Exposes Rancher metrics to Prometheus
Addition of Agent-State thanks to @johnrengelman and minor fix to internal metrics.
Agent state is now tracked for hosts. The agent state is what is set when a host connects/disconnects from the Rancher server. Useful for when hosts enter 're-connecting' state.
Fixes for V1 API backwards compatibility
HIDE_SYS
Now set to true by default. To view system services then set this bool value to false.This release encompasses somewhat of a re-write for how we handled data. In summary, the previous exporter was split into vertical functional stacks around the endpoint (hosts,services,stacks). This meant that we needed to re-write it so that code traverses a central set of functions. This proved a blocker for some of the new functionality.
This is also the first version to use the new version format courtesy of https://github.com/a-h/ver/ .
rancherURL
removed as a label, we don't support multiple rancher servers at this stage and it makes the output hard to parse. Ultimately the aim here longer term is to return the 'environment', this requires an upstream change to the API though.stack_name
field now included for all service metricsunhealthy
state, inline with how it's processed in the other state metrics.Stable Release of the go-rewrite codebase.
This is the last version that will use the existing version scheme, the next iteration will use a more robust versioning scheme.