Hi, All you need to do is copy the etc-directory from the root volume to your other volume,
Then if your root-volume fails, you floppy-boot and select your other volume as root volume, reboot and all is well...
Now if you have other stuff on your root volume besides etc, you will still need to restore it from backups.
---- Mats
-----Ursprungligt meddelande----- Från: Deepak Soneji [mailto:sonejideepak@hotmail.com] Skickat: den 8 november 2002 01:03 Till: devnull@adc.idt.com; toasters@mathworks.com Ämne: Re: Root/Boot Disk(Volume?) Safety
I was wondering if we can snapmirror vol0 of filer1 to filer2 and viceversa! or maybe use ndmpcopy across the filers. Thanks /Deepak
----- Original Message ----- From: devnull@adc.idt.com To: toasters@mathworks.com Sent: Thursday, November 07, 2002 2:16 PM Subject: Root/Boot Disk(Volume?) Safety
Dear Toasters,
I have 2 filers, a F810 and a F740. We also have some Sun servers that do NIS, sendmail, web server etc.
To protect from root/boot disk failures on my Sun, i either use rsync to copy the root disk to another machine with the same hardware architecture
or
mirror the disk to a "backup" disk on the same machine.
I have 2 volumes on my F810 vol0(which has 2 disks) and vol1(11 disks) on a DS-14 shelf.
I am trying to protect myself from the situation where i lose both my vol0 disks, which will mean that my filer will no longer know anything about itself and i cant really "connect" from my backup server to the netapp to recover everything.
If i lose 2 disks on vol1, i will lose all data, but i can slap 2 new disks on the filer and still recover all data from my backups.
Any recommendations, thoughts on the same ?
I use Legato for backup and do my backups over NFS, by mounting the volumes to my backup server.
/dev/null
devnull@adc.idt.com
mats.oberg@tietoenator.com writes:
Hi, All you need to do is copy the etc-directory from the root volume to your other volume,
Then if your root-volume fails, you floppy-boot and select your other volume as root volume, reboot and all is well...
I wonder whether such copying is entirely safe. In some releases of ONTAP certain actions cause /etc on a non-root volume to be created and files put there, the overwriting of which by the proposed copying might have evil consequences.
For example, at one time (pre-6.0, I think) turning quotas on for the non-root volume did this, creating files in /etc/db there. And more recently taking a backup dump would create files in /etc/tmp and in /etc/oldmaps/bkp. I don't know of anything that mucks with /etc on a non-root volume in 6.2 or later, though.
One could copy to /etc_safety_copy on the non-root volume instead, of course, and then use the not-so-secret "mv" command to rename it as /etc as part of the recovery process.
Chris Thompson Email: cet1@cam.ac.uk
I wonder whether such copying is entirely safe. In some releases of ONTAP certain actions cause /etc on a non-root volume to be created and files put there, the overwriting of which by the proposed copying might have evil consequences.
Thanks for the tip.
I just use rsync to "sync" between /vol/vol0/etc/ and /vol/vol1/safe_etc/
Has anyone made Boot floppies(ontap) on Solaris/Linux.
I dont know how to fit more than 1.45 Mb on a 3.5" floppy.
Ontap uses 4 disks to boot.
3 are of size 1.47M and one of size 0.76M.
Any help is appreciated.
Thanks, Ram.
This is not so easy. We used to do this and after an upgrade, it crashed every time we ran rsync to copy from root volume /etc to other volume /etc for a backup root volume. There are certain files in /etc that cannot be completely overwritten, even if it is on a non-root volume. Haven't heard if this is changed. Don't remember bug number.
At 10:03 AM -0500 11/12/02, devnull@adc.idt.com wrote:
I wonder whether such copying is entirely safe. In some releases of ONTAP certain actions cause /etc on a non-root volume to be created and files put there, the overwriting of which by the proposed copying might have evil consequences.
Thanks for the tip.
I just use rsync to "sync" between /vol/vol0/etc/ and /vol/vol1/safe_etc/
Has anyone made Boot floppies(ontap) on Solaris/Linux.
I dont know how to fit more than 1.45 Mb on a 3.5" floppy.
Ontap uses 4 disks to boot.
3 are of size 1.47M and one of size 0.76M.
Any help is appreciated.
Thanks, Ram.
devnull@adc.idt.com asks [...]
Has anyone made Boot floppies(ontap) on Solaris/Linux.
On Solaris ... often. (Well, not so often as I used to, as it's _so_ tedious to do a real floppy boot to test the results these days. But I try to make a new floppy set at least for each n.n release - e.g. 6.0, 6.1, 6.2)
I dont know how to fit more than 1.45 Mb on a 3.5" floppy.
If you did, the filer wouldn't know how to read them!
Ontap uses 4 disks to boot.
It's up to 5 now.
3 are of size 1.47M and one of size 0.76M.
Any help is appreciated.
Well, suppose you have a Solaris system with a floppy drive attached, and assume the volume manager is running.
You will have the NNN_boot_floppy_M.X files (NNN = ONTAP release, M = 1 to 5, X = i or a) downloaded from now.netapp.com and have verified their checksums.
Insert a new or reusable floppy in the drive. Use "volcheck" to make the volume manager notice it. If necesary, low-level format it using the "fdformat" command.
Now do "dd if=NNN_boot_floppy_M.X of=/vol/dev/diskette0/unlabelled bs=18k". If you have multiple floppy drives it might not be diskette0. It might not be called "unlabelled" if you are re-using an old floppy: I tend to use filename completion in the shell at that point. :-) The block size doesn't matter all that much: the copying will be pretty slow regardless. DO NOT USE "rdiskette0" instead of "diskette0", though: that will change the low-level formatting and you'll have to start over.
Now use "eject" to get your floppy disk out. Fill in the pretty coloured label. Now repeat the process for the remaining floppies ... :-(
Now schedule enough downtime on your filer to test what you've just made. Of course, it helps to have a spare filer (with the right architecture) at this point!
Chris Thompson Email: cet1@cam.ac.uk