Sign in with
Sign up | Sign in
Your question

Migrating Buffalo TeraStation Live RAID cluster to Linux Server

Last response: in Storage
Share
November 23, 2012 3:13:30 PM

I've got two old Buffalo TeraStation Live NAS servers (4x1TB HD and RAID5 config) and I need to migrate all the data (they are both filled) off to a new Linux server that I just purchased with a lot more storage.

My issue, as many know, is that the Buffalo NAS boxes are horrifically slow for transferring data and simply mounting the two NAS boxes on my Linux server and transferring the files will take about 4 days. Has anybody found a faster way?

One idea was to simply pull the drives out of the Buffalo NAS and mount them in four open bays on my Linux server, and then construct the RAID cluster on the server and transfer it directly. Has anyone done this? It would certainly be much faster, and since I believe the Buffalo is only doing software RAID, it should be possible to do on the Linux server.

Any thoughts / comments / pointers to get me started? I'll happily write up a "how to" at the end of this project!
November 23, 2012 10:12:34 PM

Well, It was much easier than I thought and my Buffalo TeraServer Live RAID array is now up and running (and copying files) infinitely faster on my Linux server. For anybody else out there struggling with the migration of an old Buffalo TeraServer Live, this is a great way to do a migration since your effective throughput goes from about 100Mbps to about 1.5Gbps. Just putting the transfer speed in context of bps for those struggling with the fact that the 1Gbps ethernet in the back of the TeraServer is useless when you factor in how underpowered the CPU is on this box. It can't achieve 1/10th the throughput of the Gb ethernet interface. Here is what I did:

- removed all 4 drives from the TeraServer, being careful to write the slot number on the drive for re-installation back on the TeraServer later
- installed the 4 drives (in order just to make it easier) in 4 open bays of my SuperMicro CentOS 6.3 server [/dev/sdc /dev/sdd /dev/sde /dev/sdf]
- used the command "mdadm -E /dev/sdc1" and cycled through partitions until I found that the RAID5 array was on /dev/sdc6 and noted the UUID string
- edited the /etc/mdadm.conf as follows:

DEVICE /dev/sdc /dev/sdd /dev/sde /dev/sdf
ARRAY /dev/md0 metadata=0.90 UUID=bbeaa0b8:9768147b:36258eb6:6d7e74f0

I rebooted for good measure (possibly not necessary) and then mounted /dev/md0 and it was all there. Be careful to not mess with other partitions on these drives if you want your TeraServer to function later when you put the drives back. How do I mark this "SOLVED"?
m
0
l
November 23, 2012 11:53:06 PM

WARNING!!! I'm not entirely certain why, but upon returning the drives to the Buffalo unit it no longer recognizes the RAID5 array although all I did was remove the drives, put them into my CentOS server and mount the array, copy the files over, and then return everything back to the Buffalo box. While the unit is still manageable, it claims that all the disks are "unformatted". I'm fine since I had everything copied off the system, but a huge caution to anyone who tries this and wants to ensure that the RAID array still works when returned... it appears that it does NOT. Anybody have an idea as to why this may be the case?
m
0
l
!