A new version of check_rancher2, a monitoring plugin to monitor Kubernetes environments managed by Rancher 2.x, is available!
The newest version is 1.7.1 and fixes the cluster status check (-t cluster). Prior to 1.7.1, the check_rancher2 plugin would only look for the different component statuses of a cluster. However if the whole cluster is in an error state, the component statuses cannot be retrieved from the cluster.
This resulted in a false indication, that all clusters are OK even though when at least one cluster is in an error state:
$ ./check_rancher2.sh -H rancher.example.com -U token-xxxxx -P secret -t cluster
CHECK_RANCHER2 OK - All clusters (20) are healthy|'clusters_total'=20;;;; 'clusters_errors'=0;;;;
Now with the fix in 1.7.1 at hand, the cluster's state is parsed correctly and immediately alerts:
$ ./check_rancher2.sh -H rancher.example.com -U token-xxxxx -P secret -t cluster
CHECK_RANCHER2 CRITICAL - cluster cluster-33 is in error state -|'clusters_total'=20;;;; 'clusters_errors'=1;;;;
As this bugfix is considered important, everyone should update to the newest version of check_rancher2.
More details can be found in issue #26 on GitHub. Thanks a lot to user txj-xyz for reporting the issue.
No comments yet.