i.e. the api server). A ClusterUrlMonitor consists of a prefix, a port, and a suffix which make up the probed URL as follows:


Getting prefix and suffix right is in the users' responsibility. In most cases the prefix will end with a . while the suffix will start with a / but this is not checked or fixed by the controller. ClusterUrlMonitors are namespace scoped.


The operator implements Multiwindow, Multi-Burn-Rate Alerts in a unique way.

The official calculation for Multiwindow Multi-burn alerting will only successfully work on services and applications that already have a sufficient baseline of metrics to work and calculate availability upon. In a case of newly created services, alerts will fire immediately and will only be remediated once there's enough data present. In some cases a user might want to create a Monitor for a newly created Route or ClusterUrl. To support this, the operator takes into account the overall number of existing probes, in a way that if there are no sufficient probes (yet), an alert will not fire.


Currently the blackbox exporter deployment is only using the default config file which only allows a limit set of probes.


In order to develop the repo follow these steps to get an env started:

  1. run make test to test build and deploy
  2. change Makefile and config/manager/manager.yaml to point to the repo you wish to use
  3. build and deploy with make docker-build docker-push 3.1. if you want to use a local image use IMG=
  4. use make deploy to deploy your operator on a cluster you are logged into 4.1. this also can have the IMG
  5. check logs with oc logs -n openshift-monitoring deploy/route-monitor-operator-controller-manager -c manager
  6. retrigger pull of pod with oc delete -n openshift-monitoring -lapp=route-monitor-operator,component=operator

Test operator locally

The makefile has a command to run the operator locally:

make run

Running integration tests

The integration test suite is located in int/. You can execute the test suite against a cluster you are currently logged into. It will expose the registry of that cluster, push the operator to it and execute a suite of tests to ensure the operator does it's work. Tested with running crc locally and being logged in as kubeadmin. If you want to run the tests as a different user, set the KUBEUSER environment variable before executing the tests, and make sure you're logged in as that user to the cluster.

make test-integration


  • add option to specify which probes to use
  • make service monitor use a different interval via modifying a line in the spec of route monitor

GitHub Stars



3mos ago








6mos ago
6mos ago
7mos ago
7mos ago
No alternatives found
No tutorials found
Add a tutorial