VXVM 4.0 建立snapshot volume
时间:2008-01-16 13:33:41 来源: 作者:
|
volume snapshot Traditional:1. 建一个Diskgroup至少3块盘(每mirror plex不能在同一块盘上) 2. 在这个Diskgroup创建一个layout=mirror的Volume nmirror=2 vxassist -g datadg make datavol 100m layout=mirror,nolog nmirror=2 alloc=datadg01 datadg02 2. vxvol set fastresync=on datavol 打开fastresync 3. vxassist -g datadg snapstart datavol(为snapshot做准备,即加入一个Plex形成3个Plex的Volume 新加的Plex是在datadg中的另一块subdisk上建立的,并且与前2个Plex进行同步 4. vxassist -g datadg snapshot datavol snapvol 新加入的Plex与datavol detach并生成snapshot volume 5. vxassist -g datadg snapback snapvol snapshot还原到原来的Volume中,和snapstart时一样 6. Clear the association to create an independent volume. vxassist -g datadg snapclear snapvol 7.Destroy PITC vxassist -g datadg remove volume snapvol FastResync must be enabled for a volume in order to use this command. Snapshot objects (ss) and detached plexes (dp) are used by FastResync to track associations between a volume and its snapshot volumes Instants: 1. vxassist -g datadg make datavol 100m datadg02建立datavol卷 2. vxsnap -g datadg prepare datavol (Enable FastResync) 此时自动建立datavol_dcl卷(100MB的datavol,datavol_dcl只有227KB,datavol有几个Plex datavol_dcl就建立几个 DCO log Plex用来记录做snapshot后datavol上数据的改变,以便通过snapshot可以快速恢复 3. vxsnap -g datadg addmir datavol 为datavol添加mirror plex(同Traditonal的第3步) 4. 建立一个空Volume作为snapshot volume用 (如果建空Volume转到5,也可以不建转到6) LEN=`vxprint -g datadg -F%len datavol` DCONAME=`vxprint -g datadg -F%dco_name datavol` RSZ=`vxprint -g datadg -F%regionsz $DCONAME` vxassist -g datadg make snapvol $LEN init=active vxsnap -g datadg prepare snapvol regionsize=$RSZ 5. vxsnap -g datadg make source=datavol/snapvol=snapvol 把datavol中的一个Plex移动到snapvol作为snapshot volume 并不把datavol里原来新加的Mirror Plex转为snapvol volume里作为Plex 而是在datadg里的另一快磁盘建立subdisk并作为snapvol的Plex 6. vxsnap -g datadg make source=datavol/newvol=snapvol/plex=datavol-01(用哪个Plex去建新的Volume) 7. 刷新snapvol与datavol同步 vxsnap -g datadg refresh snapvol source=datavol 8. 重新加入回datavol变为一个Mirror Plex vxsnap -g datadg reattach snapvol source=datavol 9. 解除snapvol与datavol的关系,snapvol变为独立的Volume vxsnap -g datadg dis snapvol 10.删除snapvol vxedit -g datadg -rf rm snapvol OHP(Off-Host Processing ): 1. Enable FastResync on the volume: # vxsnap -g datadg prepare datavol 2. Create an empty volume for use as the snapshot volume. 3. Suspend updates to the volume and unmount the file system: # umount /mnt1 4. Create a full-sized instant snapshot for the volume: # vxsnap -g datadg make source=datavol/snapvol=snapvol 5. If you temporarily suspended updates to the volume by a database, release the tables from hot-backup mode. 6. Split the snapshot volume into a separate disk group: # vxdg split datadg offhostdg snapvol 7. Deport the disk group: # vxdg deport offhostdg 8. On the off-host processing host, import the disk group: # vxdg import offhostdg 9. To perform off-host processing, check and mount the snapshot volume: # fsck -F vxfs /dev/vx/rdsk/offhostdg/snapvol # mount -F vxfs /dev/vx/dsk/offhostdg/snapvol /mnt1 10. When processing is complete, unmount the snapshot: # umount /mnt1 11. Deport the disk group: # vxdg deport offhostdg 12. On the primary host, reimport the disk group: # vxdg import offhostdg 13. Rejoin the disk group with the original disk group: # vxdg join offhostdg datadg 14. Restart the snapshot volume using one of the following: # vxrecover -g datadg -m snapvol # vxvol -g datadg startall 15. Refresh the snapshot volume with the original volume: # vxsnap -g datadg refresh snapvol datavol File System:文件系统上的snapshot,只能恢复逻辑上的错误 Traditional和Instant snapshot自己一步步做过记下的 OHP没来得及做,照着文档整理了一下 david5337 回复于:2004-11-08 11:14:38 顶上,这么好的东西,大家应该捧场啊! 另外,楼主如果能够加上针对数据库在这个snapshot的备份、数据恢复,那就是锦上添花了! 笑看风云_淡 回复于:2004-11-08 11:37:24 引用:原帖由 "david5337" 发表: 顶上,这么好的东西,大家应该捧场啊! 另外,楼主如果能够加上针对数据库在这个snapshot的备份、数据恢复,那就是锦上添花了! up anber45 回复于:2004-11-08 12:36:06 不错,虽然好久没有做过了,还是用力顶一下! 川菜厨子 回复于:2004-11-15 08:43:14 没办法,实验环境没有数据库 nimysun 回复于:2004-12-16 17:58:38 太高深,不懂。 妖刀 回复于:2005-02-06 14:33:23 简单,无他,唯手熟耳 vahmoiz 回复于:2005-02-27 15:14:46 硬件快照: http://iqstor.vicp.net/index.php?pid=solutions&p1id=2 |
原文链接:http://bbs.chinaunix.net/viewthread.php?tid=440108 转载请注明作者名及原文出处 |










文章评论
共有 位网友发表了评论 查看完整内容