• July 14, 2020, 12:51:27 AM
  • Welcome, Guest
Please login or register.

Login with username, password and session length
Advanced search  

News:

This Forum Beta is ONLY for registered owners of D-Link products in the USA for which we have created boards at this time.

Author Topic: Harddrive Failure, need suggestions  (Read 3949 times)

3kramd5

  • Level 1 Member
  • *
  • Posts: 5
Harddrive Failure, need suggestions
« on: January 05, 2013, 10:39:27 AM »

So, at least one of the drives in my -323 has failed. I believe it was the primary (in the right). I've been using the same seagate 750GB drives in a RAID1 array for years.

A few days ago, it hung while I was trying to open a file. I turned it on and off, and the right LED was glowing purple. I turned it off for a while, and when I turned it back on, the drive LEDs were both solid blue but the power LED just blinks constantly (never finishes booting up, I guess).

If I pull out the right hand drive, it will boot. However, I don't seem to have access to the left drive. I tried putting the left drive into the right bay, and it still doesn't seem to see it (the easy search tool doesn't list any drives, and when I navigate to it via windows explorer (//192.168.0.170/volume_1) it's a dead end.

I would be surprised if both drives up and failed at the same time. Any ideas about how to recover my data? If one drive failing = total loss, then RAID1 is pretty silly. I've tried reading the drives in windows with DiskInternals Linux Reader, but am unable to see anything on them.
Logged

jhtopping

  • Level 2 Member
  • **
  • Posts: 73
Re: Harddrive Failure, need suggestions
« Reply #1 on: January 05, 2013, 11:11:49 AM »

I would start with the "Sticky Topics", like Data Recovery (Windows PCs) and then Data Backup Versus Redundancy.
Logged

3kramd5

  • Level 1 Member
  • *
  • Posts: 5
Re: Harddrive Failure, need suggestions
« Reply #2 on: January 05, 2013, 11:54:57 AM »

I have been.

NAS Data Recovery can see data on the drive which prevents the 323 from booting up, however it doesn't even recognize the drive that the 323 has no problem with.

I guess best case scenario I can get my data by paying 100 dollars for software. But really I'm more concerned with make sense of this.

The failure doesn't make sense. Why does a "good" drive (readable with the software tool) hang the dlink hardware, but a "bad drive" (not readable with the software tool) poses no problem from the dlink hardware?

Assuming both drives aren't bad, is there no way to use the NAS-323 hardware itself to read data from a single drive from a mirrored array? If not, what's the point of RAID1 on the NAS-323?
Logged

ivan

  • Level 8 Member
  • ***
  • Posts: 1475
Re: Harddrive Failure, need suggestions
« Reply #3 on: January 07, 2013, 07:15:54 AM »

OK, so your setup was RAID 1 therefore you should be able to get at your data on the good drive - assuming you don't have a working backup.

Have you tried using a USB caddy or adapter to attach your good drive to a PC and, if you are using windows, an installable file system that will allow you to read EXT2/3 file systems.

If, using that setup, you don't see anything on the good drive then you may not have been using RAID 1 but RAID 0, in which case everything is gone.

If you are sure that it was RAID 1 then you could try replacing the bad drive and rebooting to see if your DNS-323 will try and rebuild the array.  This assumes there isn't a defect with the good drive that prevents the firmware of the NAS seeing it as part of a degraded RAID 1 array.

Just as an aside, we have used the USB adapter route to show clients that, as a last ditch effort assuming no physical disk problems, we can recover much of their data and it costs, especially if they don't have a backup.  The only way we use RAID 1 as a backup medium is when we are backing up one RAID 1 to another which is our standard setup.

Unfortunately, I can't answer your second last question because we have never tried.  The only time we had a disk go bad on us we just replaced it and the NAS rebuilt the array without problems.
Logged

3kramd5

  • Level 1 Member
  • *
  • Posts: 5
Re: Harddrive Failure, need suggestions
« Reply #4 on: January 09, 2013, 07:22:59 PM »

Hey Ivan -

I am certain it was RAID1. I've had no luck in Windows short of Runtime's software (NAS Data Recovery), the demo of which shows the file structure and files. I'm trying other methods to avoid buying a software solution.

I haven't found a compatible EXT file system for windows7/64.

I did boot into Ubuntu and was able to assemble the drive (it recognized it as one of two from a mirrored array), but so far have failed to mount that assembly. I suspect it's due to my severe inexperience with Linux command line stuff, though.

I put in a different known good drive and it said it would format it. However it only recognized the new drive.

Fortunately, there was nothing time critical on the drive. It's mostly my music collection (which I also have on two cloud services) and some financial documents that I can with a little leg work re-download from my various banks. I was just hoping that it would be a little easier.

Thanks for the help!
Logged

ojosch

  • Level 1 Member
  • *
  • Posts: 18
Re: Harddrive Failure, need suggestions
« Reply #5 on: January 11, 2013, 06:43:10 PM »

BTW, RAID is a false sense of security. I work for a major enterprise storage company which manufacturers NAS / SAN storage appliances, and it is well known in the industry that RAID basically is designed primarily to allow the appliance to continue to serve data after an array suffers from a single drive failure (or two-drive failure in a raid dual-parity array). Then, if the drive has failed and is causing a loop or system instability, then user is instructed to pull out the faulty drive, and the degraded array should be able to continue to serve data, and then a replacement drive should be installed as soon as possible, as the array is then functioning in a degraded mode (if there are low spare disks available in the system). I speak of a full size enterprise appliance in that example..

Even with RAID, you should always have a DR system (disaster recovery) which you make full and incremental backups to so that in the event that you loose an entire raid group then recovery is possible. This will also offer better protection in the event that files become infected by viruses or data becomes corrupted for example (which RAID has no chance at protecting).

For the above listed reasons, I like to operate my DNS-323 under JBOD mode (not raid, but individual disk mode), and I use rsync (through cron-jobs, scheduled tasks) to copy all my data from my linux and windows servers to the DNS-323 separately. I make it copy to one drive, then copy to the other one on a different schedule, so if I get a virus, there is a delay before the new data reaches the second drive. I only use my 323 for backup only (so my data is actually copied to 3 locations). I also copy crucial files to a cloud storage provider in case my house burns down.
 

In your setup where you may be using the 323 as a primary NAS storage device, during normal operation you could configure it to JBOD mode and just use rsync to copy the data from one drive to the other.

But for now, see if unit will boot up with just the one drive. If not, then get another 750GB drive, install it and see if it boots. If not, then pull the good drive, buy a USB to sata adapter (maybe $14 bucks) so you can plug it into a PC which can read ext 2/3 filesystem like a Linux machine (or just use a Linux tool cd to boot from) where you can mount the drive and pull the data to another location.

And as a last note, any time you integrate some form of backup and recovery system, you should always test the recovery strategy before trusting your data onto it.

It amazes me still how many Tier 1 enterprise companies do not test recovery strategies before deploying their systems and then they call us when they have corrupted the data in a LUN and want us to try and recover what cannot be recovered, since they had been taking inconsistent snapshots the whole time (w/o quiescing the DB first) or some similar scenario like that. Had they tested it first, they would've learned that they needed to stop I/O on the DB so it would be in a consistent state prior to taking a volume snapshot is the example there... Simple testing would've uncovered that.

The Tier 3 or 4 enterprise environments always test every possible failure scenario in a dev system, and documenting the recovery procedure there before ever deploying it to the actual production environment. Now I am not saying you should go that far necessarily, but in the future you may want to test a few scenarios with some test files loaded onto the drive or raid array (like pulling drives and seeing how easy it is to access the data in the event of a failure).

But as a last note, my opinion is you should always have your data copied to another physical device at all times so if the device completely fails your not ever dead in the water (no single point of failure).

Large companies like banks, investment firms, government office, etc, ALWAYS have their data protected with snapshots, local raid group mirrors, metrocluster sync mirroring, DR site snapshots, backup to tape archives, etc, with up-to-the-minute restore capability because it means a lot to them that they cannot forget what your balance is, how much interest you earned, or how much tax you owe.

Your family pictures and certain things like that cannot be replaced either so you may want to continue looking for a reliable backup / recovery strategy that is safer than just using RAID by itself..
« Last Edit: January 13, 2013, 07:53:13 PM by ojosch »
Logged

3kramd5

  • Level 1 Member
  • *
  • Posts: 5
Re: Harddrive Failure, need suggestions
« Reply #6 on: January 11, 2013, 07:41:10 PM »


For the above listed reasons, I like to operate my DNS-323 under JBOD mode (not raid, but individual disk mode), and I use rsync (through cron-jobs, scheduled tasks) to copy all my data from my linux and windows servers to the DNS-323 separately. I make it copy to one drive, then copy to the other one on a different schedule, so if I get a virus, there is a delay before the new data reaches the second drive.

I like that idea. If when this is all settled my DNS box is still functioning, I'll switch to JBOD.

As to your linux suggestion, I've tried but think I'm coming up short in the knowhow category. After consulting google, I ran

mdadm --assemble --run /dev/md0 /dev/sdd2

Per lsblk, that resulted in an md0 child under sdd with type raid1:
sdd 8:48 0 698.7G 0 disk
├─sdd1 8:49 0 517.7M 0 part [SWAP]
└─sdd2 8:50 0 697G 0 part
└─md0 9:0 0 697G 0 raid1

Next, I made a temporary directory and tried to mount md0

mkdir /tmp/sdd-mount && sudo mount /dev/md0 /tmp/sdd-mount

To which it says "/dev/md0: can't read superblock"

Any idea what that means?
« Last Edit: January 11, 2013, 07:44:07 PM by 3kramd5 »
Logged

JavaLawyer

  • Poweruser
  • Level 15 Member
  • *
  • Posts: 12190
  • D-Link Global Forum Moderator
    • FoundFootageCritic
Re: Harddrive Failure, need suggestions
« Reply #7 on: January 11, 2013, 07:43:11 PM »

Echoing ojosch's key points, please refer to the following sticky post: DNS-323 - Data Backup vs. Redundancy
I am thinking of renaming that post "DNS-323 - RAID is Not a Backup".  :-\
Logged
Find answers here: D-Link ShareCenter FAQ I D-Link Network Camera FAQ
There's no such thing as too many backups FFC

3kramd5

  • Level 1 Member
  • *
  • Posts: 5
Re: Harddrive Failure, need suggestions
« Reply #8 on: January 11, 2013, 09:07:34 PM »

Java-
After giving up on Linux, I installed vista on a spare HDD and the ETX file system adding from your sticky seems to be working. Thanks.
Logged