zookeeper

ZooKeeper >=3.6.0 has built-in Prometheus support, but the bundled dashboard queries the zk_* names emitted by Categraf's ZooKeeper input. Use this input for the current dashboard. If native ZooKeeper metric names differ, update the dashboard instead of only swapping the endpoint.

Overview

The Categraf input uses ZooKeeper Four Letter Words commands. The real-data test used ZooKeeper 3.9 and this input and passed all dashboard queries.

Note that zookeeper v3.4.10 and later added a whitelist for four-letter commands, so you need to add the whitelist setting to the zookeeper configuration file zoo.cfg:

4lw.commands.whitelist=mntr,ruok

In production, allow only the required commands instead of using *.

Configuration

The zookeeper plugin configuration lives in conf/input.zookeeper/zookeeper.toml. Separate the addresses of multiple instances in a cluster with spaces:

[[instances]]
cluster_name = "dev-zk-cluster"
addresses = "127.0.0.1:2181"
timeout = 10
labels = { job = "zookeeper", instance = "zk-01:2181" }

To monitor multiple zookeeper clusters, just add more instances:

[[instances]]
cluster_name = "dev-zk-cluster"
addresses = "127.0.0.1:2181"
timeout = 10
labels = { job = "zookeeper-a", instance = "zk-a-01:2181" }

[[instances]]
cluster_name = "test-zk-cluster"
addresses = "127.0.0.1:2181 127.0.0.1:2182 127.0.0.1:2183"
timeout = 10
labels = { job = "zookeeper-b" }

Run ./categraf --test --inputs zookeeper and confirm zk_up, zk_znode_count, and zk_packets_sent. Preserve the job and instance labels used by the dashboard variables.