Kubernetes: OpenShift Cluster Collector
The Kubernetes monitoring gathers information about an OpenShift cluster by
connecting to a Prometheus endpoint. This service monitors the connection to
Prometheus. Moreover, for each Prometheus query one of the following errors
- Request exception
- Invalid response
- Prometheus error
- Querying endpoint succeeded, but no samples received
- Incompatible data received: data did not match format expected from
If only incompatible data is received, the service goes
UNKNOWN. If any
other error occurs, the state of the service goes
CRIT. If all queries
succeed, the service is
OK. If multiple differing errors occur, each error
is displayed together with the correspond queries.
`Request exception` indicates that agent_kube was unable to obtain a response
from the endpoint. Examples of such errors are ssl, proxy and timeout errors.
In case the monitoring of the endpoint is setup for the first time, it is
advisable to double-check the configuration. Specifying an incorrect scheme or
token is a common source of this error. Moreover, details of the service
contain the precise endpoint, which is contacted by Checkmk. Using the
`--debug` flag of `agent_kube` may provide additional information.
`Invalid response` indicates that the contacted server did not respond in a
way that is expected from a Prometheus endpoint. Checkmk will only process
data, which matches the HTTP API specified by Prometheus. A common source of
this error is that a proxy server or an OAuth server interjected itself in the
communication. Debugging can be done in similiar manner as for `Request
`Prometheus error` indicates that Prometheus reported an error while
responding to the query. In such a case, Checkmk does not process any data,
even though some data may be transmitted. Checkmk will provide the error code
and message as reported by Prometheus.
`Querying endpoint succeeded, but no samples received` indicates that the
query succeeded. However, the metric requested by Checkmk could not be
obtained. Which metrics are available depends on the collectors inside of
cluster and how they are scraped by Prometheus. Checkmk only maintains
metrics which match the internals of an OpenShift cluster.
`Incompatible data received` indicates that the labels of the received metrics
did not match those expected by Checkmk. The labels depend on the collectors
of the cluster.
Note, that the information displayed by this service changes, if the
corresponding collector is changed.
One service is created if the option `Use data from Openshift` is set.