[rear-users] Thanks a lot for this great software

Andreas Tarp andreas.tarp at ericsson.com
Fri Dec 7 16:13:32 CET 2012


the nodes have two internal disks mirrored by hardware raid controller, so OS only sees one internal disk unless you have a look on the device with raid controller software (hpacucli tool).

Thanks a lot for the hint about possible installation option to use normal device names right from the beginning, but the nodes were installed first, than we had to start integrating rear ;-)

You are right, it sounds realy scary that ID names are also used inside of initrd, but luckily this did not cause any problems on our nodes. So seems like the names inside of initrd are not that often used (if at all) 


-----Original Message-----
From: rear-users-bounces at lists.relax-and-recover.org [mailto:rear-users-bounces at lists.relax-and-recover.org] On Behalf Of Johannes Meixner
Sent: Freitag, 7. Dezember 2012 15:13
To: Relax-and-Recover users list
Subject: Re: [rear-users] Thanks a lot for this great software


On Dec 7 13:08 Andreas Tarp wrote (excerpt):
> HP DL380 nodes running SLES11

Do those HP DL380 nodes have a built-in local harddisk that you use to install the operating system or are they diskless and/or do you perhaps even use SAN with multipath?

> we are on rear-1.13.0-45.git201208022252 version SLES standard 
> installation uses device IDs instead of normal device names.
> As our rear release was not able to scope with device IDs we changed 
> device names in /etc/fstab, /boot/grub/menu.lst and 
> /boot/grub/device.map back to normal device names.

Yes, you may have a look at

Off the top of my head I think you could somehow specify in the installation settings that a SLES installation uses traditional device nodes /dev/sd* but right now I don't find the documentation.

Right now a colleague told me that the persistent storage device names are also somehow stored in the initrd - scaring!
On my SLED11-SP2 workstation I can verify this:
# grep ' / ' /etc/fstab
/dev/disk/by-id/ata-ST3320418AS_9VM9JAXK-part1 / ...

# zcat /boot/initrd | strings | grep ST3320418AS ... fallback_rootdev='/dev/disk/by-id/ata-ST3320418AS_9VM9JAXK-part1'
... rootdev='/dev/disk/by-id/ata-ST3320418AS_9VM9JAXK-part1'
... resumedev='/dev/disk/by-id/ata-ST3320418AS_9VM9JAXK-part3'

Luckily it worked also for me to change only /etc/fstab, /boot/grub/menu.lst and /boot/grub/device.map to use traditional device nodes.

Currently I don't know for what the persistent storage device names are used in initrd.

To be really 100% correct I assume one would also need to re-generate the initrd so that everywhere traditional device nodes are used (hoping that initrd is the last file where it is stored).

Kind Regards
Johannes Meixner
SUSE LINUX Products GmbH -- Maxfeldstrasse 5 -- 90409 Nuernberg -- Germany HRB 16746 (AG Nuernberg) GF: Jeff Hawn, Jennifer Guild, Felix Imendoerffer _______________________________________________
rear-users mailing list
rear-users at lists.relax-and-recover.org

More information about the rear-users mailing list