In our env, we have around 100 more solaris 8,9,10 servers,
we quaterly apply patches & emergency fixes, for this we take prior OS Ufsdump bkp,
In order to reduce the efforts to go DC & load each time the tapes on all servers & take the ufsdump backup, we planned to set up a central server, where we can collect the OS images & store & re-use in case of system crash.(Eg: Like NIM server in AIX)
The main point around to setup is to :
Take all the ufsdump to disk instead of tapes.
Load patches in central server & patch all the clients from the central server.
**In case system crash can we recover the OS image having it on central server by booting via LAN ? can we recover like how tape restore?
You can backup all your filesystems using ufsdump to a remote disk over NFS. (You should use "fssnap" before ufsdump on each filesystem to snapshot.)
If your system crashes and won't boot then you can boot from CDROM and mount your hard disk root filesystem on /a. You can, if you need, then "newfs" your root slice ready to take a clean restore.
Use the ifconfig utility to plumb and configure a network interface so that you can ping the remote box where your root ufsdump is sitting.
Then mount the remote filesystem containing your ufsdump on /mnt.
Change directory to /a and ufsrestore your dump file of the root.
Job done!!
Of course, you may need to restore other filesystems in the same way (eg /var) before your system will boot depending on how you installed everything in the first place.
As far as patching is concerned you could store patchsets on your central box but you should ftp them to each server for patching because it is strongly recommended that patching is done in single user.
Just post if you need any of the above expanded.
---------- Post updated at 10:07 AM ---------- Previous update was at 10:04 AM ----------
Another route would be to use flarcreate to create a flash on your cental server.
To recover boot from CDROM and do a "flash" install.