You are on page 1of 21

MONITORING DEEP DIVE

Using Elastic Stack in production


X-Pack Monitoring

Monitor Elasticsearch
• Real-time statistics and metrics for all clusters and nodes
Diagnose Issues
• Analyze historical or real-time data for root cause analyses
Optimize Performance
• Utilize in-depth analyses to improve cluster performance
Monitoring Licenses

Basic
• Free license
• Monitor the Elastic Stack – cluster, node, and indices in Kibana
• Monitoring of Kibana
• Real-time statistics and metrics for all clusters and nodes

Gold and Platinum


• Multicluster support to compare health and performance of multiple clusters
• Analyze historical or real-time data for root cause analyses
• Utilize analyses to proactively optimize and improve cluster performance
• Configure data retention policy
Advantages

• The official Elasticsearch monitoring tool


‒ Created by Elasticsearch developers
‒ Shipped together with Elastic Stack release

• Realtime and historical analysis


‒ Store metric data as Elasticsearch indices

• All-in-one cluster health management


‒ Indices: search and indexing rate, latency
‒ Nodes: CPU, JVM, load average and free disk space

• Also available in Elastic Cloud

4
So what does it tell me - Index

• Overview • Advanced
▪ Search Rate (/s) ▪ Index Memory – Lucene
▪ Search Latency (ms) ▪ Index Memory – Elasticsearch
▪ Indexing Rate (/s) ▪ Request Rate
▪ Indexing Latency (ms) ▪ Request Time (ms)
▪ Segment Count ▪ Refresh Time (ms)
▪ Document Count ▪ Throttle Time (ms)
▪ Shard Location / Legend ▪ Disk (MB)
▪ Segment Count

5
So what does it tell me - Nodes

• Overview • Advanced
▪ JVM Heap (GB) ▪ GC Count
▪ Index Memory (KB) ▪ GC Duration (ms)
▪ CPU Utilization (%) ▪ JVM Heap (GB) – Max and Used
▪ System Load ▪ CPU Utilisation (%)
▪ Latency (ms) ▪ Index Memory – Lucene
▪ Segment Count ▪ Index Memory – Elasticsearch
▪ Indexing Time (ms)
▪ Request Rate
▪ Indexing Threads
▪ Read Threads

6
UI Overview

7
Monitoring is now for Kibana as well as Elasticsearch

8
Switch to
Monitoring in
the new
navigation bar
– shows list of
clusters being
monitored

9
Star denotes
Master node

High-level overview of Node stats

10
Node
Details

11
Index
Details

12
13
New Kibana
Node
Monitoring

14
Monitoring Demo

15
Deployment
Single Cluster

Kibana ES node ES node ES node


monitoring-UI monitoring-agent monitoring-agent monitoring-agent

Elasticsearch Cluster

17
Separate Monitoring Cluster

ES node ES node ES node


monitoring-agent monitoring-agent monitoring-agent

Cluster

Kibana ES node
monitoring-UI monitoring-agent

Monitoring Cluster

18
Integrations
• Alerting
• Security
Alerting

• Alert when:
• Cluster health changes
• High memory usage
• High cpu usage
• High file descriptor usage
• Nodes joining or leaving the cluster

20
Security

• Password protection for Kibana access


• Role-based access control (e.g. cluster
monitoring only users)
• Secure communication between the
marvel-agent and the monitoring cluster
• Monitoring of Shield Audit Indices

21

You might also like