prometheus/prometheus

kubernetes_sd_config (alertmanagers): reloading config doesn't work

已關閉

#11,013 建立於 2022年7月13日

 (6 則留言) (2 個反應) (1 位負責人)Go (10,408 個分叉)batch import
component/service discoverycomponent/service discovery/kuberneteshelp wantedkind/bugpriority/P2

倉庫指標

星標
 (64,042 顆星)
PR 合併指標
 (平均合併 11天 5小時) (30 天內合併 118 個 PR)

描述

What did you do?

It is possible to automatically discovery alertmanagers if they are hosted in Kubernetes, something like this:

alerting:
  alertmanagers:
    - kubernetes_sd_configs:
        - role: pod
          namespaces:
            names:
              - monitoring
      relabel_configs:
        - source_labels: [__meta_kubernetes_pod_label_app_kubernetes_io_name]
          regex: alertmanager
          action: keep

What did you expect to see?

Updating the configuration when adding new alertmanagers.

What did you see instead? Under which circumstances?

Prometheus only discovery alertmanagers at startup. If the pods with the alertmanagers were recreated (and their ip changed), then Prometheus will not know about them.

System information

No response

Prometheus version

Version	2.36.2
Revision	d7e7b8e04b5ecdc1dd153534ba376a622b72741b
Branch	HEAD
BuildUser	root@f051ce0d6050
BuildDate	20220620-13:21:35
GoVersion	go1.18.3

Prometheus configuration file

...
    alerting:
      alertmanagers:
        - kubernetes_sd_configs:
            - role: pod
              namespaces:
                names:
                  - monitoring
          relabel_configs:
            - source_labels: [__meta_kubernetes_pod_label_app_kubernetes_io_name]
              regex: alertmanager
              action: keep
...

Alertmanager version

No response

Alertmanager configuration file

No response

Logs

No response

貢獻者指南