Corosync+pacemaker enable high availability of web clusters

Pacemaker: Pacemaker, is a cluster resource manager. It achieves maximum availability for Cluster service node and resource level failure detection and recovery using the messaging and membership capabilities of the preferred cluster infrastructure.

Pacemaker is a resource manager and does not provide heartbeat information.

heartbeat3.0 components after splitting:

Heartbeat: The original message communication layer is independent of the Heartbeat project, the new heartbeat is only responsible for maintaining the information of the nodes of the cluster and their previous communication.

Cluster-glue: Equivalent to an intermediate layer, which is used to correlate heartbeat and pacemaker, consisting mainly of 2 parts, namely LRM and Stonith.

Resource-agent: A collection of scripts used to control service start and stop, monitor server status

Pacemaker: Cluster resource Manager (CRM), used to manage the entire HA control center, the client through the pacemaker to configure the management to monitor the entire cluster.

Pacemaker Features

Host and application-level fault detection and recovery

Support for virtually any redundant configuration

Supports multiple cluster configuration modes at the same time

Configure policy to handle quorum loss

Support app startup/shutdown sequence

Applications that support multiple modes

Can test for any failure or cluster status of a cluster

Pacemaker internal structure

650) this.width=650; "src=" Http://clusterlabs.org/mwiki/images/thumb/9/9f/Stack.png/600px-Stack.png "alt=" 600px-stack.png "/>

Cluster Component Description

STONITHD: Heartbeat System

LRMD: The local resource management daemon. It provides a common interface for supporting resource types and directly invokes resource proxies.

Pengine: Policy engine.

CIB: The cluster information base.

CRMD: Cluster resource management daemon. The main message agent Penine and LRM, also elected a leader of the DC co-ordinated activities of the cluster

Openais:openais Message and Member layer

Heartbeat: Heartbeat message layer, an alternative to Openais

CCM: Consensus cluster member, Heartbeat member layer

Function

The CIB uses XML to represent the configuration and current state of all resources in the cluster, and the CIB content is automatically synchronized across the cluster, using Pengine to compute the ideal state of the cluster, generating a list of instructions, and then conveying it to the DC.

The DC node of all nodes in the pacemaker cluster is elected as the main decision node. If the node of the selected DC is down, it will quickly establish a new DC on all nodes. DCS pass Pengine generated policies to LRMD or CRMD on other nodes through the cluster messaging infrastructure.

The ideal strategy is recalculated when there are node outages in the cluster pengine. In some cases, it may be necessary to close the node to protect the shared data or to complete the recycling of the resources, because the pacemaker is configured with the STONITHD device. The Stonith can knock the other nodes out of the head, usually with the remote power supply disconnected. Pacemaker will configure the Stonith device as a resource to be saved in the CIB, making it easier for them to monitor resource failures or outages.

Experimental Operation CENTOS6. X

Topology diagram

650) this.width=650; "src=" Http://s3.51cto.com/wyfs02/M02/58/98/wKiom1S02QTQcxOvAAFJSxjKOk0076.jpg "alt=" Wkiom1s02qtqcxovaafjsxjkok0076.jpg "/>

Installation configuration

Yum Install Corosync pacemaker-y

This article is from the "ngames" blog, make sure to keep this source http://ngames.blog.51cto.com/3187187/1607503

Corosync+pacemaker enable high availability of web clusters