1.环境及说明:
vmware 6.0
node1 (192.168.0.5) node2 (192.168.0.6) 网关 192.168.0.1
硬件设置均为 2cpu/256M/sda(4G)/sdb(1G)/sdc(1G) (sdb,sdc在系统安装完成后增加)
相关说明
1)#command command要在node1与node2分别执行一次。
2)node[12]#command 指command只在node1或2上执行.
3)----/路径/文件 配置文件,下面是文件内容.
2.安装:
1)操作系统安装(此时sdb,sdc未增加)
centos4.5 (2.6.9-55.ELsmp) 最小化安装到sda,sda的分区在安装过程中自动生成.安装完成后,增加两个新盘sdb(1G),sdc(1G).
2)更新系统,安装相关软件包.
cd /etc/yum.repos.d/
wget
yum -y install yum-plugin-fastestmirror
yum -y update
yum remove kernel-smp-2.6.9-55.0.2.EL kernel-2.6.9-55.0.2.EL
yum -y install device-mapper-multipath perl-Net-Telnet magma magma-devel magma-plugins ccs ccs-devel cman-kernel cman-kernel-smp cman-kernheaders cman cman-devel dlm-kernel dlm-kernel-smp dlm-kernheaders dlm dlm-devel fence GFS GFS-kernel GFS-kernel-smp GFS-kernheaders gnbd-kernel gnbd-kernel-smp gnbd-kernheaders gnbd gulm gulm-devel iddev iddev-devel piranha ipvsadm lvm2-cluster pyorbit gcc kernel-smp-devel-2.6.9-55.EL.i686 kernel-devel-2.6.9-55.EL.i686 rpm-build flex
3)安装drbd:(使用8.0.*版,因为在gfs下支持primary/primary方式)
cd /root
wget
tar -zxvf drbd-8.0.4.tar.gz
cd drbd-8.0.4
make rpm KDIR=/usr/src/kernels/2.6.9-55.EL-smp-i686
rpm -ivh dist/RPMS/i386/drbd-8.0.4-3.i386.rpm dist/RPMS/i386/drbd-km-2.6.9_55.ELsmp-8.0.4-3.i386.rpm
make clean
make rpm KDIR=/usr/src/kernels/2.6.9-55.EL-i686
rpm -ivh dist/RPMS/i386/drbd-km-2.6.9_55.EL-8.0.4-3.i386.rpm
3.配置与相关操作:
1)配置启动
/boot/grub.conf
default=0
timeout=5
splashimage=(hd0,0)/grub/splash.xpm.gz
hiddenmenu
title CentOS-4 i386 (2.6.9-55.ELsmp)
root (hd0,0)
kernel /vmlinuz-2.6.9-55.ELsmp ro root=/dev/VolGroup00/LogVol00
initrd /initrd-2.6.9-55.ELsmp.img
title CentOS-4 i386-up (2.6.9-55.EL)
root (hd0,0)
kernel /vmlinuz-2.6.9-55.EL ro root=/dev/VolGroup00/LogVol00
initrd /initrd-2.6.9-55.EL.img
使用2.6.9-55ELsmp内核为默认启动内核
2)配置drbd
----/etc/drbd.conf
global { usage-count yes; }
common { syncer { rate 100M; } }
resource r0 {
protocol C;
net {
cram-hmac-alg sha1;
shared-secret "FooFunFactory";
allow-two-primaries;
}
on node1 {
device /dev/drbd0;
disk /dev/sdb;
address 192.168.0.5:7789;
flexible-meta-disk internal;
}
on node2 {
device /dev/drbd0;
disk /dev/sdb;
address 192.168.0.6:7789;
flexible-meta-disk internal;
}
}
resource r1 {
protocol C;
net {
cram-hmac-alg sha1;
shared-secret "FooFunFactory";
allow-two-primaries;
}
on node1 {
device /dev/drbd1;
disk /dev/sdc;
address 192.168.0.5:7788;
flexible-meta-disk internal;
}
on node2 {
device /dev/drbd1;
disk /dev/sdc;
address 192.168.0.6:7788;
flexible-meta-disk internal;
}
}
分别镜像两个机子的sdb与sdc,生成drbd0与drbd1.
#drbdadm create-md r0 (上面定义的resource)
#drbdadm create-md r1
#service drbd start
#drbdadm -- --overwrite-data-of-peer primary all (系统启动是 drbd0,drbd1均为secondary,不可访问,此命令要加入到合适的服务启动脚本中,以使后续使用drbd0,drbd1设备的服务能正常启动)
#service drbd status
drbd driver loaded OK; device status:
version: 8.0.4 (api:86/proto:86)
SVN Revision: 2947 build by , 2007-07-03 16:04:10
0: cs:Connected st:Primary/Primary ds:UpToDate/UpToDate C r---
ns:151596 nr:16 dw:92 dr:154552 al:0 bm:26 lo:0 pe:0 ua:0 ap:0
resync: used:0/31 hits:9455 misses:13 starving:0 dirty:0 changed:13
act_log: used:0/127 hits:18 misses:0 starving:0 dirty:0 changed:0
1: cs:Connected st:Primary/Primary ds:UpToDate/UpToDate C r---
ns:68 nr:12 dw:80 dr:2328 al:0 bm:0 lo:0 pe:0 ua:0 ap:0
resync: used:0/31 hits:0 misses:0 starving:0 dirty:0 changed:0
act_log: used:0/127 hits:16 misses:0 starving:0 dirty:0 changed:0
好了,drbd安装正常。
3)配置cluster
----/etc/cluster/cluster.conf
使用fence_gnbd.
----/etc/sysconfig/cluster (目的,是在集群开始前插入模块)
modprobe lock_dlm
modprobe gnbd
modprobe gfs
----/etc/hosts
127.0.0.1 localhost.localdomain localhost
192.168.0.5 node1
192.168.0.6 node2
#service ccsd start
#service cman start
#service fenced start
#ccs_tool lsnode
Cluster name: cluster1, config_version: 1
Nodename Votes Nodeid Iface Fencetype
node1 1 human
node2 1 human
# ccs_tool lsfence
Name Agent
human fence_gnbd
# cman_tool services
Service Name GID LID State Code
Fence Domain: "default" 2 2 run -
[1 2]
好了,一切正常.
4)配置clvm
----/etc/lvm/lvm.conf
devices {
dir = "/dev"
scan = [ "/dev" ]
filter = [ "r|/dev/cdrom|", "r|/dev/sdb|","r|/dev/sdc|", "a|/dev/sda|","a|/dev/drbd|","r|.*|"]
(此处修改是为的避免出现duplicate uuid的错误,与实战1)和2)相关)
#filter = [ "r|/dev/cdrom|", "r|/dev/sdb|","r|/dev/sdc|", "a|/dev/sda|","a|/dev/mpath|","r|.*|"]
(此处修改是为的避免出现duplicate uuid的错误,与实战3)相关,使用此设置,注释掉上面的filter)
cache = "/etc/lvm/.cache"
write_cache_state = 1
sysfs_scan = 1
md_component_detection = 1
ignore_suspended_devices = 0
}
log {
verbose = 0
syslog = 1
file = "/var/log/lvm2.log"
overwrite = 0
level = 0
indent = 1
command_names = 0
prefix = " "
}
backup {
backup = 1
backup_dir = "/etc/lvm/backup"
archive = 1
archive_dir = "/etc/lvm/archive"
retain_min = 10
retain_days = 30
}
shell {
history_size = 100
}
global {
library_dir = "/usr/lib"
umask = 077
test = 0
activation = 1
proc = "/proc"
locking_type = 3 (此处改为3,使clvm生效)
fallback_to_clustered_locking = 1
fallback_to_local_locking = 1
locking_dir = "/var/lock/lvm"
}
activation {
missing_stripe_filler = "/dev/ioerror"
reserved_stack = 256
reserved_memory = 8192
process_priority = -18
mirror_region_size = 512
mirror_log_fault_policy = "allocate"
mirror_device_fault_policy = "remove"
}
#service clvmd start
5)配置multipath
----/etc/multipath.conf
devnode_blacklist {
devnode "^sd[a-z]" (此处两行目的是不检查本地磁盘)
devnode "^hd[a-z]"
}
defaults {
user_friendly_names yes
udev_dir /dev
polling_interval 10
selector "round-robin 0"
prio_callout /bin/true
path_checker readsector0
rr_min_io 100
rr_weight priorities
failback immediate
no_path_retry fail
}
device {
vendor "GNBD"
product "GNBD"
path_grouping_policy multibus
getuid_callout "/sbin/gnbd_import -q -U /block/%n"
path_checker directio
}
#service multipath start
至此所有配置完成。
#chkconfig drbd on
#chkconfig ccsd on
#chkconfig cman on
#chkconfig fenced on
#chkconfig clvmd on
#chkconfig multipath on
#chkconfig gfs on
确保相关服务自动启动。重启系统。
4.实战篇:
由于有多种顺序组合来形成最后的系统,下面开始分别实践主要的几种.
注意/etc/lvm/lvm.conf中filter的不同.
由于clvm的使用,lvm的相关命令只需在其中的一台机子上执行即可.
1)在本地形成lv,然后导出(导入).
node1#pvcreate /dev/drbd0 /dev/drbd1
node1#vgcreate test /dev/drbd0 /dev/drbd1
#service clvmd restart (不知道rhel 5.0中是否有改进,可以不重启)
#gnbd_serv -v
node1#lvcreate -L 1.99G test
node1#gnbd_export -d /dev/test/lvol0 -e gfs -u 1
node2#gnbd_export -d /dev/text/lvol0 -e gfs1 -u 1 (两个导出设备的uid相同,视为对同一个设备的多路径导出)
node1#gnbd_import -i node1 (此处是为了实验multipath,不建议导入本机的导出设备)
node1#gnbd_import -i node2
node1#multipath -ll
mpath0 (1)
[size=1 GB][features="0"][hwhandler="0"]
\_ round-robin 0 [active]
\_ #:#:#:# gnbd0 252:0 [active][ready]
\_ #:#:#:# gnbd1 252:1 [active][ready]
node1#gfs_mkfs -j 2 -p lock_dlm -t cluster1:gfs /dev/mpath/mpath0
#mount -t gfs /dev/mpath/mpath0 /mnt
好了,可以正常使用了。
/***********************************
下面回到未导出(导入)的状态.
#umount /mnt
#dmsetup remove /dev/mpath/mpath0
#gnbd_import -R
#gnbd_export -R
*************************************/
2)在本地形成gfs,然后导出(导入).
node1#gfs_mkfs -j 2 -p lock_dlm -t cluster1:gfs /dev/test/lvol0
node1#gnbd_export -d /dev/test/lvol0 -e gfs -u 1
node2#gnbd_export -d /dev/test/lvol0 -e gfs1 -u 1 (两个导出设备的uid相同,视为对同一个设备的多路径导出)
#gnbd_import -i node1 (此处是为了实验multipath,不建议导入本机的导出设备)
#gnbd_import -i node2
#multipath -ll
mpath0 (1)
[size=1 GB][features="0"][hwhandler="0"]
\_ round-robin 0 [active]
\_ #:#:#:# gnbd0 252:0 [active][ready]
\_ #:#:#:# gnbd1 252:1 [active][ready]
node1#mount -t gfs /dev/mpath/mpath0 /mnt
/***********************************
下面回到未使用lvm前的状态.
#umount /mnt
#dmsetup remove /dev/mpath/mpath0
#gnbd_import -R
#gnbd_export -R
node1#lvchange -an /dev/test/lvol0
node1#lvremove /dev/test/lvol0
node1#vgremove test
node1#pvremove /dev/drbd0 /dev/drbd1
*************************************/
3)直接导出(导入)设备drbd0,drbd1.(推荐使用,client分担了server的设备管理的部分负载,可以提高性能)
node1#gnbd_export -d /dev/drbd0 -e gfs -u 1
node2#gnbd_export -d /dev/drbd0 -e gfs1 -u 1
node1#gnbd_export -d /dev/drbd1 -e gfs2 -u 2
node2#gnbd_export -d /dev/drbd1 -e gfs3 -u 2
#gnbd_import -i node1
#gnbd_import -i node2
#multipath -ll
mpath1 (2)
[size=1023 MB][features="0"][hwhandler="0"]
\_ round-robin 0 [enabled]
\_ #:#:#:# gnbd2 252:2 [active][ready]
\_ #:#:#:# gnbd0 252:0 [active][ready]
mpath0 (1)
[size=1023 MB][features="0"][hwhandler="0"]
\_ round-robin 0 [enabled]
\_ #:#:#:# gnbd3 252:3 [active][ready]
\_ #:#:#:# gnbd1 252:1 [active][ready]
node1#pvcreate /dev/mpath/mpath0 /dev/mpath/mpath1
node1#vgcreate test /dev/mpath/mpath0 /dev/mpath/mpath1
#service clvmd restart (不知道rhel 5.0中是否有改进,可以不重启)
node1#lvcreate -L 1.9G test
node1#gfs_mkfs -j 2 -p lock_dlm -t cluster1:gfs /dev/test/lvol0
#mount -t gfs /dev/test/lvol0 /mnt
5.结束语
写此文档的目的是为了提供给大家一个关于gfs应用的一个基本的范例,将关于clvm,drbd,multipath的概念变成实在的东西供大家来学习讨论。
参照此文,稍加修改后,多gnbd_server,多路径及使用drbd来构建便宜的SAN应用都不会有太多的问题。
6.不足之处
对于更细化的概念有赖于读者的自行补充,尚可加入lvs,未提及。fence,multipath未充分使用。
还有更多更多,希望有人不吝指教。