五、gfs文件系统的挂载
1、可挂载gfs文件系统节点的必要条件
节点必须是gfs所在cluster的一个节点，须防止非gfs所在cluster节点挂载gfs文件系统
检查挂载节点的cluster是否就是gfs的cluter，可以查看gfs的superblock内容
[root@node1 ~]# gfs_tool df /var/www/html
/var/www/html:
SB lock proto = "lock_dlm"
SB lock table = "cluster8:gfslv01" #gfs所属cluster
SB ondisk format = 1309
SB multihost format = 1401
Block size = 4096
Journals = 4
Resource Groups = 14
Mounted lock proto = "lock_dlm"
Mounted lock table = "cluster8:gfslv01"
Mounted host data = "jid=1:id=196611:first=0"
Journal number = 1
Lock module flags = 0
Local flocks = FALSE
Local caching = FALSE
Oopses OK = FALSE
Type Total Blocks Used Blocks Free Blocks use%
------------------------------------------------------------------------
inodes 5 5 0 100%
metadata 192 0 192 0%
data 1441471 0 1441471 0%
2、mount选项
格式：mount -o StdMountOpts,GFSOptions -t gfs Device Mountpoint
#支持标准选项和gfs特定选项
lockproto=[lock_dlm,lock_nolock]：锁机制，lock_dlm用于cluster，lock_nolock用于单机
locktable=clustername:fsname ：cluster和文件系统名
upgrade : 提升，多用新gfs版本挂载
acl ：开启访问控制列表
[root@node1 ~]# mount -o lockproto=lock_dlm /dev/vg01/lv01 /var/www/html
[root@node1 ~]# df -h
文件系统 容量 已用 可用 已用% 挂载点
/dev/mapper/vg00-lv00 49G 2.9G 44G 7% /
/dev/sda1 251M 23M 216M 10% /boot
tmpfs 2.0G 0 2.0G 0% /dev/shm
/dev/mapper/vg01-lv01 5.5G 24K 5.5G 1% /var/www/html
[root@node1 ~]# umount /var/www/html #卸载文件系统
六、Journals日志管理
1、查看superblock信息
[root@node1 ~]# gfs_tool df /var/www/html
/var/www/html:
SB lock proto = "lock_dlm" #锁机制
SB lock table = "cluster8:gfslv01" #cluster
SB ondisk format = 1309
SB multihost format = 1401
Block size = 4096 #块大小，gfs_mkfs -b指定块大小
Journals = 4 #日志数量
Resource Groups = 14
Mounted lock proto = "lock_dlm"
Mounted lock table = "cluster8:gfslv01"
Mounted host data = "jid=2:id=196611:first=0"
Journal number = 2
Lock module flags = 0
Local flocks = FALSE
Local caching = FALSE
Oopses OK = FALSE
Type Total Blocks Used Blocks Free Blocks use%
------------------------------------------------------------------------
inodes 5 5 0 100%
metadata 192 0 192 0%
data 1441471 0 1441471 0%
2、检查日志数量
[root@node1 ~]# gfs_tool jindex /var/www/html | grep Journal
Journal 0:
Journal 1:
Journal 2:
Journal 3:
3、增加日志数量
要求：lv有足够的空间支持日志数量的增加
必须在拉升gfs文件系统前前增加日志数量
[root@node1 ~]# gfs_jadd -Tv -j 2 /var/www/html #增加日志数前测试，确保磁盘空间足够
Requested size (65536 blocks) greater than available space (0 blocks)
[root@node1 ~]# gfs_jadd -j 2 /var/www/html #增加2个日志
Requested size (65536 blocks) greater than available space (0 blocks)
七、GFS系统的扩展
1、扩展逻辑卷
[root@node1 ~]# pvcreate /dev/sde
Physical volume "/dev/sde" successfully created
[root@node1 ~]# vgextend vg01 /dev/sde
Volume group "vg01" successfully extended
[root@node1 ~]# lvextend -L +5120M /dev/vg01/lv01 #扩展逻辑卷
Extending logical volume lv01 to 6.00 GB
Logical volume lv01 successfully resized
2、扩展gfs（gfs_grow -v Device|Mount_point）
[root@node1 ~]# gfs_grow -v /dev/vg01/lv01 #扩展gfs
FS: Mount Point: /var/www/html
FS: Device: /dev/mapper/vg01-lv01
FS: Options: rw,hostdata=jid=1:id=196611:first=0
FS: Size: 262142
RGRP: Current Resource Group List:
RI: Addr 245760, RgLen 2, Start 245762, DataLen 16380, BmapLen 4095
RI: Addr 229376, RgLen 2, Start 229378, DataLen 16380, BmapLen 4095
RI: Addr 212992, RgLen 2, Start 212994, DataLen 16380, BmapLen 4095
RI: Addr 196608, RgLen 2, Start 196610, DataLen 16380, BmapLen 4095
RI: Addr 49157, RgLen 2, Start 49159, DataLen 16376, BmapLen 4094
RI: Addr 32778, RgLen 2, Start 32780, DataLen 16376, BmapLen 4094
RI: Addr 16399, RgLen 2, Start 16401, DataLen 16376, BmapLen 4094
RI: Addr 17, RgLen 2, Start 19, DataLen 16380, BmapLen 4095
RGRP: 8 Resource groups in total
JRNL: Current Journal List:
JI: Addr 163840 NumSeg 2048 SegSize 16
JI: Addr 131072 NumSeg 2048 SegSize 16
JI: Addr 98304 NumSeg 2048 SegSize 16
JI: Addr 65536 NumSeg 2048 SegSize 16
JRNL: 4 Journals in total
DEV: Size: 1572864
RGRP: New Resource Group List:
RI: Addr 262142, RgLen 6, Start 262148, DataLen 91120, BmapLen 22780
RI: Addr 353269, RgLen 15, Start 353284, DataLen 243904, BmapLen 60976
RI: Addr 597188, RgLen 15, Start 597203, DataLen 243904, BmapLen 60976
RI: Addr 841107, RgLen 15, Start 841122, DataLen 243904, BmapLen 60976
RI: Addr 1085026, RgLen 15, Start 1085041, DataLen 243904, BmapLen 60976
RI: Addr 1328945, RgLen 15, Start 1328960, DataLen 243904, BmapLen 60976
RGRP: 6 Resource groups in total
Preparing to write new FS information...
Done.
[root@node1 ~]# df -h
文件系统 容量 已用 可用 已用% 挂载点
/dev/mapper/vg00-lv00
49G 2.9G 44G 7% /
/dev/sda1 251M 23M 216M 10% /boot
tmpfs 2.0G 0 2.0G 0% /dev/shm
/dev/mapper/vg01-lv01
5.5G 20K 5.5G 1% /var/www/html
八、gfs文件系统inode节点的动态分配
1、GFS有inode节点空间自动分配的能力（就是说不用为inode预先分配磁盘空间，GFS自己会去动态分配）；
2、GFS的inode有4K（默认）一个块那么大（普通的fs在2K左右），为了增进效率，如果文件内容足够小，会被直接存放在inode内；
3、手动回收无用的inode所占空间:gfs_tool reclaim <mountpoint>
[root@node1 ~]# gfs_tool reclaim /var/www/html
Don't do this if this file system is being exported by NFS (on any machine).
Are you sure you want to proceed? [y/n] y
Reclaimed:
version 0
inodes 0
metadata 192