Andrew Mercer
on this page
    • Enable the rancher-monitoring add-on: Grafana dashboards for VMs (per-VM CPU/memory/disk/network from KubeVirt metrics), Longhorn, nodes, etcd.
    • Key alerts to configure:
    • etcd fsync latency and leader changes.
    • Longhorn volumes degraded/faulted, replica rebuild counts, disk usage > 75%.
    • Node memory pressure (VM OOM risk).
    • VM migration failures.
    • Backup job failures.
    • Certificate expiry.
    • Forward logs with rancher-logging (Loki, Elasticsearch, syslog, Splunk, S3).
    • Export events — many Harvester problems announce themselves as Kubernetes events first.
    • Support bundles (Support → Generate Support Bundle) collect logs/state across the cluster; invaluable for support tickets and self-diagnosis.

    ← Multi-tenancy and RBAC · Index · Upgrades and lifecycle →