Skip to content

Latest commit

 

History

History
106 lines (74 loc) · 3.13 KB

README.md

File metadata and controls

106 lines (74 loc) · 3.13 KB

Ceph Cluster Checkmk Plugin

Checkmk agent plugin for monitoring Ceph clusters.

Features

Provides 7 services from a single agent plugin:

Service Monitors Thresholds
Ceph Health Overall HEALTH_OK/WARN/ERR with detail per check State-based
Ceph OSD Status OSDs up/in, alerts on down/out State-based
Ceph OSD Latency Worst-case commit/apply latency, per-OSD graphs Configurable (default 50/100 ms)
Ceph Scrub Overdue scrub/deep-scrub PG counts, scrub activity Configurable (default 10/50 PGs)
Ceph Pool Usage Cluster capacity %, per-pool usage graphs Configurable (default 80/90%)
Ceph PG Status PG state distribution, non-clean PG alerting State-based
Ceph I/O Client read/write throughput and IOPS (bidirectional graphs) Info only

Per-instance metrics (per-OSD latency, per-pool usage) are yielded dynamically and auto-graphed by Checkmk. Aggregate graphs are provided for cluster I/O, scrub backlog, and capacity.

Requirements

  • Checkmk 2.3.0p1 or later
  • Agent host must have ceph CLI access (typically a mon node)

Installation

MKP Package (recommended)

Download the latest .mkp file from the Releases page and install it:

mkp install oposs_ceph-<version>.mkp

Manual Installation

Copy the plugin files into your Checkmk site:

local/lib/python3/cmk_addons/plugins/oposs_ceph/
├── agent_based/
│   └── oposs_ceph.py
├── graphing/
│   └── oposs_ceph.py
└── rulesets/
    └── oposs_ceph.py

local/lib/python3/cmk/base/cee/plugins/bakery/
└── oposs_ceph.py

local/share/check_mk/agents/plugins/
└── oposs_ceph

Bakery Deployment

Enable the "Ceph Monitoring (Linux)" agent bakery rule to automatically deploy the agent plugin to hosts. The collection interval is configurable (default 60s).

Cluster Setup

Since all Ceph mon nodes report the same cluster-wide data, the plugin supports Checkmk's native cluster hosts to avoid duplicate services:

  1. Add each mon node as a regular host in Checkmk
  2. Deploy the agent plugin to all mon nodes (via bakery rule or manually)
  3. Create a cluster host in Setup > Hosts > Create cluster host
    • Add the mon nodes as cluster nodes
  4. Assign the Ceph services to the cluster in the cluster host's service configuration — use "Clustered services" rules to move the 7 Ceph services from the individual nodes to the cluster host
  5. Discover services on the cluster host

The cluster check function picks data from whichever mon node responds. If one mon goes down, monitoring continues from the remaining nodes.

Agent Data

The agent plugin runs three commands:

  • ceph status -f json — health, OSD map, PG states, I/O stats
  • ceph osd perf -f json — per-OSD commit/apply latency
  • ceph df detail -f json — cluster and per-pool capacity

Troubleshooting

Test the agent plugin directly on the mon node:

/usr/lib/check_mk_agent/plugins/oposs_ceph

Verify Ceph CLI access:

ceph status -f json

License

MIT - OETIKER+PARTNER AG