GlusterFS適合存儲大文件,小文件性能較差。生產上可使用GlusterFS+Cinder爲Openstack提供塊存儲,存儲虛擬機鏡像。node
GlusterFS是Scale-Out存儲解決方案Gluster的核心,它是一個開源的分佈式文件系統,具備強大的橫向擴展能力,經過擴展可以支持數PB存儲容量和處理數千客戶端。GlusterFS藉助TCP/IP或InfiniBand RDMA網絡將物理分佈的存儲資源彙集在一塊兒,使用單一全局命名空間來管理數據。GlusterFS基於可堆疊的用戶空間設計,可爲各類不一樣的數據負載提供優異的性能。linux
本文使用如下環境進行部署。git
主機名centos |
IP地址服務器 |
角色網絡 |
備註架構 |
linux-node1.unixhot.com併發 |
192.168.56.11異步 |
Server、Clienttcp |
|
linux-node2.unixhot.com |
192.168.56.12 |
Server、Client |
目前Gluster的最新版本爲3.4.3版本,直接使用官方提供的repo進行安裝。
[root@linux-node2 ~]# yum install centos-release-gluster
[root@linux-node2 ~]# yum install glusterfs-server
驗證安裝
[root@linux-node1 yum.repos.d]# glusterfs -V
glusterfs 3.4.3 built on Apr 3 2014 16:02:46
Repository revision: git://git.gluster.com/glusterfs.git
Copyright (c) 2006-2013 Red Hat, Inc. http://www.redhat.com/
啓動Glusterd
[root@linux-node1 yum.repos.d]# /etc/init.d/glusterd start
請安裝並啓動linux-node2
在開始建立ClusterFS卷以前,須要建立一個稱之爲Trusted Storage的池,是一個可信的網絡存儲服務器,能夠理解爲集羣。爲卷提供bricks。
gluster peer probe<主機名或者IP地址>,建議是IP地址,或者內網用DNS作主機名解析,或者在/etc/hosts作好綁定。
添加成員服務器到Poll中:例
[root@linux-node1 ~]# gluster peer probe 192.168.1.12
peer probe: success
刪除集羣中的成員服務器:例
[root@linux-node1 ~]# gluster peer detach linux-node2.unixhot.com
peer detach: success
注意不要添加本地的服務器。
[root@linux-node1 ~]# gluster peer probe 10.6.16.1
peer probe: success: on localhost not needed
[root@linux-node1 ~]# gluster peer probe 10.6.16.2
peer probe: success
[root@linux-node1 ~]# gluster peer status
Number of Peers: 3
GlusterFS的卷共有三種基本類型,能夠組合共7種類型,其中Distributed、Replicated、Strped、爲基本類型。Distributed Striped、Distributed Replicated、Distributed Striped Replicated、Striped Replicated。爲組合類型。
本文爲了測試7種卷類型,須要在各個服務器分別建立7個目錄。
[root@linux-node1 ~]# mkdir /data && cd /data && mkdir type{1,2,3,4,5,6,7}/exp{1,2,3,4,5,6,7} –p
建立Volumes使用如下命令進行:
gluster volume create NEW-VOLNAME [stripe COUNT | replica COUNT]
[transport [tcp | rdma | tcp,rdma]] NEW-BRICK1 NEW-BRICK2 NEW-BRICK3...
分佈式卷架構示意圖:
[root@linux-node1 data]# gluster volume create test1-volume 10.6.16.1:/data/type1/exp1 10.6.16.2:/data/type1/exp1 10.6.11.31:/data/type1/exp1 10.6.11.32:/data/type1/exp1 force
volume create: test1-volume: success: please start the volume to access data
[root@linux-node1 data]# gluster volume info test1-volume
Volume Name: test1-volume
Type: Distribute
Volume ID: 1ed9c3fe-3bc7-4525-b194-219a8e85f7fb
Status: Created
Number of Bricks: 4
Transport-type: tcp
Bricks:
Brick1: 10.6.16.1:/data/type1/exp1
Brick2: 10.6.16.2:/data/type1/exp1
Brick3: 10.6.11.31:/data/type1/exp1
Brick4: 10.6.11.32:/data/type1/exp1
複製卷的架構示意圖。
[root@linux-node1 data]# gluster volume create test2-volume replica 4 10.6.16.1:/data/type2/exp1 10.6.16.2:/data/type2/exp1 10.6.11.31:/data/type2/exp1 10.6.11.32:/data/type2/exp1 force
volume create: test2-volume: success: please start the volume to access data
[root@linux-node1 data]# gluster volume info test2-volume
Volume Name: test2-volume
Type: Replicate
Volume ID: 035b6768-2d07-4530-aa91-38571cacc90c
Status: Created
Number of Bricks: 1 x 4 = 4
Transport-type: tcp
Bricks:
Brick1: 10.6.16.1:/data/type2/exp1
Brick2: 10.6.16.2:/data/type2/exp1
Brick3: 10.6.11.31:/data/type2/exp1
Brick4: 10.6.11.32:/data/type2/exp1
條帶卷架構示意圖:
[root@linux-node1 ~]# gluster volume create test3-volume stripe 4 10.6.16.1:/data/type3/exp1 10.6.16.2:/data/type3/exp1 10.6.11.31:/data/type3/exp1 10.6.11.32:/data/type3/exp1 force
volume create: test3-volume: success: please start the volume to access data
[root@linux-node1 ~]# gluster volume info test3-volume
Volume Name: test3-volume
Type: Stripe
Volume ID: 6d754a08-c0e9-4fb8-bc8e-54c8f1f3cbe4
Status: Created
Number of Bricks: 1 x 4 = 4
Transport-type: tcp
Bricks:
Brick1: 10.6.16.1:/data/type3/exp1
Brick2: 10.6.16.2:/data/type3/exp1
Brick3: 10.6.11.31:/data/type3/exp1
Brick4: 10.6.11.32:/data/type3/exp1
分佈式條帶卷架構示意圖:
[root@linux-node1 ~]# gluster volume create test4-volume stripe 2 10.6.16.1:/data/type4/exp1 10.6.16.2:/data/type4/exp1 10.6.11.31:/data/type4/exp1 10.6.11.32:/data/type4/exp1 force
volume create: test4-volume: success: please start the volume to access data
[root@linux-node1 ~]# gluster volume info test4-volume
Volume Name: test4-volume
Type: Distributed-Stripe
Volume ID: 2b14574c-79c6-4329-91b9-8dc036969f2f
Status: Created
Number of Bricks: 2 x 2 = 4
Transport-type: tcp
Bricks:
Brick1: 10.6.16.1:/data/type4/exp1
Brick2: 10.6.16.2:/data/type4/exp1
Brick3: 10.6.11.31:/data/type4/exp1
Brick4: 10.6.11.32:/data/type4/exp1
分佈式複製卷架構示意圖:
[root@linux-node1 ~]# gluster volume create test5-volume replica 2 10.6.16.1:/data/type5/exp1 10.6.16.2:/data/type5/exp1 10.6.11.31:/data/type5/exp1 10.6.11.32:/data/type5/exp1 force
volume create: test5-volume: success: please start the volume to access data
[root@linux-node1 ~]# gluster volume info test5-volume
Volume Name: test5-volume
Type: Distributed-Replicate
Volume ID: fc94f6f7-a3e4-471b-82b0-4e74d5a529b9
Status: Created
Number of Bricks: 2 x 2 = 4
Transport-type: tcp
Bricks:
Brick1: 10.6.16.1:/data/type5/exp1
Brick2: 10.6.16.2:/data/type5/exp1
Brick3: 10.6.11.31:/data/type5/exp1
Brick4: 10.6.11.32:/data/type5/exp1
[root@linux-node1 ~]# gluster volume create test6-volume stripe 2 replica 2 10.6.16.1:/data/type6/exp1 10.6.16.1:/data/type6/exp2 10.6.16.2:/data/type6/exp1 10.6.16.2:/data/type6/exp2 10.6.11.31:/data/type6/exp1 10.6.11.31:/data/type6/exp2 10.6.11.32:/data/type6/exp1 10.6.11.32:/data/type6/exp2 force
Multiple bricks of a replicate volume are present on the same server. This setup is not optimal.
Do you still want to continue creating the volume? (y/n) y
volume create: test6-volume: success: please start the volume to access data
[root@linux-node1 ~]# gluster volume info test6-volume
Volume Name: test6-volume
Type: Distributed-Striped-Replicate
Volume ID: e6d681e3-b006-41ce-a466-c4c16cf03cce
Status: Created
Number of Bricks: 2 x 2 x 2 = 8
Transport-type: tcp
Bricks:
Brick1: 10.6.16.1:/data/type6/exp1
Brick2: 10.6.16.1:/data/type6/exp2
Brick3: 10.6.16.2:/data/type6/exp1
Brick4: 10.6.16.2:/data/type6/exp2
Brick5: 10.6.11.31:/data/type6/exp1
Brick6: 10.6.11.31:/data/type6/exp2
Brick7: 10.6.11.32:/data/type6/exp1
Brick8: 10.6.11.32:/data/type6/exp2
[root@linux-node1 ~]# gluster volume create test7-volume stripe 2 replica 2 10.6.16.1:/data/type7/exp1 10.6.16.2:/data/type7/exp1 10.6.11.31:/data/type7/exp1 10.6.11.32:/data/type7/exp1 force
volume create: test7-volume: success: please start the volume to access data
[root@linux-node1 ~]# gluster volume info test7-volume
Volume Name: test7-volume
Type: Striped-Replicate
Volume ID: 3f007e9e-3c42-4dd3-935e-60da041659d1
Status: Created
Number of Bricks: 1 x 2 x 2 = 4
Transport-type: tcp
Bricks:
Brick1: 10.6.16.1:/data/type7/exp1
Brick2: 10.6.16.2:/data/type7/exp1
Brick3: 10.6.11.31:/data/type7/exp1
Brick4: 10.6.11.32:/data/type7/exp1
在開始掛載使用卷以前,必需先啓動卷。使用如下命令啓動卷。
[root@linux-node3 yum.repos.d]# gluster volume start
Usage: volume start <VOLNAME> [force]
GlusterFS支持三種客戶端類型。Gluster Native Client、NFS和CIFS。官方推薦使用Native Client,可使用GlusterFS的所有功能。
Gluster Native Client是基於FUSE的,因此須要保證客戶端安裝了FUSE。這個是官方推薦的客戶端,支持高併發和高效的寫性能。
[root@linux-node4 ~]# yum -y install glusterfs-client
[root@linux-node4 ~]# mkdir /mnt/glusterfs
[root@linux-node4 ~]# mount.glusterfs 10.6.16.1:/test1-volume /mnt/glusterfs/
[root@linux-node4 ~]# mount
/dev/sda3 on / type ext4 (rw)
proc on /proc type proc (rw)
sysfs on /sys type sysfs (rw)
devpts on /dev/pts type devpts (rw,gid=5,mode=620)
tmpfs on /dev/shm type tmpfs (rw)
/dev/sda1 on /boot type ext4 (rw)
none on /proc/sys/fs/binfmt_misc type binfmt_misc (rw)
10.6.16.1:/test1-volume on /mnt/glusterfs type fuse.glusterfs (rw,default_permissions,allow_other,max_read=131072)
掛載參數
當使用mount –t glusterfs的使用,能夠執行相關的參數以下:
backupvolfile-server=server-name
volfile-max-fetch-attempts=number of attempts
log-level=loglevel
log-file=logfile
transport=transport-type
direct-io-mode=[enable|disable]
For example:
# mount -t glusterfs -o backupvolfile-server=volfile_server2 --volfile-max-fetch-attempts=2 log-level=WARNING,log-file=/var/log/gluster.log server1:/test-volume /mnt/glusterfs
If backupvolfile-server option is added while mounting fuse client, when the first volfile server fails, then the server specified in backupvolfile-server option is used as volfile server to mount the client.
In --volfile-max-fetch-attempts=X option, specify the number of attempts to fetch volume files while mounting a volume. This option is useful when you mount a server with multiple IP addresses or when round-robin DNS is configured for the server-name..
NFS默認是使用的UDP進行通訊,因爲GlusterFS不支持UDP。因此須要設置爲TCP的方式掛載。
mount -o mountproto=tcp -t nfs server1:/test-volume /mnt/glusterfs
[root@linux-node1 ~]# gluster volume stop test1-volume
[root@linux-node1 ~]# gluster volume delete test1-volume
GlusterFS支持在線進行卷的擴展。
[root@linux-node3 data]# gluster peer probe 10.6.11.21
peer probe: success
[root@linux-node3 data]# gluster volume add-brick test1-volume 10.6.11.21:/data/type1/exp1 force
volume add-brick: success
[root@linux-node3 data]# gluster volume info test1-volume
Volume Name: test1-volume
Type: Distribute
Volume ID: 1ed9c3fe-3bc7-4525-b194-219a8e85f7fb
Status: Stopped
Number of Bricks: 5
Transport-type: tcp
Bricks:
Brick1: 10.6.16.1:/data/type1/exp1
Brick2: 10.6.16.2:/data/type1/exp1
Brick3: 10.6.11.31:/data/type1/exp1
Brick4: 10.6.11.32:/data/type1/exp1
Brick5: 10.6.11.21:/data/type1/exp1
收縮卷和擴展卷類似據以Brick爲單位。
[root@linux-node3 data]# gluster volume remove-brick test1-volume 10.6.11.21:/data/type1/exp1 start
volume remove-brick start: success
ID: 15764d05-5675-46b5-8467-b9eb12570e61
[root@linux-node3 data]# gluster volume remove-brick test1-volume 10.6.11.21:/data/type1/exp1 status
[root@linux-node3 data]# gluster volume remove-brick test1-volume 10.6.11.21:/data/type1/exp1 commit
Removing brick(s) can result in data loss. Do you want to Continue? (y/n) y
volume remove-brick commit: success
[root@linux-node3 data]# gluster volume info test1-volume
Volume Name: test1-volume
Type: Distribute
Volume ID: 1ed9c3fe-3bc7-4525-b194-219a8e85f7fb
Status: Started
Number of Bricks: 4
Transport-type: tcp
Bricks:
Brick1: 10.6.16.1:/data/type1/exp1
Brick2: 10.6.16.2:/data/type1/exp1
Brick3: 10.6.11.31:/data/type1/exp1
Brick4: 10.6.11.32:/data/type1/exp1
進行卷遷移前須要確保遷移目標已經加入集羣。或者遷移的Brick目錄存在。遷移卷能夠執行暫停和停止的操做。
[root@linux-node3 data]# gluster volume replace-brick test1-volume 10.6.16.1:/data/type1/exp1 10.6.16.1:/data/type1/exp2 commit force
All replace-brick commands except commit force are deprecated. Do you want to continue? (y/n) y
暫停遷移
gluster volume replace-brick test1-volume 10.6.16.1:/data/type1/exp1 10.6.16.1:/data/type1/exp2 pause
停止遷移
gluster volume replace-brick test1-volume 10.6.16.1:/data/type1/exp1 10.6.16.1:/data/type1/exp2 abort
2.3.3.3替換Brick
# gluster volume replace-brick VOLNAME BRICKNEW-BRICK start/pause/abort/status/commit
#gluster volume replace-brick dht_vol server0:/mnt/sdb1 server0:/mnt/sdc1 start
//如上,執行replcace-brick卷替換啓動命令,使用start啓動命令後,開始將原始Brick的數據遷移到即將須要替換的Brick上。
#gluster volume replace-brick dht_vol server0:/mnt/sdb1 server0:/mnt/sdc1 status
//在數據遷移的過程當中,能夠查看替換任務是否完成。
#gluster volume replace-brick dht_vol server0:/mnt/sdb1 server0:/mnt/sdc1 abort
//在數據遷移的過程當中,能夠執行abort命令終止Brick替換。
#gluster volume replace-brick dht_vol server0:/mnt/sdb1 server0:/mnt/sdc1 commit
//在數據遷移結束以後,執行commit命令結束任務,則進行Brick替換。使用volume info命令能夠查看到Brick已經被替換。
開啓/關閉系統配額
# gluster volume quota VOLNAME enable/disable
//在使用系統配額功能時,須要使用enable將其開啓;disable爲關閉配額功能命令。
設置(重置)目錄配額
# gluster volume quota VOLNAME limit-usage /directory limit-value
#gluster volume quota dht_vol limit-usage /quota 10GB
//如上,設置dht_vol卷下的quota子目錄的限額爲10GB。
PS:這個目錄是以系統掛載目錄爲根目錄」/」,因此/quota即客戶端掛載目錄下的子目錄quota
配額查看
# gluster volume quota VOLNAME list
# gluster volume quota VOLNAME list /directory name
//可使用如上兩個命令進行系統卷的配額查看,第一個命令查看目的卷的全部配額設置,第二個命令則是執行目錄進行查看。
//能夠顯示配額大小及當前使用容量,若無使用容量(最小0KB)則說明設置的目錄多是錯誤的(不存在)。
地域複製(geo-replication)
# gluster volume geo-replication MASTER SLAVE start/status/stop
地域複製是系統提供的災備功能,可以將系統的所有數據進行異步的增量備份到另外的磁盤中。
#gluster volume geo-replication dht_vol 192.168.2.104:/mnt/sdb1 start
//如上,開始執行將dht_vol卷的全部內容備份到2.104下的/mnt/sdb1中的task,須要注意的是,這個備份目標不能是系統中的Brick。
Profile Command 提供接口查看一個卷中的每個brick的IO信息。
#gluster volume profile VOLNAME start
//啓動profiling,以後則能夠進行IO信息查看
#gluster volume profile VOLNAME info
//查看IO信息,能夠查看到每個Brick的IO信息
# gluster volume profile VOLNAME stop
//查看結束以後關閉profiling功能
Top command 容許你查看bricks的性能例如:read, write, file open calls, file read calls, file write calls, directory open calls, and directory real calls
全部的查看均可以設置top數,默認100
# gluster volume top VOLNAME open [brick BRICK-NAME] [list-cnt cnt]
//查看打開的fd
#gluster volume top VOLNAME read [brick BRICK-NAME] [list-cnt cnt]
//查看調用次數最多的讀調用
#gluster volume top VOLNAME write [brick BRICK-NAME] [list-cnt cnt]
//查看調用次數最多的寫調用
# gluster volume top VOLNAME opendir [brick BRICK-NAME] [list-cnt cnt]
# gluster volume top VOLNAME readdir [brick BRICK-NAME] [list-cnt cnt]
//查看次數最多的目錄調用
# gluster volume top VOLNAME read-perf [bs blk-size count count] [brick BRICK-NAME] [list-cnt cnt]
//查看每一個Brick的讀性能
# gluster volume top VOLNAME write-perf [bs blk-size count count] [brick BRICK-NAME] [list-cnt cnt]
//查看每一個Brick的寫性能