Support Policies for RHEL High Availability Clusters - Cluster Stacks and Resource Managers

Updated -

Contents

Overview

Applicable Environments

  • Red Hat Enterprise Linux (RHEL) with the High Availability Add-On

Useful References and Guides

Introduction

This guide offers Red Hat's policies and requirements in relation to the component stacks and/or resource-managers that can be used in a RHEL High Availability cluster. Users of the RHEL High Availability software should adhere to these policies in order to be eligible for assistance from Red Hat with valid support subscriptions.

Policies

Multiple cluster stacks are unsupported: Running multiple cluster stack solutions (ex. pacemaker and HP ServiceGuard) on the same set of hosts (or cluster nodes) is unsupported.

RHEL 7 and 8 supported cluster stack and resource-manager: Red Hat only provides and supports one configuration of cluster components to make up the cluster "stack" in these releases of RHEL:

  • corosync provides membership, quorum, and member-communication functionality
  • pacemaker provides high-availability resource-management, STONITH, remote-node management, and other functionality
  • resource-agents provides agents that enable the pacemaker resource-manager to interact with various specific types of software and resources in a cluster.
  • fence-agents provides agents that enable the pacemaker stonith management daemon to manage cluster-member states or access to shared resources
  • Other components may be involved in providing additional functionality in conjunction with this overall stack

RHEL 6 supported cluster stacks and resource-managers: Red Hat offers two different configurations of cluster components in RHEL 6 to make up the cluster "stack", and organizations can choose either one to deploy based on their needs. These options are described in the next two sections, along with their conditions and requirements.

Red Hat introduced the pacemaker-based configuration several releases into RHEL 6 for the purpose of adding further high-availability capabilities for users. Red Hat has maintained the previously-available cman+rgmanager stack for the remaining life of RHEL 6, so existing users would have the option to keep using it until they are ready to migrate to the alternate configuration or new release of the product.

RHEL and RHEL High Availability subscriptions offer access and support for both cluster stack configurations in RHEL 6, in accordance with the RHEL Lifecycle and Red Hat Scope of Coverage:

RHEL 6 cluster-stack option #1: cman+pacemaker: Red Hat offers support for the following cluster stack configuration in RHEL 6, subject to the conditions described further below:

  • corosync provides membership and member-communication functionality
  • cman provides quorum; an API layer for membership and communication functions for other components; and the Distributed Lock Manager (DLM) for cluster-locking and RHEL Resilient Storage
  • pacemaker provides high-availability resource-management, STONITH, remote-node management, and other functionality
  • resource-agents provides agents that enable the pacemaker resource-manager to interact with various specific types of software and resources in a cluster.
  • fence-agents provides agents that enable the pacemaker stonith management daemon to manage cluster-member states or access to shared resources.
  • Other components may be involved in providing additional functionality in conjunction with this overall stack

The conditions for use of this configuration are:

  • RHEL 6 High Availability Update 5 or later packages: The cluster members must use no earlier than pacemaker-1.1.10-14.el6, pcs-0.9.90-2.el6, cman/clusterlib 3.0.12.1-59.el6, resource-agents-3.9.2-40.el6, libqb-0.16.0-2.el6, and selinux-policy-3.7.19-226.el6
  • Mixed use of conga cluster administration (luci and ricci components) in pacemaker clusters is not supported. pcs is the supported tool for administration of pacemaker clusters.

RHEL 6 cluster-stack option #2: cman+rgmanager Red Hat has offered a combination of cman and rgmanager for high availability management since the initial release of RHEL 6. This configuration consists of:

  • corosync provides membership and member-communication functionality
  • cman provides quorum; an API layer for membership and communication functions for other components; STONITH; and the Distributed Lock Manager (DLM) for cluster-locking and RHEL Resilient Storage
  • rgmanager provides high-availability resource-management
  • resource-agents provides agents that enable the rgmanager resource-manager to interact with various specific types of software and resources in a cluster.
  • fence-agents provides agents that enable the cman fenced STONITH-management daemon to manage cluster-member states or access to shared resources.
  • Other components may be involved in providing additional functionality in conjunction with this overall stack

Comments