Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[vreid47362]

I have noticed that my RAID 1 array is now very susceptible to entering the degraded state when power is suddenly removed from the DNS-323 (i.e. a power failure, etc).  Once it enters the degraded state, my DNS-323 requires the re-introduction of freshly wiped drive to format it and re-sync the array.  Each time this happens, the drive affected is the drive on the right side of the unit, when facing it from the front side.

I have tried multiple Western Digital and Seagate drives and they all seem to exhibit the same behavior.   If I manually wipe the formatting from the drives and re-insert them, the DNS-323 will find them, format them, and start using the drives again.  In addition, I have confirmed that the drives are good when they degrade (I'm able to mount them and read/write from Ubuntu).

I believe that there is a problem with the RAID 1 implementation of the DNS-323 with the 1.04 firmware.

To fix this, I would like to see the re-introduction of ext3 or reiserfs or some other journaling file system and some more hard drive management options on the web gui for the next firmware release.

[vreid47362]

I have also noticed the following behavior when I have a degraded RAID 1 on the DNS-323 with firmware 1.04:

If I downgrade to firmware 1.03, the two drives in the degraded array immediately start re-syncing and the degraded array message goes away. 

[fordem]

I have also noticed the following behavior when I have a degraded RAID 1 on the DNS-323 with firmware 1.04:

If I downgrade to firmware 1.03, the two drives in the degraded array immediately start re-syncing and the degraded array message goes away. 

Is that - to your mind - good or bad?

As you have noticed there have been changes made to whatever disk error/failure detection mechanisms the DNS-323 uses in the upgrade to firmware 1.04 - I have not yet explored these changes - I will however make the following comments, based on your posts.

I'm running a pair of Seagate Barracuda 7200.9 drives in a RAID1 configuration and even with frequent (daily, sometimes more than once a day) power outages causing an improper shutdown, I have not experienced any degraded array messages, possibly because my unit was not being written to at the time of the improper shutdown.

To my mind the problem does not lie with the RAID implementation, but rather the improper shutdown and perhaps the lack of a UPS monitoring and automated shutdown capability - I've implemented many a RAID array over the last fifteen years, more hardware RAID than software, I'm only too well aware of just how much damage an improper shutdown can do to your array, and in my opinion, it's unavoidable, I've seen it happen on equipment where the RAID controller alone costs two & three times the price of a DNS-323 - the solution is to shut the system down properly.

On the firmware 1.03 causing an immediate resync of the degraded array - I wouldn't feel too happy - do you know why there was a degraded array message in the first place?  It doesn't necessarily mean that there is a failed disk, which your first post suggests that you seem to think is the case (and by the way - being able to mount the disk and read from/write to it doesn't mean that it's good, but just that the area you are reading from/writing to is good), a degraded array message can mean that the data on the drives is no longer in synch, and that now leaves you wondering - which of the two drives had invalid or corrupt data, and was that the drive that fw 1.03 chose to do the resync from.

[ECF]

vreid47362

Have you tried formatting both drives after the firmware update to see if the issue persists? it seem your original RAID1 configuration is from when you had firmware 1.03 and you have had the degraded state and may have formatted one drive to re-sync under the 1.04 firmware. We have had some changes and in firmware 1.04 when formatted now supports Unicode this may be an issue in your case.

[vreid47362]

I appreciate the thoughts of all who have responded to my posts on this topic. 

To answer some questions that have been posed:

First, I agree that the improper shutdown is the bigger problem than the raid re-sync.  I found that one of the batteries in one of my UPS's had gone bad.  Still, since this device is aimed at the consumer market, it is safe to say that most people in that category do not have their DNS-323 attached to a UPS.  So, it seems safe to assume that the degraded RAID recovery mechanisms should be robust enough to survive a power failure.

I have experienced quite a few raid failures caused by power issues over the past few years myself with 3Ware, Intel, High Point, Adaptec, and other controllers.  In most instances, I have been able to successfully recover RAID 1 arrays.  When I haven't been able to successfully recover, it has been either because of a power surge that fried hardware or because the underlying file system was not journalized.  I know that the last 2 or 3 DNS-323 firmware's have ditched journaling in favor of more speed and disk space (which I think is a mistake -- personally, I'd like the choice of ext2, ext3, or reiser).

I have tried reformatting the drives after installing the 1.04 firmware, but this seems to make no difference with my loss of sync problems.

I have 2 identical DNS-323's and they both lose sync with an unexpected power failure.  One thing that may very well be a factor in this is that I consistently use the enterprise grade hard drives (usually Western Digital RE's or RE2's) which have less built-in recovery firmware and are supposed to work better with RAID controllers.  Perhaps the DNS-323's are tuned to respond better to "average" hard drives.  Perhaps I'll drop a couple drive of these next week and give them a try to see what happens.

[ECF]

Thank you for all your input this will be investigated for a resolution if confirmed and all your input will help to improve our products

[fordem]

I appreciate the thoughts of all who have responded to my posts on this topic. 

To answer some questions that have been posed:

First, I agree that the improper shutdown is the bigger problem than the raid re-sync.  I found that one of the batteries in one of my UPS's had gone bad.  Still, since this device is aimed at the consumer market, it is safe to say that most people in that category do not have their DNS-323 attached to a UPS.  So, it seems safe to assume that the degraded RAID recovery mechanisms should be robust enough to survive a power failure.

I have experienced quite a few raid failures caused by power issues over the past few years myself with 3Ware, Intel, High Point, Adaptec, and other controllers.  In most instances, I have been able to successfully recover RAID 1 arrays.  When I haven't been able to successfully recover, it has been either because of a power surge that fried hardware or because the underlying file system was not journalized.  I know that the last 2 or 3 DNS-323 firmware's have ditched journaling in favor of more speed and disk space (which I think is a mistake -- personally, I'd like the choice of ext2, ext3, or reiser).

I have tried reformatting the drives after installing the 1.04 firmware, but this seems to make no difference with my loss of sync problems.

I have 2 identical DNS-323's and they both lose sync with an unexpected power failure.  One thing that may very well be a factor in this is that I consistently use the enterprise grade hard drives (usually Western Digital RE's or RE2's) which have less built-in recovery firmware and are supposed to work better with RAID controllers.  Perhaps the DNS-323's are tuned to respond better to "average" hard drives.  Perhaps I'll drop a couple drive of these next week and give them a try to see what happens.

Once upon a time (sounds like a fairy tale, doesn't it ?) computers were huge devices isolated in environmentally controlled chambers and attended by the white robed high priests - networks were unheard of, and all was good, no mere mortal had to worry about loss of data.

Fast forward to today - computers and networks are everywhere, network attached storage has made into the home network - it's time for the common folk to acquire some of the discipline of the white robed acolytes of yore - surely the consumer who can afford multiple computers, a network, and a NAS - can afford a UPS to protect his data.

BTW - I don't think (but I don't really know) that the discontinuation of ext3 had anything to do with speed or space, I do know there was an issue with data corruption when it was used, so my guess is that's what led to it being dropped.

[Fatman]

I almost thought I was about to hear about multivac for a minute with that intro.

Thank you fordem for saying what I have been trying to all along, a value of data implies a value in it's protection.

And so I leave you with "The Last Question" of these white robed high priests.
"How can the net amount of entropy of the universe be massively decreased?" - Isaac Asimov

[vreid47362]

I'm curious about how DNS-323 hardware version may be affecting the raid instability that I'm having. 

I was comparing the differences in support and documentation between D-Link Australia, and the USA, and I came across something rather interesting.

On the Australia D-Link site, the most recent recommended firmware for the DNS-323 version A1 is 1.03.  Firmware 1.04 is reserved for version B1 only on the Australian support site.

Both of my DNS-323's are version A1.  It is possible, I suppose, that 1.04 is optimized for B1 and is experiencing some instability on version A1 hardware.

Also, I have noticed that my Raid 1 instability appears to not be noticed until there are actually some 10's of gigabytes of files on its hard drives.

Also, off topic, but the DNS-343 looks like it will be a really nice product!!!!!

[ECF]

The firmware on the Australia website is not the same firmware as on our US support website.