Werk #22430: Prometheus alerting rules special agent: clarify that it queries Prometheus, not Alertmanager
| Component | Checks & agents | ||||||
| Title | Prometheus alerting rules special agent: clarify that it queries Prometheus, not Alertmanager | ||||||
| Date | Sep 16, 2026 | ||||||
| Level | Trivial Change | ||||||
| Class | Bug Fix | ||||||
| Compatibility | Compatible - no manual interaction needed | ||||||
| Checkmk versions & editions |
|
The special agent formerly presented in Setup as "Prometheus Alertmanager" does
not talk to Alertmanager at all. It queries the Prometheus server HTTP API
(GET /api/v1/rules) to read the configured alerting rules and their current
state. Alertmanager does not offer this endpoint, so users who pointed the agent
at an Alertmanager instance (usually port 9093) received HTTP 404 errors, and
users could wrongly assume that silences, receivers or Alertmanager cluster
status are monitored by this agent.
The wording is now corrected:
- The Setup ruleset is titled "Prometheus alerting rules (via Prometheus API)".
- The help text of the "URL server address" option states that the address
must be the Prometheus server (typically port 9090), that
/api/v1/rulesis queried, and that Alertmanager itself is not supported by this agent. - The manual pages of the three checks are now titled "Prometheus: Alerting rule", "Prometheus: Alerting rule group" and "Prometheus: Alerting rule summary" instead of "Alertmanager: Rules", "Alertmanager: Groups" and "Alertmanager: Summary".
Only names and help texts change. The ruleset ID, the agent section
<<<alertmanager>>>, the check plug-in names and the service names are
unchanged, so existing rules, discovered services and their history are not
affected. A deeper rename or merging this agent into the Prometheus special
agent is deliberately left for a separate change, as that would affect
persisted state.