Recovering Software Raid Array

[glennda]

As some may have seen from another thread I have a Snap server which has died, all the disks are present and correct and boot in a alternate machine fine. The only problem with this is that the OS the snap server ran (guardian OS) has no desktop shell or command line from the console. But i cannot access this machine from a web browser or remote shell as the console tells me.

 

I think i will need to try and recreate the software raid array on the machine using a live CD - anybody got any tips/good instructions??

 

Toby

 

EDIT: if a Mod can edit the title to Recover Software raid array please!

Edited November 27, 2012 by glennda

[dhicks]

I think i will need to try and recreate the software raid array on the machine using a live CD - anybody got any tips/good instructions??

 

In the past when I've done this using System Rescue CD, MD arrays have simply been recreated on boot - simply make sure all disks are connected, boot from CD and see what's in /dev/md*. This assumes your snap server was Linux based, of course, and used MD arrays.

[glennda]

In the past when I've done this using System Rescue CD, MD arrays have simply been recreated on boot - simply make sure all disks are connected, boot from CD and see what's in /dev/md*. This assumes your snap server was Linux based, of course, and used MD arrays.

 

Will try that currently trying to use a ubuntu live cd and it just says all the drives are locked but can see the MD array running mdadm --assemble. will give the system rescue a go.

 

Guardian OS seems to be linux as far as i know and wikipedia describes it as "The GuardianOS is a UNIX-like operating system based on the mainstream Linux kernel" Snap Server - Wikipedia, the free encyclopedia

[dhicks]

Will try that currently trying to use a ubuntu live cd and it just says all the drives are locked but can see the MD array running mdadm --assemble.

 

After reassembly, I think mdadm sets arrays to be read-only by default (you might want to keep it read-ony while you tinker with it, of course). "cat /proc/mdstat" should show you your RAID array's status (whether it thinks it is functioning, whether it is rebuilding).

 

What are you expecting to see when you have an assembled MD array? A file serving OS could well be using LVM on top of RAID, in which case you won't see a handily mountable file system in /dev/mdx, any file systems with be in LVM volumes within that array.

[glennda]

After reassembly, I think mdadm sets arrays to be read-only by default (you might want to keep it read-ony while you tinker with it, of course). "cat /proc/mdstat" should show you your RAID array's status (whether it thinks it is functioning, whether it is rebuilding).

 

What are you expecting to see when you have an assembled MD array? A file serving OS could well be using LVM on top of RAID, in which case you won't see a handily mountable file system in /dev/mdx, any file systems with be in LVM volumes within that array.

 

appreciate that - trying to see anything I can, its a device somebody has come to us with and asked if we can recover.

 

System rescue cd states the following on boot "mdadm failed to restore critical section for reshape" Sorry

 

Any thing else I can try?

[dhicks]

System rescue cd states the following on boot "mdadm failed to restore critical section for reshape"

 

A quick Google suggests mdadm might be getting confused between whole devices and partitions - i.e. the difference between /dev/sda1 and /dev/sda. To my mind, this rather implies something's overwritten a partition table somewhere along the way on one or more harddrives. A partition table is, of course, just a pretty simple set of data saying how the harddrive should be split up - if there isn't a backup of it available (worth Googling to see what Guardian OS does with partition information) you can try and reconstruct it by hand - each hardrive will probably be one simple, large partition.

[glennda]

A quick Google suggests mdadm might be getting confused between whole devices and partitions - i.e. the difference between /dev/sda1 and /dev/sda. To my mind, this rather implies something's overwritten a partition table somewhere along the way on one or more harddrives. A partition table is, of course, just a pretty simple set of data saying how the harddrive should be split up - if there isn't a backup of it available (worth Googling to see what Guardian OS does with partition information) you can try and reconstruct it by hand - each hardrive will probably be one simple, large partition.

 

Yeah not sure I want to play around with it too much incase I risk the data. There appears to be a few companies online who specilise in getting data from these devices so might have to send it off to them.