A Buffalo LinkStation went missing from the office network without warning. It was no longer visible as a share through “This PC” and was increasingly reluctant to start up at all. Yet it did respond to a ping, which is a lot more helpful information than it seems.
This is a scenario, not a case study. It illustrates a common failure type that we address on a regular basis, and how it is approached, written from the technical side (versus as part of a single customer project). Please visit our documented case files here.
When the unit is powered on, the indicators look good. And the unit responds to a ping. Also, the web interface may even load. However, there are no shares. Or if they are available, they have no files in them. And the unit indicates that there is either no volume or a volume that is not mounted.
The majority of NAS devices run their OS on a small partitioned area, separate from the data volume, that has been duplicated on each member disk. Therefore, even if the data array is completely broken, the device will still boot. The web interface will be functional. And it will continue to answer on the network (ping), because it is booted off a partition that is functioning properly.
Thus, when you have a responsive NAS with no shares, you know something useful. Specifically, the hardware is working. Your network is working. And the issue is strictly related to the data volume. That is significantly easier to work with than a non-responsive unit.
It also indicates the unit has its diagnostic processes successfully running, and therefore it will confidently offer to repair the issues.
That the unit is functional. In order to answer a ping, the unit needs to boot, load the operating system and initialise networking. The fault is located in the storage layer below (either the array will not assemble or the volume will not mount), but typically your data remains intact.
Often, it is going to be the same problem. Many units will check and attempt to mount their volume during startup, and if they get hung up on a section that is unreadable, they could become stalled for such a length of time that they would appear to not start. Typically, the unit is starting and waiting.
No. Underneath the interface it's all traditional Linux storage (mdadm) with a standard file system, so after the disks have been imaged, the appliance and its firmware will have no bearing on how it is rebuilt.
No. Bad sectors in one drive mean every time you start up the unit you'll be going through the exact same sector that isn't readable, and possibly writing to the array. Just shut it off and leave it.
Yes. For instance, in this situation, a small part of the drive was affected (the file system metadata) and it prevented a volume from being mounted, while the vast majority of the disk read perfectly. It is position, more than size, which determines whether a volume will mount.
From £500 plus VAT (written costs agreed after a 48-hour no-charge diagnostic). Please send us the disks but keep the enclosure; we would appreciate it if you could label each disk according to its position (for example bay #1, bay #2, bay #3 etc.) and make sure you number each disk as it comes out.
A member dropped and the array is degraded past its tolerance. This means two disks out of a RAID 5 configuration or one disk out of a striped set. Typically the disks themselves are fine but the RAID will simply refuse to assemble.
File system damage on an otherwise healthy array. The array can be assembled but there are issues with the ext4 or Btrfs file system that sits on top of it, and this corruption frequently happens after an unclean shutdown.
Metadata damage in a layer between the two. This could include a Logical Volume Manager (LVM), thin provisioning or even the vendor’s own volume management tool. In all cases, the storage array is “good” and the file systems themselves have no errors; however, there is an issue with how they are mapped together.
The purpose of the diagnostic is to figure out which of these three things happened, since each has a different path to recovery.