Elastic Stack on Docker

Preconfigured Security, Tools, and Self-Monitoring

Configured to be ready to be used for Log, Metrics, APM, Alerting, Machine Learning, and Security (SIEM) usecases.

Introduction

Elastic Stack (ELK) Docker Composition, preconfigured with Security, Monitoring, and Tools; Up with a Single Command.

Suitable for Demoing, MVPs and small production deployments.

Stack Version: 9.2.3 🎉 - Based on Official Elastic Docker Images

You can change Elastic Stack version by setting ELK_VERSION in .env file and rebuild your images. Any version >= 9.0.0 is compatible with this template.

⚠️ Upgrading from 8.x? See the Upgrade Notes section below for breaking changes and migration steps.

---

Main Features 📜

• Configured as a Production Single Node Cluster. (With a multi-node cluster option for experimenting).
• Security Enabled By Default.
• Configured to Enable:
  • Logging & Metrics Ingestion
    • Option to collect logs of all Docker Containers running on the host. via make collect-docker-logs.
  • APM
  • Alerting
  • Machine Learning
  • Anomaly Detection
  • SIEM (Security information and event management).
  • Enabling Trial License
• Use Docker Compose and .env to configure your entire stack parameters.
• Persist Elasticsearch's Keystore and SSL Certifications.
• Self-Monitoring Metrics Enabled (using Metricbeat for ES 9+).
• Prometheus Exporters for Stack Metrics.
• Embedded Container Healthchecks for Stack Images.

More points

And comparing Elastdocker and the popular deviantony/docker-elk

Expand...

One of the most popular ELK on Docker repositories is the awesome deviantony/docker-elk. Elastdocker differs from deviantony/docker-elk in the following points.

• Security enabled by default using Basic license, not Trial.

• Persisting data by default in a volume.

• Run in Production Mode (by enabling SSL on Transport Layer, and add initial master node settings).

• Persisting Generated Keystore, and create an extendable script that makes it easier to recreate it every-time the container is created.

• Parameterize credentials in .env instead of hardcoding elastich:changeme in every component config.

• Parameterize all other Config like Heap Size.

• Add recommended environment configurations as Ulimits and Swap disable to Docker Compose.

• Make it ready to be extended into a multinode cluster.

• Configuring the Self-Monitoring and the Filebeat agent that ship ELK logs to ELK itself. (as a step to shipping it to a monitoring cluster in the future).

• Configured Prometheus Exporters.

• The Makefile that simplifies everything into some simple commands.

Automatic Docker Container Log Collection

Collect logs from all Docker containers on your host with a single command:

make collect-docker-logs

Filebeat automatically discovers containers, parses logs, and ships them to Elasticsearch. View and analyze everything in Kibana with zero configuration.

---

Requirements

• Docker 20.05 or higher with Docker Compose v2
• 4GB RAM (For Windows and MacOS make sure Docker's VM has more than 4GB+ memory.)

Setup

1. Clone the Repository

   git clone https://github.com/sherifabdlnaby/elastdocker.git

2. Initialize Elasticsearch Keystore and TLS Self-Signed Certificates

   $ make setup

   For Linux's docker hosts only. By default virtual memory is not enough so run the next command as root sysctl -w vm.max_map_count=262144

3. Start Elastic Stack

   $ make elk           <OR>         $ docker compose up -d

4. Visit Kibana at https://localhost:5601 or https://<your_public_ip>:5601

   Default Username: elastic, Password: changeme

   • Notice that Kibana is configured to use HTTPS, so you'll need to write https:// before localhost:5601 in the browser.
   • Modify .env file for your needs, most importantly ELASTIC_PASSWORD that setup your superuser elastic's password, ELASTICSEARCH_HEAP & LOGSTASH_HEAP for Elasticsearch & Logstash Heap Size.

Whatever your Host (e.g AWS EC2, Azure, DigitalOcean, or on-premise server), once you expose your host to the network, ELK component will be accessible on their respective ports. Since the enabled TLS uses a self-signed certificate, it is recommended to SSL-Terminate public traffic using your signed certificates.

🏃🏻‍♂️ To start ingesting logs, you can start by running make collect-docker-logs which will collect your host's container logs.

Additional Commands

Expand

To Start Monitoring and Prometheus Exporters

$ make monitoring

To Ship Docker Container Logs to ELK

$ make collect-docker-logs

To Start Elastic Stack, Tools and Monitoring

$ make all

To Start 2 Extra Elasticsearch nodes (recommended for experimenting only)

$ make nodes

To Rebuild Images

$ make build

Bring down the stack.

$ make down

Reset everything, Remove all containers, and delete DATA!

$ make prune

Configuration

• Some Configuration are parameterized in the .env file.
  • ELASTIC_PASSWORD, user elastic's password (default: changeme pls).
  • ELK_VERSION Elastic Stack Version (default: 9.2.3)
  • ELASTICSEARCH_HEAP, how much Elasticsearch allocate from memory (default: 1GB -good for development only-)
  • LOGSTASH_HEAP, how much Logstash allocate from memory.
  • Other configurations which their such as cluster name, and node name, etc.
• Elasticsearch Configuration in elasticsearch.yml at ./elasticsearch/config.
• Logstash Configuration in logstash.yml at ./logstash/config/logstash.yml.
• Logstash Pipeline in main.conf at ./logstash/pipeline/main.conf.
• Kibana Configuration in kibana.yml at ./kibana/config.
• Metricbeat Configuration in metricbeat.yml at ./metricbeat/config (for Stack Monitoring in ES 9+).

Setting Up Keystore

You can extend the Keystore generation script by adding keys to ./setup/keystore.sh script. (e.g Add S3 Snapshot Repository Credentials)

To Re-generate Keystore:

make keystore

Notes

• ⚠️ Elasticsearch HTTP layer is using SSL, thus mean you need to configure your elasticsearch clients with the CA in secrets/certs/ca/ca.crt, or configure client to ignore SSL Certificate Verification (e.g --insecure in curl).

• Adding Two Extra Nodes to the cluster will make the cluster depending on them and won't start without them again.

• Makefile is a wrapper around Docker Compose commands, use make help to know every command.

• Elasticsearch will save its data to a volume named elasticsearch-data

• Elasticsearch Keystore (that contains passwords and credentials) and SSL Certificate are generated in the ./secrets directory by the setup command.

• Make sure to run make setup if you changed ELASTIC_PASSWORD and to restart the stack afterwards.

• For Linux Users it's recommended to set the following configuration (run as root)

  sysctl -w vm.max_map_count=262144
  

  By default, Virtual Memory is not enough.

---

Working with Elastic APM

After completing the setup step, you will notice a container named apm-server which gives you deeper visibility into your applications and can help you to identify and resolve root cause issues with correlated traces, logs, and metrics.

Authenticating with Elastic APM

In order to authenticate with Elastic APM, you will need the following:

• The value of ELASTIC_APM_SECRET_TOKEN defined in .env file as we have secret token enabled by default
• The ability to reach port 8200
• Install elastic apm client in your application e.g. for NodeJS based applications you need to install elastic-apm-node
• Import the package in your application and call the start function, In case of NodeJS based application you can do the following:

const apm = require('elastic-apm-node').start({
  serviceName: 'foobar',
  secretToken: process.env.ELASTIC_APM_SECRET_TOKEN,
  
  // https is enabled by default as per elastdocker configuration
  serverUrl: 'https://localhost:8200',
})

Make sure that the agent is started before you require any other modules in your Node.js application - i.e. before express, http, etc. as mentioned in Elastic APM Agent - NodeJS initialization

For more details or other languages you can check the following:

• APM Agents in different languages

Monitoring The Cluster

Via Stack Monitoring (Metricbeat)

Elasticsearch 9+ uses Metricbeat for Stack Monitoring (the recommended approach). When you start monitoring with make monitoring, Metricbeat will collect metrics from all stack components and send them to Elasticsearch.

Head to Stack Monitoring tab in Kibana to see cluster metrics for all stack components.

Architecture Change in ES 9:

• ES 8.x and earlier: Used internal xpack.monitoring for self-monitoring
• ES 9.x: Uses external Metricbeat collection (more scalable and reliable)

In Production, cluster metrics should be shipped to another dedicated monitoring cluster.

Via Prometheus Exporters

If you started Prometheus Exporters using make monitoring command. Prometheus Exporters will expose metrics at the following ports.

Prometheus Exporter	Port	Recommended Grafana Dashboard
elasticsearch-exporter	9114	Elasticsearch by Kristian Jensen
logstash-exporter	9304	logstash-monitoring by dpavlos

Note: Elasticsearch Exporter uses updated flags for ES 9 compatibility (--es.indices instead of deprecated --collector.indices).

---

Upgrade Notes from 8.x to 9.x

Expand to see breaking changes and migration details...

Elasticsearch 9 introduced several breaking changes. This section documents the changes made to ElastDocker for ES 9 compatibility.

Breaking Changes Fixed

1. Logstash Configuration Changes

File: logstash/config/logstash.yml

• http.host → api.http.host

File: logstash/pipeline/main.conf

• ssl → ssl_enabled
• ssl_certificate_verification → ssl_verification_mode
• cacert → ssl_certificate_authorities

2. Monitoring Architecture Change

Before (ES 8.x):

• Used internal xpack.monitoring.collection.enabled setting
• Components self-reported metrics

After (ES 9.x):

• Uses external Metricbeat for metric collection
• More scalable and follows Elastic's recommended approach
• New component: metricbeat service in docker-compose.monitor.yml

Files Modified:

• elasticsearch/config/elasticsearch.yml - Removed xpack.monitoring.collection.enabled
• logstash/config/logstash.yml - Removed xpack.monitoring settings
• apm-server/config/apm-server.yml - Removed monitoring section
• metricbeat/config/metricbeat.yml - NEW FILE for Stack Monitoring

3. Filebeat Migration to Filestream Input

The container input type is deprecated in Filebeat 9. Migrated to the modern filestream input with container parser - the ES 9+ recommended approach.

Files Modified:

• filebeat/filebeat.docker.logs.yml - Now uses type: filestream with container parser
• filebeat/filebeat.monitoring.yml - All module inputs migrated to filestream

Key Changes:

• type: container → type: filestream with unique IDs
• Added parsers.container configuration for Docker log parsing
• Added prospector.scanner.symlinks: true for Docker log paths
• No deprecation warnings - fully ES 9 compliant

4. Certificate Generation Script

File: setup/setup-certs.sh

• Updated password generation to work without openssl command (not available in ES 9 containers)
• Now uses /dev/urandom for random password generation

5. Elasticsearch Exporter Flags

File: docker-compose.monitor.yml

• Updated exporter flags for compatibility with exporter v1.10.0+
• --collector.indices → --es.indices

Known Deprecation Warnings

The following deprecation warnings are expected and originate from upstream Elastic components. They will be resolved in future component releases:

1. Beats using ?local parameter (CRITICAL) - ~446 occurrences

   • Source: Metricbeat
   • Will be fixed in future Beats releases
   • Note: Filebeat no longer generates these warnings after migrating to filestream input

2. Behavioral Analytics deprecated (WARN) - ~37 occurrences

   • Source: Kibana cleanup process
   • Expected during ES 9 migration
   • Will resolve once cleanup completes

3. APM System Index Access (WARN) - ~13 occurrences

   • Source: APM Server
   • Will be fixed in future APM Server releases

These warnings don't affect functionality and are logged to the deprecation data stream for visibility.

Upgrade Path

Important: You must upgrade to Elasticsearch 8.19.x before upgrading to 9.x.

Recommended Path:

8.17.0 → 8.19.x (run Upgrade Assistant) → 9.x

For a clean installation on ES 9, simply:

1. Set ELK_VERSION=9.2.3 in .env
2. Run make setup
3. Run make elk (or make all for full stack with monitoring)

---

License

MIT License Copyright (c) 2022-2026 Sherif Abdel-Naby

Contribution

PR(s) are Open and Welcomed.