redhat6.5 based on GUI virtualization cluster

For various reasons, can not do experiments, here is a brief description of the GUI virtualization cluster

NAT: Host to open the Ip_forward function, so as to communicate with the external network.
Enter/etc/libvirt/qemu/network/default.xml
CP Default.xml Net1.xml
Vim Net1.xml
Modified Place Name=net1,uuid,bridge NAME,MAC,IP address Real Machine IP assignment range of Ip,range virtual machine
CD Autostart directory under Ln-s. /net1.xml./
Service LIBVIRTD Restart
Enter the corresponding XML file for the virtual machine and modify the network section to

Bridging: 1. Close NetworkManager
2.CP Ifcfg-eth0 Ifcfg-br0
3.vim ifcfg-br0
Device=br0
Type=bridge
4.vim Ifcfg=eth0
Ipaddr,netmask,gateway,dns all deleted First
Added: bridge=br0
5.service Network Restart
Enter the virtual machine corresponding file, modify the network section can
* (Modifying virtual machine files is required to shut down the virtual machine first)

A GUI-based virtual cluster

SELinux and Iptables shut down, then do the next NTP time synchronization 1.yum install-y Luci Ricci (the second node only needs to install Ricci) or Yum groupinstall-y "high availablity" (two sections 2.service Luci Start (Luci is graphical management) 3.service Ricci start (two nodes with a password) 4./etc/cluster/cluster.conf (master profile) or access https:// The ip:8084 of your machine the user and password of your machine are then all configured in the GUI, the mouse dot 5.create (create cluster) cluster name---Cluster name node name section named Select Download package S Auto Pack 6.clustat (view cluster status) 7.fence devices (create two virtual fence devices associated with a real machine) fence Virt8. Click on the node name to enter the settings, add a virtual fence, named before you create a good fence (two nodes do) 9. Add failover domain (invalid domain) name restricted (tick) Two nodes tick on 10.RESOURC ES (resources) IP address x.x.x.100 (VIP) netmask 24 Add a script name httpd full path/etc/init.d/http D 11.service Groups Service name automatically start this service (tick on) failover domain recovery Policy Select an action to add Child resource VIP http (on the node to first load the package) 12 open the Web page, Access x.x.x.100 can access to a node of the Web site GUI end 13. Before failover, Do a fence verification (here is a link to see How to configure, remind a little) 1. You must do this in your host, and then copy the key file to the corresponding node path, the path must not be wrong 14. Command line Add sectionPoint 1. Modify/etc/cluster/cluster.conf version number +1 the node file below adds a new node 15.cman_tool version-r (Synchronous version number) 16. New node start three services, Cman RGM Anager modclusterd17.clusvcadm-r Web name-M New node (migrate service to new node) 18 new node pulls the fence device over 19. When you close two nodes in a row, the third node does not provide services to the outside, and this involves the weight problem 20. Arbitration disk: New machine, add a disk, partition give him 100M space yum   Install-y scsi* vim/etc/tgt/targets.conf in the 38th row around the Backing-store path to your partition, the next two lines also open, three behavior an example. Service TGTD start chkconfig tgtd on21.node1-3 install iscsi* iscsiadm-m discovery-t st-p node4 iscsiadm -M node-l mkqdisk-c/dev/your new disk-L quorum disk name 22.service Luci stop rm-f/var/lib/luci/data/luci.db Service         Luci startGUI23. Add a cluster directly, and then click the last configure Qdisk added using quorum disk by device label--quorum disk name heuristics /bin/ping host Ip-c 1 interval 10 (ping every 10 seconds) score 1 (ping plus one point) TKO 2 (ping two times ping does not pass him off. ) Minimum total score 1 (min. 1) 24. Shared storage 1. Each node goes to hook up for NFS, giving the corresponding user Rights 2. Which node provides services on the External hook NFS in the GUI, point to resources, add n Fs/ciFS Mount name--a name mount point---hook up locally? Host---NFS server IP (best in the Intranet) NFS export directory name or CIFS share (which file for NFS server?) ) 25.service groups add the resources to the NFS name 26. High availability is complete, but how do you ensure a high availability based on load balancing? Hardware load balancing exclusion, the Linux system can also do the LVS load balance, the realization of cost reduction. 27.lvs is based on kernel work, when compiling the kernel, the LVS function is added, can be in the form of files, can also be used as a module to load standing use. The software is used to define the rules that the kernel uses to define functions.  28.lvs-nat (two nodes to do) vim/etc/sysctl.conf Net.ipv4.ip_forward = 1 sysctl-p yum groupinstall-y "Load Balancer" (if there is no package, find the yum Source yourself) piranha-passed (set password) service Piranha-gui start29. Open page ip:3636 access GLO        BAL SETTINGS Public IP Intranet IP mode--nat router IP Intranet IP last one changed to 254 device eth1:1 redundancy Another machine's public intranet ipvirtual servers add rules, open rules, edit rules Nameportprotocolvipdevice eth0:1 (sub-interface) REAL Server add two intranet port monitoring SCRIPT s first/bin/ceshi%h the third yes in the native Yum install-y links (two units are installed) vim/bin/ceshi#! /bin/bash/usr/bin/links-dump 1 >/dev/null 2>&1if [0-eq $?]; Then Echo Yeselse echo No fichmod +x ceshiscp/etc/sysconfig/ha/lvs.cf another machine Scp/bin/ceshi another machine service pulse start (two units) via ETH1 Login to Machine (two machines like) Ifdown eth0route Add default GW x.x.x.1route add default GW x.x.x.254service httpd start

Fence use

redhat6.5 based on GUI virtualization cluster