Monitor CTDB cluster health in on active/active Samba/CTDB nodes.
An agent plugin runs locally on each node in the CTDB cluster, reading ctdb status and ctdb ip all, and reports both this node's CTDB-reported health and the cluster's shared state - recovery mode and public IP allocation.

What it monitors

• Node health - CTDB-reported state (OK / PARTIALLYONLINE / DISABLED / STOPPED / UNHEALTHY / DISCONNECTED / BANNED), distinct from whether ctdbd is merely running
• Recovery mode - cluster-wide NORMAL vs RECOVERY, with a grace period so a normal brief failover doesn't alert while a stuck recovery does
• Public IP allocation - orphaned IPs with no owning node, and uneven distribution across the cluster
• Recovery duration - graphed metric for spotting recoveries trending longer over time

Highlights

• Reads local CTDB state only
• Every severity independently configurable per state
• Intentional per-node redundancy - each node reports the full cluster picture
• Deployed and configured via the Checkmk Agent Bakery

Requirements

• A Linux CTDB cluster with the ctdb CLI installed
• Checkmk 2.3.0p1+
• No credentials required

Full documentation and configuration guide: https://github.com/turretsec/cmk-ctdb