gfs 基本用法

 chkconfig --list |grep cman
cman            0:off   1:off   2:on    3:on    4:on    5:on    6:off

chkconfig --list |grep rgmanager
rgmanager       0:off   1:off   2:on    3:on    4:on    5:on    6:off


###md3200 管理软件
chkconfig --list |grep SMagent
SMagent         0:off   1:off   2:off   3:off   4:off   5:off   6:off


关闭时先关闭，rgmanager 再关闭cman

启动时先启动cman在启动rgmanager

#clustat
Cluster Status for datacluster @ Wed Jun 26 21:13:14 2013
Member Status: Quorate

 Member Name                                                     ID   Status
 ------ ----                                                     ---- ------
 host01                                                              1 Online
 host02                                                              2 Online
 host03                                                              3 Online, Local
 /dev/block/253:3                                                    0 Online, Quorum Disk

# cman_tool status
Version: 6.2.0
Config Version: 17
Cluster Name: datacluster
Cluster Id: 38426
Cluster Member: Yes
Cluster Generation: 3892
Membership state: Cluster-Member
Nodes: 3
Expected votes: 3
Quorum device votes: 2
Total votes: 5
Node votes: 1
Quorum: 3 
Active subsystems: 10
Flags:
Ports Bound: 0 177 
Node name: host03
Node ID: 3
Multicast addresses: 239.192.150.176
Node addresses: 192.168.128.3

# cat /etc/cluster/cluster.conf


       
               
                       
                               
                       
               
               
                       
                               
                       
               
               
                       
                               
                       
               
       
       
               
               
               
       
       
       



表决盘制定为2票，每个节点的每个分区是1票， 启动的票数必须等于或大于 cman期望的票数3票

防止脑裂，某节点因为服务器负载或各种忙，无法及时回应其他节点的应答请求，导致其他节点认为某节点坏，而夺取共享磁盘的读写权限，而某节点认为他是正常的而继续访问共享磁盘，导致整个共享文件系统不一致崩溃！