On This Page

Home / Stream/ Deploying Cribl Stream Software· Manual Deployment·Distributed Deployment

Distributed Deployment ​

Getting started with a distributed Cribl Stream deployment

To sustain higher incoming data volumes, and/or increased processing, you can scale from a single instance up to a multi-instance, distributed deployment. The Worker instances are centrally managed by a single Leader Node, which is responsible for keeping configurations in sync, and for tracking and monitoring the instances’ activity metrics.

To sustain high availability, Cribl Stream offers failover to an alternate Leader.

See common distributed deployment use cases in Worker Groups - What Are They and Why You Should Care.

As of version 3.0, Cribl Stream’s former “master” application components are renamed “leader.” While some legacy terminology remains within CLI commands/​options, configuration keys/values, and environment variables, this document will reflect that.

Concepts ​

Single Instance - a single Cribl Stream instance, running as a standalone (not distributed) installation on one server.

Leader Node - a Cribl Stream instance running in Leader mode, used to centrally author configurations and monitor Worker Nodes in a distributed deployment.

Worker Node - a Cribl Stream instance running as a managed Worker, whose configuration is fully managed by a Leader Node. (By default, will poll the Leader for configuration changes every 10 seconds.)

Worker Group - a collection of Worker Nodes that share the same configuration. You map Nodes to a Worker Group using a Mapping Ruleset.

Worker Process - a Linux process within a Single Instance, or within Worker Nodes, that handles data inputs, processing, and output. The process count is constrained by the number of physical or virtual CPUs available; for details, see Sizing and Scaling.

Mapping Ruleset - an ordered list of filters, used to map Workers Nodes into Worker Groups.

Options and Constraints ​

A Worker Node’s local running config can be manually overridden/changed, but changes won’t persist on the filesystem. To permanently modify a Worker Node’s config: Save, commit, and deploy it from the Leader. See Deploying Configurations below.

With an Enterprise license, you can configure Permissions-based (Stream 4.2 and later) or Role-based access control at the Worker Group level. Non-administrator users will then be able to access Workers only within those Worker Groups on which they’re authorized.

Aggregating Workers ​

To clarify how the above concepts add up hierarchically, let’s use a military metaphor involving toy soldiers:

  • Worker Process = soldier.
  • Worker Node = multiple Worker Processes = squad.
  • Worker Group = multiple Worker Nodes = platoon.

Multiple Worker Groups are very useful in making your configuration reflect organizational or geographic constraints. E.g., you might have a U.S. Worker Group with certain TLS certificates and output settings, versus an APAC Worker Group and an EMEA Worker Group, each with their own distinct certs and settings.

Architecture ​

This is an overview of a distributed Cribl Stream deployment’s components.

Distributed deployment architecture
Distributed deployment architecture

Here is the division of labor among components of the Leader Node and Worker Node.

Leader Node ​

  • API Process - Handles all the API interactions.

  • N Config Helpers - One process per Worker Group. Helps with maintaining configs, previews, etc.

Worker Node ​

  • API Process - Handles communication with the Leader Node (i.e., with its API Process) and handles other API requests.

  • N Worker Processes - Handle all the data processing.

Single-Instance Architecture ​

For comparison, here’s the simpler division of labor on a single-instance deployment, where the separate Leader versus Worker Nodes are essentially condensed into one stack:

  • API Process - Handles all the API interactions.

  • N Worker Processes - Handle all data processing,

  • One of the Worker Processes is called the leader Worker Process. (Not to be confused with the Leader Node.) This is responsible for writing configs to disk, in addition to data processing.

So here, the API Process handles the same responsibilities as a Leader Node’s API Process, while the Worker Processes correspond to the Worker Nodes’ Worker Processes. The exception is that one Worker Process does double duty, also filling in for one of the Leader Node’s Config Helpers.

System Requirements ​

Leader Node Requirements ​

  • OS:

    • Linux: RedHat, CentOS, Ubuntu, AWS Linux (64bit).

  • System:

    • +4 physical cores, +8 GB RAM.
    • 25 GB free disk space.

  • Git: git must be available on the Leader Node. See details below.

  • Browser Support: The five most-recent versions of Chrome, Firefox, Safari, and Microsoft Edge.

We assume that 1 physical core is equivalent to 2 virtual/hyperthreaded CPUs (vCPUs). All quantities listed above are minimum requirements.

We recommend deploying the Leader on stable, highly available infrastructure, because of its role in coordinating all Worker instances.

Worker Node Requirements ​

Port Requirements ​

During installation of any Leader or Worker instance, firewalls on that instance’s host must enable outbound communication to https://cdn.cribl.io on port 443.

Other ports must remain open continuously for a Distributed deployment to function.

If any of this traffic must go through a proxy - except for cluster communication, which can’t be proxied - see System Proxy Configuration for configuration details.

Installing on Linux ​

See Single-Instance Deployment, as the installation procedures are identical.

Version Control with git