How to Build a Two-Node Nginx HA Cluster on CentOS 7 with Corosync & Pacemaker

Outline

Preface

Experiment Introduction

Experiment Topology

Experiment Environment

Experiment Steps

Conclusion

Preface

These days I am studying high‑availability clusters and will share several experiments; this series will contain about five articles.

Experiment Introduction

This experiment is simple: implement two‑node Nginx high availability on CentOS 7 using corosync + pacemaker.

Experiment Topology

Experiment Environment

Host        IP        Purpose node1.anyisalin.com  172.16.1.2  Web service, HA node node2.anyisalin.com  172.16.1.3  Web service, HA node nfs.anyisalin.com  172.16.1.4  Provides NFS service for website files

Note: SELinux and iptables are disabled on all hosts in this experiment.

Experiment Steps

Preparation

All HA nodes must trust each other, have consistent hostname resolution, and synchronized time.

Configure hosts file synchronization

Configure trust

Time synchronization

Install HA Cluster Components

Since RHEL 6.4, the pcs tool can manage the full lifecycle of a cluster. First install pcs on both nodes and generate configuration files.

[root@node1 ~]# yum install pcs -y # node1 install pcs
[root@node1 ~]# ssh node2.anyisalin.com 'yum install pcs -y'  # node2 install pcs
[root@node1 ~]# echo passwd | passwd --stdin hacluster # set hacluster password
[root@node1 ~]# ssh node2.anyisalin.com "echo passwd | passwd --stdin hacluster" # set password on node2
[root@node1 ~]# systemctl start pcsd ; ssh node2.anyisalin.com "systemctl start pcsd" # start pcsd

[root@node1 ~]# pcs cluster auth node1.anyisalin.com node2.anyisalin.com -u hacluster # authenticate
Password:
node1.anyisalin.com: Authorized
node2.anyisalin.com: Authorized

[root@node1 ~]# pcs cluster setup --name mycluster node1.anyisalin.com node2.anyisalin.com # create cluster
[root@node1 ~]# pcs cluster start --all # start cluster

Install Nginx and Configure NFS

# Install nginx
[root@node1 ~]# yum install nginx -y       # note: requires epel repo
[root@node1 ~]# ssh node2.anyisalin.com "yum install nginx -y"
[root@node1 ~]# systemctl enable nginx   # enable autostart
[root@node1 ~]# ssh node2.anyisalin.com "systemctl enable nginx"

# Configure NFS
[root@nfs ~]# mkdir /www/htdocs -pv
[root@nfs ~]# echo "<h1>This is NFS on 172.16.1.4</h1>" > /www/htdocs/index.html
[root@nfs ~]# vim /etc/exports
/www/htdocs    172.16.1.0/24(ro)

[root@nfs ~]# systemctl start rpcbind.service
[root@nfs ~]# systemctl start nfs-server.service

# Test NFS
[root@node1 ~]# mount -t nfs 172.16.1.4:/www/htdocs /usr/share/nginx/html/
[root@node1 ~]# df
Filesystem               1K-blocks    Used Available Use% Mounted on
172.16.1.4:/www/htdocs   52403200 1134336 51268864   3% /usr/share/nginx/html
[root@node1 ~]# umount /usr/share/nginx/html/

Configure Cluster Resources with crmsh

Although pcs can install the cluster, configuring resources with pcs is cumbersome; crmsh offers a more convenient syntax.

# Install crmsh
[root@node1 ~]# yum install crmsh-2.1.4-1.1.x86_64.rpm

# Configure resources
[root@node1 ~]# crm configure
crm(live)configure# primitive webip ocf:heartbeat:IPaddr params ip=172.16.1.8
crm(live)configure# primitive nginx systemd:nginx

crm(live)configure# primitive webstore ocf:heartbeat:Filesystem params device="172.16.1.4:/www/htdocs" directory="/usr/share/nginx/html" fstype="nfs" op start timeout=60s op stop timeout=60s op monitor interval=20s timeout=40s

# Define resource group
crm(live)configure# group webservice webip webstore nginx
crm(live)configure# verify
crm(live)configure# property stonith-enabled=false
crm(live)configure# commit

Testing

Conclusion

This article demonstrates a simple HA cluster implementation; the key is understanding HA principles, which are covered in a separate blog post.