System Recovery

Frank DiPrete fdiprete at comcast.net
Wed Jul 16 08:05:06 EDT 2008


It appears that /dev/sda is an 80G IDE drive(?) and /dev/sdb is a 300G 
SATA drive. Unfortunately having IDE drives go completely fubar even 
when they still have that 'new drive smell' is completely normal these 
days. I send love notes to seagate in the form of rma's frequently.

cd rescue mode is loading drive controllers (the pix driver is for the 
sata drive and possibly the ide drive depending on the chipset) but it's 
probably not the root cause of the problem.

there is a prompt during rescue boot to try and detect partitions and 
mount them on /mnt/sysimage. Is this when the system reboots itself?

To narrow this down, disconnect the failed drive /dev/sda and boot the 
rescue cd again. if it does not reboot itself, theres the problem.

Second test is to reconnect the drive, boot in rescue mode, but *do not* 
try and detect the partitions and mount them on /mnt/sysimage

If this fails too, then the drive is really fubar.

If this is the case then there is little point in trying to get your 
root (/) partition info off this disk.

However if you are able to boot the rescue environment without mounting 
the partitions there may be some hope. lvm complicates matters somewhat.

If you can boot it into rescue mode with the drive attached but *not 
mounted* you can try:

pvscan
vgscan
lvchange -ay
lvscan

then try to fsck that root partition on lvm
fsck /dev/VolGroup00/LogVol00

btw - k3b may use temp space on var by default to stage and create its 
iso image. (I'd have to check) so it may have done nothing wrong but try 
to write to a crappy drive.

On a side note, I know that fedora/rh uses lvm for everything as the 
default install, but I stopped using lvm for the /(root) partition 
because it make recovery more difficult.




Labitt, Bruce wrote:
> How many physical hard disk drives are part of this system?  
> [Labitt, Bruce] Two.  The first drive 80GB has the system on it.
> Are there any hardware RAID controllers or such in use?[Labitt, Bruce]
> no.
> 
>> It appears I can access /data and its contents.
> 
>   Appearances can be deceiving.  Have you run fsck against
> /dev/lvol1/data yet?  
> 
> [Labitt, Bruce] no.  actually there is no data worth anything there.  
> 
> (Even that is far from perfect; if something
> scrambled just the contents of file data blocks but left the
> filesystem metadata structures intact, fsck won't see anything wrong.
> But best to start somewhere.)
> 
>> How do I figure out what drive has what on it?
>> sda, sdb, etc.
> 
>   "fdisk -l /dev/sda" will show the partition table on the "sda" disk.
>  Run that command as needed for all the disks in the system.
> [Labitt, Bruce] I'll try that, once I get thru the 'boot' sequence
> again...
> 
> sda 80GB: 255 heads, 63 sectors/track, 9726 cylinders
> Device	boot	Start		End	Blocks	Id	System
> /dev/sda1	*	1		13	104391	83	Linux
> /dev/sda2		14		9726	78019672+	8e
> Linux LVM
> 
> sdb 300GB: 255 heads, 63 sectors/track, 36481 cylinders
> /dev/sdb1		1		36481	293033601	83
> Linux
> 
> 
> 
>   "sda" will be the first SCSI, SATA, or other non-IDE device.  The
> second would be "sdb", the third "sdc", and so on.  IDE disks are
> "hda", "hdb", etc., instead.
> 
>   "pvs" and "lvs" will show you the physical and logical volumes LVM
> knows about.
> 
> [Labitt, Bruce] response for both commands
> /var/lock: mkdir failed: Not a directory
> Locking type 1 initialization failed.
> 
> -- Ben
> _______________________________________________
> gnhlug-discuss mailing list
> gnhlug-discuss at mail.gnhlug.org
> http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
> 
> _______________________________________________
> gnhlug-discuss mailing list
> gnhlug-discuss at mail.gnhlug.org
> http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
> 
> 


More information about the gnhlug-discuss mailing list