ClusterD
Apache Mesos®, actively continued

Cluster infrastructure,
continued.

ClusterD carries the proven Apache Mesos® model forward—efficient resource isolation and sharing across distributed frameworks, built for modern Linux systems.

Support ClusterD Explore the docs View on GitHub
ClusterD architecture: framework schedulers exchange resource offers and task descriptions with the ClusterD master, which launches tasks through executors on agents
01Mesos® lineage
02Shared resource pool
03Framework-driven
04Open source
Start building

Launch your first ClusterD cluster.

Follow the focused installation guide for a minimal Debian-based setup.

Open Quickstart
Why ClusterD

A durable foundation for distributed systems.

Some infrastructure ideas outlast the trend cycle. ClusterD preserves the Mesos® two-level scheduling model while maintaining the system for today's Linux and container environments.

01 / RESOURCE MODEL

One pool. Many frameworks.

Share CPUs, memory, ports, and other resources across distributed applications without forcing every workload into a single scheduler.

02 / ISOLATION

Modern Linux controls.

cgroups v2 is the required resource-control foundation from ClusterD 1.11.0-0.6.0 onward.

03 / CONTAINERS

Choose the execution path.

Run workloads through the native Mesos® containerizer or integrate with the Docker containerizer.

04 / OPERATIONS

Designed to be operated.

High-availability master setups, maintenance primitives, quotas, reservations, and observability are documented parts of the platform.

05 / OPEN SOURCE

Infrastructure you can inspect.

Build with CMake, enable SSL support, study the implementation, and contribute under the Apache License 2.0.

$ cmake -DUNBUNDLED_LIBEVENT=true \
  -DENABLE_SSL=true -DENABLE_LIBEVENT=true ../
$ make
06 / EXTENSIBILITY

Build the scheduler you need.

Scheduler and executor APIs let teams implement framework-specific placement and lifecycle logic rather than forcing every workload through one orchestration model.

Operate from the terminal

See the ClusterD CLI in action.

List agents, inspect running tasks, and retrieve cluster state directly from the command line.

View ClusterD CLI on GitHub
A short ClusterD CLI walkthrough using synthetic agent, framework, and task data.
Text alternative for the CLI demonstration
  1. ./clusterd-cli agent list lists the active agent synthetic-agent-001 at node-01.example.test.
  2. ./clusterd-cli framework list --all lists three active synthetic frameworks and one archived synthetic framework.
  3. ./clusterd-cli task list --all shows synthetic-task-running in TASK_RUNNING state and synthetic-task-finished in TASK_FINISHED state.
  4. ./clusterd-cli compose list synthetic-compose shows the demo:web:frontend task running on node-01.example.test.
  5. ./clusterd-cli m3s status synthetic-m3s --m3s --kubernetes reports a healthy m3s cluster with one server and two agents, followed by a ready Kubernetes cluster with three nodes.
  6. ./clusterd-cli task inspect synthetic-task-running returns JSON identifying its synthetic executor, framework, agent, and TASK_RUNNING status.
ClusterD WebUI

Explore your cluster from the browser.

The new React-based WebUI brings the cluster overview, tasks, frameworks, agents, resources, and master details together in a modern interface with light and dark themes.

View ClusterD WebUI on GitHub
Animated tour of the ClusterD WebUI showing the cluster overview, task and framework details, agents, resources, master information, and light and dark themes
A tour of the new ClusterD WebUI and its light and dark themes.
Architecture

Coordination without a monolith.

ClusterD separates resource coordination from application scheduling. Masters offer resources; framework schedulers decide what to run; agents execute the work.

Understand the architecture
  1. 01

    Coordinate

    Masters track agents and make available resources visible to registered frameworks.

  2. 02

    Schedule

    Each framework applies its own workload-specific scheduling logic.

  3. 03

    Execute

    Agents launch and isolate tasks across the dynamically shared node pool.

ClusterD mark
“Good infrastructure should not disappear just because the industry moves on to its next idea.”

ClusterD exists to keep the Apache Mesos® architecture maintained, secure, and useful for the people who depend on it.

Why ClusterD

Why ClusterD?

Explore how the shared resource pool and framework-specific scheduling model give microservice environments room to use the platform that best fits each workload.

Why ClusterD?
ClusterD architecture with framework schedulers, a master, and agents
Professional services

Expert help for production ClusterD.

AVENTER provides professional ClusterD support as well as architecture and software development for integrations, extensions, and project-specific requirements.

Professional support Software development
Start building

Put the cluster back in your hands.

Read the operational documentation, install from the package repository, or join the project on Matrix.

Get ClusterDJoin Matrix