Rebuilding a NAS appliance RAID array Print

Author: Simon Bisson
Category: Server

Disks fail, and RAID arrays need rebuilding. Be prepared for the worst – and keep your clients’ data safe.

Hard disks don’t last forever. The bathtub curve gets you every time, with old drives wearing out and losing data. That’s why RAID systems are a good idea, with RAID 1 mirroring information between two disks and RAID 5 using an array of three or more disks to create a recoverable disk array that can rebuild after any one disk in the array fails.

Business-grade Network Attached Storage (NAS) appliances form the backbones of many small office networks, providing a central file and print service where there’s no need to invest in a full-scale server. They also provide additional storage capacity which can reduce the load on an SBS server. Many of these appliances are embedded Linux systems using consumer disks and running them with a workload very different from their original design specifications. Disks will fail, and you will have to rebuild the array.

Use the Web interface on a NAS appliance to manage RAID arrays.
Use the Web interface on a NAS appliance to manage RAID arrays.
Rebuilding a NAS array is a relatively simple (if time-consuming) task. You’ll need to have a disk of the same size as the failed disk. Some NAS arrays support using larger disks as a path to increasing the available storage, so it’s worth checking the documentation. If possible, you should have appropriate disks in stock to reduce client downtime – if you don’t know what’s in the NAS use the hardware diagnostics and think about buying spares when you populate a bare NAS carcass. Once the disk has been replaced, use the NAS appliance’s user interface to rebuild the array. The process is relatively simple, and most systems will prepare the drive and copy data automatically. There can be an issue with early versions of the Buffalo TeraStation Pro firmware. The system flash memory on these devices only contains a boot loader – most of the OS is held as a partition on the RAID array. In early OS releases a disk failure prevented the OS from booting and a disk replacement meant either purchasing a preformatted disk from Buffalo or installing new firmware. To avoid similar problems, make sure that all your clients’ arrays are running the latest software releases.

A 500GB rebuild can take more than eight hours – so be prepared for a late night with plenty of coffee while you watch the progress bars and refresh your browser, or make sure the customer knows the drive won’t be available until you’ve tested it the next morning. 


Show other articles by this author