Glusterfs 分布式存储系统配置
在服务器集群中,为了同步更新APP内容,利用Glusterfs分布式存储功达到相同的APP只更新一次,其他机器同步更新
试验环境:VmwareFusion3.1.1CentOS5.5
1.安装
#yum-yinstallglusterfs-clusterglusterfs-server
2.设置两台机器的别名分别为svr1,svr2
3.两台机器的服务端配置文件和客户端文件如下:
服务端配置文件glusterfsd.vol
[root@svr1~]#cat/etc/glusterfs/glusterfsd.vol
###file:server-volume.vol.sample
#####################################
###GlusterFSServerVolumeFile##
#####################################
####CONFIGFILERULES:
###"#"iscommentcharacter.
###-Configfileiscasesensitive
###-Optionswithinavolumeblockcanbeinanyorder.
###-Spacesortabsareusedasdelimitterwithinaline.
###-Multiplevaluestooptionswillbe:delimitted.
###-Eachoptionshouldendwithinaline.
###-Missingorcommentedfieldswillassumedefaultvalues.
###-Blank/commentedlinesareallowed.
###-Sub-volumesshouldalreadybedefinedabovebeforereferring.
###Exportvolume"brick"withthecontentsof"/home/export"directory.
volumeposix
typestorage/posix#POSIXFStranslator
optiondirectory/opt/export#Exportthisdirectory
end-volume
volumelocks
typefeatures/locks
#optionmandatory-lockson
subvolumesposix
end-volume
volumebrick
typeperformance/io-threads
optionthread-count8
subvolumeslocks
end-volume
###Addnetworkservingcapabilitytoabovebrick.
volumeserver
typeprotocol/server
optiontransport-typetcp
#optiontransport-typeunix
#optiontransport-typeib-sdp
#optiontransport.socket.bind-address192.168.1.10#Defaultistolistenonallinterfaces
#optiontransport.socket.listen-port6996#Defaultis6996
#optiontransport-typeib-verbs
#optiontransport.ib-verbs.bind-address192.168.1.10#Defaultistolistenonallinterfaces
#optiontransport.ib-verbs.listen-port6996#Defaultis6996
#optiontransport.ib-verbs.work-request-send-size131072
#optiontransport.ib-verbs.work-request-send-count64
#optiontransport.ib-verbs.work-request-recv-size131072
#optiontransport.ib-verbs.work-request-recv-count64
#optionclient-volume-filename/etc/glusterfs/glusterfs-client.vol
subvolumesbrick
#NOTE:Accesstoanyvolumethroughprotocol/serverisdeniedby
#default.Youneedtoexplicitlygrantaccessthrough#"auth"
#option.
optionauth.addr.brick.allow127.0.0.1,192.168.*#Allowaccessto"brick"volume
end-volume
客户端配置文件glusterfs.vol
[root@svr1~]#cat/etc/glusterfs/glusterfs.vol
###file:client-volume.vol.sample
#####################################
###GlusterFSClientVolumeFile##
#####################################
####CONFIGFILERULES:
###"#"iscommentcharacter.
###-Configfileiscasesensitive
###-Optionswithinavolumeblockcanbeinanyorder.
###-Spacesortabsareusedasdelimitterwithinaline.
###-Eachoptionshouldendwithinaline.
###-Missingorcommentedfieldswillassumedefaultvalues.
###-Blank/commentedlinesareallowed.
###-Sub-volumesshouldalreadybedefinedabovebeforereferring.
###Addclientfeatureandattachtoremotesubvolume
volumesvr1
typeprotocol/client
optiontransport-typetcp
optionremote-hostsvr1
optionremote-subvolumebrick#nameoftheremotevolume
end-volume
volumesvr2
typeprotocol/client
optiontransport-typetcp
optionremote-hostsvr2
optionremote-subvolumebrick
end-volume
volumereplicate1
typecluster/replicate
subvolumessvr1
end-volume
volumereplicate2
typecluster/replicate
subvolumessvr2
end-volume
volumedistribute
typecluster/distribute
subvolumesreplicate1replicate2
#subvolumesreplicate1
end-volume
volumewritebehind
typeperformance/write-behind
optioncache-size1MB
subvolumesdistribute
end-volume
volumecache
typeperformance/io-cache
optioncache-size64MB
subvolumeswritebehind
end-volume
4.接下来启动服务端
[root@svr1~]#glusterfsd-f/etc/glusterfs/glusterfsd.vol
执行psfax|grepgluseterfs查看是否有启动
使用以上命令操作svr2
5.启动客户端
启动客户端之前先挂起fuse
[root@svr1~]#modprobefuse
启动客户端
[root@svr1~]#glusterfs-l/var/log/glusterfs/opt-glusterfs.log-f/etc/glusterfs/glusterfs.vol/opt/clusterfs
使用df-h查看挂载是否成功
[root@svr1~]#df-h
文件系统容量已用可用已用%挂载点
/dev/hda25.7G4.6G858M85%/
/dev/hda32.9G69M2.7G3%/home
/dev/hda199M12M82M13%/boot
tmpfs502M0502M0%/dev/shm
glusterfs#/etc/glusterfs/glusterfs.vol
12G9.0G1.9G84%/opt/clusterfs
如果没有如上显示，则说明挂载没有成功，可以查看/var/log/glusterfs/opt-glusterfs.log下建立的日志进行排错
最后使用以上命令进行svr2的操作
注:假如重启到glusterfs挂连接不上,可以使用命令取消挂起
[root@svr1~]#df-h
FilesystemSizeUsedAvailUse%Mountedon
/dev/mapper/vg_webserver2-lv_root
50G2.0G45G5%/
tmpfs1.5G01.5G0%/dev/shm
/dev/sda1485M37M423M8%/boot
/dev/mapper/vg_webserver2-lv_home
46G180M44G1%/home
df:`/opt/clusterfs':Transportendpointisnotconnected
[root@svr1~]#umount/opt/clusterfs
root@svr2~]#umount/opt/clusterfs
================================
最后的效果是,不管在哪台机器上的/opt/glusterfs下新建更新文件,都会同步更新到另一台机器上,当你相同的APP在N台机器上时,这是多么轻松的事啊