Master Elasticsearch Node Commands: Inspect, Monitor, and Troubleshoot Your Cluster

Node‑related Commands

1. View basic node information

The GET /_nodes/<node_id> API returns all details of a node, including process, JVM, plugins, roles, attributes, and settings. GET /_nodes/<node_id> To fetch only specific sections, append the sub‑path, e.g., GET /_nodes/<node_id>/process: GET /_nodes/<node_id>/process Sample response:

{
  "_nodes" : {"total" : 1,"successful" : 1,"failed" : 0},
  "cluster_name" : "es-xxxx",
  "nodes" : {
    "fdaOV16OQPq6-AUVihmq4A" : {
      "name" : "162680s31430001456s32",
      "transport_address" : "xx.0.96.22:9300",
      "host" : "xx.0.96.22",
      "ip" : "xx.0.96.22",
      "version" : "7.10.1",
      "roles" : ["master","ml","remote_cluster_client"],
      "attributes" : {
        "ml.machine_memory" : "16478932992",
        "rack" : "cvm_4_200003",
        "xpack.installed" : "true",
        "set" : "200003",
        "transform.node" : "false",
        "ip" : "xx.20.58.221",
        "temperature" : "hot",
        "ml.max_open_jobs" : "20",
        "region" : "4"
      },
      "process" : {"refresh_interval_in_millis" : 1000,"id" : 24918,"mlockall" : false}
    }
  }
}

2. View node statistics

The GET /_nodes/stats API provides comprehensive metrics such as JVM, CPU, memory, disk usage, index counts, request counts, cache statistics, segment and translog data.

GET /_nodes/stats
GET /_nodes/<node_id>/stats
GET /_nodes/stats/<metric>

Example: retrieve only JVM metrics for a specific node. GET /_nodes/1626803143000145632/stats/jvm Sample response (truncated):

{
  "_nodes" : {"total" : 1,"successful" : 1,"failed" : 0},
  "cluster_name" : "es-xxx",
  "nodes" : {
    "fdaOV16OQPq6-AUVihmq4A" : {
      "timestamp" : 1639878427713,
      "jvm" : {
        "uptime_in_millis" : 13037976328,
        "mem" : {
          "heap_used_in_bytes" : 214161760,
          "heap_used_percent" : 2,
          "heap_committed_in_bytes" : 8555069440,
          "heap_max_in_bytes" : 8555069440
        },
        "threads" : {"count" : 41,"peak_count" : 43},
        "gc" : {"collectors" : {"young" : {"collection_count" : 4718,"collection_time_in_millis" : 162825},"old" : {"collection_count" : 3,"collection_time_in_millis" : 359}}}
      }
    }
  }
}

Other useful stats include merge information, segment & translog details, and store size:

GET /_nodes/1626803143000145632/stats/indices/merge
GET /_nodes/1626803143000145632/stats/indices/segments,translog
GET /_nodes/stats/indices/store

For a quick overview of node memory usage, the cat API is handy:

GET _cat/nodes
GET _cat/nodes?h=name,segments.memory,segments.index_writer_memory,heap.percent,fielddata.memory_size,query_cache.memory_size,request_cache.memory_size\&v

3. View thread‑pool usage

Use the cat thread‑pool API to see active, queued, and rejected tasks. High values in queue or rejected indicate that the node’s read/write capacity is saturated, often caused by high CPU usage.

GET /_cat/thread_pool
GET /_cat/thread_pool/search,write?v

Sample response:

node_name           name   active queue rejected
1626803143000145832 search      0     0        0
1626803143000145832 write       0     0        0
...

Typical symptoms such as query rejections are illustrated below:

When the thread‑pool queue is full, the logs look similar to the following screenshot:

4. View hot threads

The hot‑thread API helps pinpoint CPU‑intensive tasks on a node.

GET /_nodes/hot_threads
GET /_nodes/<node_id>/hot_threads

Sample output (truncated):

::: {1626803143000145832}{...}
   Hot threads at 2021-12-19T02:21:12.334Z, interval=500ms, busiestThreads=3, ignoreIdleThreads=true:
   ...

Analyzing hot threads together with CPU metrics helps identify root causes such as frequent merges, heavy text analysis, or snapshot operations.

Command Summary Table