Chinaunix

标题: 配置gfs遇到的问题。帮忙哦！ [打印本页]

作者: 04120103 时间: 2008-05-09 11:11
标题: 配置gfs遇到的问题。帮忙哦！
我的步骤：
1.在三台机子上改/etc/hosts(三台一样)
127.0.0.1             localhost.localdomain localhost
192.168.18.240          gfs-node01
192.168.20.224          gfs-node02
192.168.0.141          gnbd-server
//127.0.0.1那行原来就有，我没去掉
2.用system-config-cluster生成配置：
  在192.168.18.240机子上生成配置，保存为：/etc/cluster/cluster.conf
  复制一份到192.168.20.224上，保存为/etc/cluster/cluster.conf
  生成配置过程如下：
选择使用DLM
添加两个节点，名字分别为gfs-node01，gfs-node02，Quorum Votes值都设置为1
添加fence devices type：Glodbal Network Block Device Name：gnbd Server:gnbd-server
添加Failover Domains 名字为gnbd-server,并加入刚才创建的节点gnbd-node1,gnbd-node2
在gfs-node01,gfs-node02,中"manange fencing for this node"，选择"add a fence level"
//服务器不是虚拟机，两个节点是虚拟机
  生成的配置文件如下：
<cluster config_version="2" name="alpha_cluster">
      <fence_daemon clean_start="0" post_fail_delay="0" post_join_delay="3"/>
      <clusternodes>
            <clusternode name="gfs-node01" votes="1">
                     <fence>
                              <method name="1"/>
                     </fence>
            </clusternode>
            <clusternode name="gfs-node02" votes="1">
                     <fence>
                              <method name="1"/>
                     </fence>
            </clusternode>
      </clusternodes>
      <cman expected_votes="1" two_node="1"/>
      <fencedevices>
               <fencedevice agent="fence_gnbd" name="gnbd" server="gnbd-server"/>
      </fencedevices>
      <rm>
            <failoverdomains>
                     <failoverdomain name="gnbd-server" ordered="0" restricted="0">
                              <failoverdomainnode name="gfs-node01" priority="1"/>
                              <failoverdomainnode name="gfs-node02" priority="1"/>
                     </failoverdomain>
            </failoverdomains>
            <resources/>
      </rm>
</cluster>
3.在两个节点机子上加载模块
  modprobe gnbd
  modprobe gfs
  modprobe lock_dlm
4.在两个节点机子上启动服务
  service ccsd start
  service cman start
  service fenced start
  service clvmd start
  service gfs start
  service rgmanager start
5.在gnbd-server上执行如下命令：
  gnbd_serv -n
  gnbd_export -c -d /dev/sda1 -e global_disk
结果
  gnbd_export: created GNBD global_disk serving file /dev/sda1
6.在两个节点上执行如下命令：
  gnbd_import -i gnbd-server
结果
  gnbd_import: created directory /dev/gnbd
  gnbd_import: created gnbd device global_disk
  gnbd_recvd: gnbd_recvd started
fence_tool join
//执行ccsd 和 cman_tool join 提示already running 因为我前面已经启动这些服务了
7.在gfs-node01上
  执行gfs_mkfs -p lock_dlm -t alpha_cluster:gfs -j 2 /dev/gnbd/global_disk
提示：gfs_mkfs: Partition too small for number/size of journals
  我想是空间不够。改小一点：//最小是32 默认是128
gfs_mkfs -p lock_dlm -t alpha_cluster:gfs -j 2 －J 32 /dev/gnbd/global_disk
执行结果
This will destroy any data on /dev/gnbd/global_disk.
It appears to contain a EXT2/3 filesystem.

Are you sure you want to proceed? [y/n] y

Device:                   /dev/gnbd/global_disk
Blocksize:                4096
Filesystem Size:          9680
Journals:                2
Resource Groups:          8
Locking Protocol:       lock_dlm
Lock Table:             alpha_cluster:gfs

Syncing...
All Done
可以了哦。
8.在两个节点机子的/ 目录下建立文件夹 gfstest
执行：
  mount -t gfs /dev/gnbd/global_disk /gfstest
问题：
在三个节点上执行：cat /proc/cluster/status
发现如下结果：
gfs-node01:
Protocol version: 5.0.1
Config version: 4
Cluster name: alpha_cluster
Cluster ID: 50356
Cluster Member: Yes
Membership state: Cluster-Member
Nodes: 1
Expected_votes: 1
Total_votes: 1
Quorum: 1
Active subsystems: 6
Node name: gfs-node01
Node addresses: 192.168.18.240
gfs-node02:
Protocol version: 5.0.1
Config version: 4
Cluster name: alpha_cluster
Cluster ID: 50356
Cluster Member: Yes
Membership state: Cluster-Member
Nodes: 1
Expected_votes: 1
Total_votes: 1
Quorum: 1
Active subsystems: 6
Node name: gfs-node02
Node addresses: 192.168.20.224
gfs-server:
gnbd_export: created GNBD global_disk serving file /dev/sda1
[root@gnbd-server ~]# cat /proc/cluster/status
Protocol version: 5.0.1
Config version: 0
Cluster name:
Cluster ID: 0
Cluster Member: No
Membership state: Not-in-Cluster

执行 cat /proc/cluster/nodes时：
结果如下：
gfs-node01:
  Node  Votes Exp Sts  Name
1 1 1 M gfs-node01
gfs-node02
Node  Votes Exp Sts  Name
1 1 1 M gfs-node02
gfs-server
Node  Votes Exp Sts  Name

上述情况看感觉这三台机子是分开的。。。为什么呢？
怎么会这样！！！！
这是什么问题呢？怎么会没有节点呢？
help me !!!!

[ 本帖最后由 04120103 于 2008-5-9 11:24 编辑 ]

作者: caicheng1015 时间: 2008-05-10 22:36
GNBD怎么配置的呀?我怎么看见你GNBD没有配置么

欢迎光临 Chinaunix (http://bbs.chinaunix.net/)