Print

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

[jeffers.r]

OK. I have NAS-323 with 2 WD15EARS as RAID-1 with EXT3  already 30% filled with files. What should I expect in the future with these "unsupported" drives? Data loss? Speed degradation?
So far speed has been on 1 Gb link with no jumbo  130-160 Mbit/s (on new empty drives) and 7-8 MBytes/s (72-85 Mbit/s) over 100 Mbit/s network (I ran NAStester and watched link utilization at the same time).
Any forecast?

I'm in almost the exact same situation as above. I've been running 2 WD20EARS on RAID1 since January. 70% full. I've noticed that the speeds were slower then I expected, but didn't think too much of it until finding this thread. I'm more concerned about the potential for data loss, can anyone elaborate?

[jamieburchell]

If it was me, I'd replace the drives with supported ones. If you are running a RAID array, you should be able to swap one at a time. Before you do anything, and if you are that concerned about data loss, you should have another backup of your data.

[jeffers.r]

Honestly, I'm a bit baffled by this entire issue.

As I mentioned above, I've been running a DNS-323 using two WD20EARS drives since January in a RAID1 config without any issues what-so-ever. After reading this thread, I became concerned and decided to test things out on my second DNS-323 I purchased recently (with the exact same drive type) by removing the RAID1 and reformatting to the drives (according to some of the step-by-steps guides on these forums) in order to sync them correctly.

Now, when I compare my original DNS-323 with RAID1 without proper syncing, to the new DNS-323 without RAID and the drives properly synced, the performance is, for all intensive purposes, identical.

I'd love to simply go out and buy new "compatible" drives, but that's not in the budget at the moment. So in the meantime, it's tough to take any steps when I'm not having any issues.

Any thoughts?

[mastermayhem]

I received my 2 X Western Digital WD15EARS Caviar Green 1.5TB SATA 64MB Cache today and plugged them in, got the "format" message and attempted to format them. The format failed with an error message and I was prompted to restart. Upon restart, the drives did not show up at all. I used one of my external enclosures (eSATA) to check the drives. One appears to have failed immediately, and the other looks fine to Win7, but once I put it back in the DNS-323 it will not format. Status shows that a drive is inserted, but I cannot format it and the "Physical Disc Info" section will not show any info for the disc. It won't even show the side that it is inserted in. (ex. Drive is in "right" slot and only "left" is showing)

I have updated and then downgraded firmware, formatted the disc in Win7 using Acronis Disk Director Home in EXT2 and even completely deleted all partitions. Anyone else run into this or have a thought for a solution?

[serco]

For guys like me that use the retail WD hard disk WD20000CSRTL; and theses drives are WD20EARS since a month (then non exchangeable, non refundable).
 
I have firmware 1.09
I have 2 x WD20EARS drives formatted in RAID 1 (mirroring)
I use Windows 7 Ultimate (64 bits)

What are the suggestions to keep the DNS-323 with WD EARS drives up and running, with the minimum of inconvenients?

Also, no words from D-Link Staff in this thread since May 10th.

1) The D-Link position of not supporting the WD 2TB EARS drive is final?

2) D-Link continue to discuss with WD to find a way to help their users?

Serco

[gunrunnerjohn]

Different brand of NAS?

[SalsaNChips]

Regarding this question about WD drives...

Folks, this is a problem with WD. Specifically, sometime in early 2009 they made a decision to intentionally cripple the firmware in their non-enterprise (i.e., Caviar) product line in a way to make them ONLY function reliably as stand-alone drives in desktop applications. No RAID functionality.

The reason for this was to protect the market space for the more expensive enterprise class RE3/RE4, etc. drives. Think about it -- why would somebody want to pay 20-30% more for enterprise-class drives for a clustered storage/RAID application when the cheaper desktop drives function just as well? Solution -- make the cheaper desktop drives NOT FUNCTION SO WELL in those type applications (RAID).

One of the ways WD did this was by disabling a feature known as TLER (Time Limited Error Recovery) on the Caviar drives. Google it for details. Causes drives to drop out of RAID arrays when they encounter read error lasting more than a few seconds (actually, 7 seconds). Thus forcing you to unplug the drive from the hot-swap backplane, plug it back in and force a rebuild, whereafter it functions just fine for a while (sometimes weeks) then does it again.

How do I know all this? Because I just built a new Windows 7 (64) PC using an ASUS mb with an Intel RAID controller and a Corsair chassis with a SATA II hot swap backplane and made the mistake of selecting 4 WD 1TB Caviar drives for my 2TB RAID 10 array. Everything works great for 2, sometime 3 weeks then I will lose a drive from the array, have to hot re-plug and re-build it in Intel RST v9.6. I did a lot of research on it and discovered the TLER issue. Used WD tools to probe the time limedted error recovery parameter in the firmware on my drives. It was turned off and DISABLED in firmware.

One of the things that sort of obfuscates the issue and masks WD's involvement in this is that prior to about August 2009, the TLER parm was totally functional and yes, you could buy a cheaper Caviar drive, stick it in a RAID array and everything would be great. Those were the days. Then WD started turning TLER off in firmware and a few months later, actually DISABLED (crippled) the TLER parm in the firmware so you couldn't turn it back on, even with the correct WD utilities (which about the same time, disappeared from WD's download site).

So, if you have a Caviar drive made before a certain date, you may have no problems at all running it in a RAID configuration. But you WILL have a problem if you use recent, current production WD non-enterprise drives.

WD changed their marketing or sales materials to start using the work "desktop" a lot more in the product description of the Caviar drives. At the same time, they are now trumpeting the RE3/RE4 enterprise class drives as being "TLER enabled". Whatever...

Myself, I am ditching my TLER disabled but otherwise fully functional 1TB WD Caviar Blacks when I get the chance for a pair of Seagate Barracuda or Hitachi"s. And that *especially* applies to the drive selection for my soon to be acquired the DLINK DNS-323 NAS as well. Just wish I had known all this 4 months ago before I made a $450.00 mistake purchasing WD Caviar drives for RAID 10

[jamieburchell]

If your drives are dropping out of RAID because of read errors, doesn't that mean the disks are not happy anyway - regardless of TLER settings?

[SalsaNChips]

If your drives are dropping out of RAID because of read errors, doesn't that mean the disks are not happy anyway - regardless of TLER settings?

Nope. Google "TLER" for specific details on why it is required in a RAID application.

My understanding is this -- in a RAID application, the amount of time required to sucessfully complete a disk write operation needs to be relaxed somewhat from what it would be if the drive was operating as a stand-alone drive in a non-RAID configuration. That is, wait longer before timing out and reporting a disk error to the controller. An error reported by a disk WILL cause the controller to flag the disk as having malfunctioned and go into a recovery mode which degrades RAID performance. Or, as in my case, causes the Intel RST service to flag the drive as "offline" until I hot-plug it, whereafter it initiates a re-build (lasting a couple hours) and everything is okay until the next time it happens.

One additional detail to my original post, FYI -- I originally tried RAID 5 with two WD Blacks and had MAJOR problems. RAID 10 with 4 drives worked MUCH better, still the occasional drop-outs due to the TLER issue but tolerable until I can sell the four WD Blacks and get something else (NOT WESTERN DIGITAL).

[SalsaNChips]

What are the suggestions to keep the DNS-323 with WD EARS drives up and running, with the minimum of inconvenients?

Suggestion would be to get new drives that will work in a RAID configuration.

See here:
http://community.wdc.com/t5/forums/searchpage/tab/message?q=EARS+RAID

[gunrunnerjohn]

Obviously, WD is becoming "customer hostile" and forcing NAS users into more expensive (bur not more capable) drives.

Simple answer?  Don't buy from WD, the obviously don't need our business!

[jamieburchell]

Nope. Google "TLER" for specific details on why it is required in a RAID application.

What I was getting at is if your drives are healthy in the first place and aren't timing out on writes to bad sectors- where's the problem? You only run in to the TLER if you have errors on your drives?

[tentimes]

What I was getting at is if your drives are healthy in the first place and aren't timing out on writes to bad sectors- where's the problem? You only run in to the TLER if you have errors on your drives?

All drives give frequent read errors which are quietly corrected without you knowing - it's part of how they work with redundancy. If one of these normally benign (and unnoticeable) errors randomly hangs for more than a certain amount of time then you have a down drive. Normally the TLER would be part of the benign operation of the drive, but now it causes a critical failure. That's why it's a problem. I would recommend reading up on how modern drives work and that will give you a better understanding of error correction and recovery which are part of the normal proces of drive management.

[tentimes]

Nobody has yet replied on whether these drives work ok in JBOD operation. Are they going to be running with the misallignment error in JBOD? In that case the answer is don't use them in any situation. If they do align ok and the issue is only in RAID then I would love to hear that as I have one on it's way to me now I was going to use it to replace a fast SAmsung spinpoint 1TB in the 323 which I was going to pinch for my desktop duer to it's speed.

Can anyone comment further? I'm reading up on the problem right now but am unsure of why it is an issue for the DNS-323. CAn the 323 align the drive correctly or not under ext2 (in non raid)?

[tentimes]

EDIT: The drive cyclinders are misaligned if you format these drives as they are and put them in the 323. I have confirmed this by checking after format on a linux machine. The following comment is a link to how to align them properly so you won't be getting delays. The biggest delays will be in small random writes (up to 20x slower if not aligned). Just thought I would make that clear as my first comment might have been taken wrongly.


These drives can be used ok in standalone mode - I spent 3 days investigating and have written a guide on how to do it: http://forums.dlink.com/index.php?topic=14484.0
JBOD shouldn't be a problem either, but I don't know how that works in terms of formatting (and I haven't the spare drives to find out) or I would have written that too.

The missingt TLER bit seems to have banjaxed RAID though - I can't see a way round that without a kernel update and firmware update