Devices · array sets & servers
The actual end of an array is almost never caused by the first failure, but rather by what happens after that: a rebuild driving a tired survivor through one long read until it stops, an ejected member forced back into service and/or disks moved between bays to see what happens. Even mirrors do not provide safe harbour, as each half of the mirrored pair has been running for equal amounts of time since they were installed. There is no counter in Maidenhead, and therefore most of the arrays we receive from there arrive via tracked postal delivery. Please call us (or complete our contact form) to obtain a quote before you send the disks. Once you've received a quote, remove the disks from the enclosure, label each disk with the number of the bay it was located in, insure the package and post the disks to our Guildford intake lab using a tracked service, or simply bring them there yourself (Monday to Friday, 9am to 5:30pm; no appointment is required). Please be assured that, since we will ultimately be working solely with images of your original disks, there is no additional loss or damage once your disks arrive.
Describe the fault to an engineer
0800 6890668
| The symptom | The likely cause | Start here |
|---|---|---|
| Foreign Configuration Found (Dell PERC) | The metadata stored on each member contradicts what the card itself stores | Don't touch either option; clone the disks first |
| 1786 — Drive Array Recovery Needed (HP) | It has no cover left; it is either waiting to be rebuilt or is in the middle of being rebuilt | Image it after you shut it down |
| 1784 — Drive Array Drive Failure (HP) | One of the drives has failed as part of a RAID set | Do not choose your new drive based on an educated guess |
| 1788 — Drive Array Reports Incorrect Drive Replacement (HP) | The disks have been returned to the wrong bays | Everything is determined by sequence; stop |
| Virtual Drive is Degraded / Offline | The volume has lost its cover, or has gone missing completely | Shut down the whole group |
| 1720 — SMART drive detects imminent failure (HP) | A functioning disk tells you it will go dead soon | Make an image today |
Nothing is chargeable until you say yes. Every item gets a case number of its own immediately upon receipt, an engineer determines what has failed, and you get a clear picture of which files have a real possibility of being recovered. The single quote that follows is fixed, written down and does not cost anything.
Every member is read onto a separate piece of imaging hardware (including any your card had already rejected), so when we do all the other things to make this array operational, they happen to copies, and at no stage is there a write operation performed on the disks you submitted for imaging.
We extract from the data structure: stripe width, member order, and how parity changes as we follow a path around the array. Then we create the array using the images we created on our own systems. In doing so, your card is used for nothing, and we do not perform a rebuild during the process either.
After the array is constructed, we repair in succession: the file system, virtual machine containers and databases. Finally, after each item is repaired, we verify them against a full list before we release anything.
You don't get charged anything until you've confirmed and authorised that everything is fine after seeing the entire list of what was recovered. Any recovered items are returned via brand-new storage media and the postage costs for these are covered by us. We do not consider a project complete here until we receive confirmation that you have been able to successfully access and review every item you received back.
Why single parity got a bad name: consumer SATA specs allow for approximately one unreadable sector per every 1014 bits read. That is roughly equivalent to one in every 12.5TB. When you ask a large single-parity array to rebuild, each surviving member has to provide more than that continuously until the rebuild completes. That is when rebuilds fall apart. The number above is quoted by the manufacturers themselves rather than being generated by an engineer. However, there continues to be disagreement over just how severely this will hurt in practice.
Production came to a stop after a Friday rebuild lost a second member and thus locked away the production data. Every disk was imaged individually. Then using the clean reads from three of the disks, and any good reads from the fourth, we were able to reassemble the stripe, and get back online before Monday morning.
First obtain a quote, then send the entire set of disks, each clearly labelled with the bay it was removed from. Often a disk that has been declared "dead" contains the most recent version of some stripes, and every block will be drawn from the image that returned it most cleanly.
No. The three attributes that define the layout of an array are readable within the structure of the data itself, so working them out is routine analysis; you do not have to use guesswork (trial and error) to determine the stripe width, how the members are ordered, or how parity is rotated.
Yes. Once a functional set has been created, the next step is to remove the VMDK and VHDX containers and the database store and open each on a bench to verify function rather than just logging it as present.
Typically four to seven working days for a set of multiple disks. If trade has genuinely ceased, the work can be run continuously; let us know about the deadline when you contact us for your quote, so we may schedule our work accordingly. You can either post your disks or take them to our Guildford intake lab (Monday to Friday, 9am to 5:30pm, no appointment required).
Every additional power-up asks for even more of a drive that is already starting to fail. First open a case with us, and let the free diagnosis tell you how much remains.