Hi,
If the private region of all the disks within an dg is corrupted,then if I reconfigure by hand all the logical volumes(thus the private region) with vxassist,is this safety?The data will not be lost from logical volumes.right?
>> partially correct. Reconfigure by hand means you can use "vxmake" commands to do a bottom up rebuild of entire dg however first, you need to reinitialize the disk with correct offset values (very important) for e.g lets say if entire DG configuration is corrupted ,you will intialize the disks with
# vxdisksetup -i <disk> format=<format> puboffset=<offset> privoffset=<offset> pubslice=<pubslice> privslice=<privslice>
All above information could be find in old veritas explorers in old "vxdisk list" outputs
Post this, you will need to initalize a blank diskgroup (with same name) for e.g
# vxdg init <dgname> disk1=<disk> disk2=<disk2>
Then using vxmake, you need to create each subdisk, plex & volume, this can become tedious if number of volumes are more & you have layered volumes.
These outputs explained above is also available in /etc/vx/cbr/bk/<dgname> directory. This is nothing but backup taken by vxconfigbackupd. This is easy way to restore DG config than doing a above explained bottom up approach. You can simply do a "vxconfigrestore" to get your entire dg back.
=====================================================================
if I use this vxprivutil I could generate an dump of private region,next I use vxassist.right?
the same I could recover with vxconfigrestore if I have an backup taken with vxconfigbackup.
>>> yes you can generate the dump of config however next you need to use vxmake to regenerate your DG.
# vxmake -g <diskgroup> -d <dump_vxprivutil> (double check the switches, I may be little wrong)
=====================================================================
In an article I saw that if I like to replace an faulty disk from an dg,and the volumeas are spread over more than one subdisk,then if I perform vxdg -k rmdisk-the data could be corrupted.instead if I use this vxdiskadm with replace disk option,then this script tries to move off the faulty disk the data.but if there is no space to move the plexes,then the recommendation is to remove some plexes from other logical volumes.My question is if it is safe to remove an plex from an logical volume,who is mirror volume.In this case until the faulty disk is replaced,this volume will use only one submirror.what if this submirror fails in mean time?
>> That is actually the purpose of mirror right ? if you loose one submirror, volume will still continue to run with 1 submirror. In the mean time, you can safely replace disk using vxdiskadm menu
G
