Drobo

Anyone experience False Hard Drive Failure?

Until DRI improves Drobo Dashboard to be more informative as to the possibility of failures, I STRONGLY recommend that you extract a diagnostic log and send it to Tech Support of analysis. In my case, out of 8 drives in my DroboPro, they flagged one as having had a failure (a 25 second failure to respond to a command, and another one that was having an excessive number of errors.

Both of those drives then passed WD’s diagnostic tests, with no problems reported.

However, I believe that says more about the WD tests than it does about Drobo. If they see a problem, after hundreds or thousands of hours of real life operations, why on earth would you believe ANY test that only runs for 15 minutes? Do you really think they make up these problems just to annoy you?

In my case, I immediately upgraded those particular disks to 2TB drives, and recycled the old ones to a test/temporary use. I strongly urge you to do the same, or return them to the dealer, even if (especially if) the drives are still new and under warranty.

I agree, Suite B, that its better to trust Drobo and its decision made on real operating data in lieu of a short test from the maker of the drive. Follow the money - the drive maker wants to deflect and minimize returns. Drobo just wants drives that work.

I agree with Suite B as well. Jives with my real-world experience running large arrays. I’ve gone through over 30+ drives in the years I’ve had a homebuilt array.

I had a drive that would fail to spin up sometimes, and sometimes the system would hang for a few seconds in the middle of doing something. Diagnostics said it was fine. This went on for about a month. Then one day the drive just didn’t spin up at all no matter what I did. Probably motor failure, but regardless the drive wasn’t functional.

Hello all,

let me describe my experience with possibly false hard drive failure.

I have a Drobo v2 with 3*1TB Seagate drives installed. 2 of them are exactly the same model. 2 days ago 1 of them was reported as a faulty drive my Drobo. I removed and installed a WD 1TB Green Drive. Relayout process begun and after approximately 30 hours everything went Green and A OK.

I took the faulty drive to my office and connected it to my computer with a USB External Case. The drive seemed to be ok. Partitioned it and run all kind of diagnostic tests checking all clusters one by one, etc. Everything seemed to work just fine.

I read some posts in these forums saying that some HDs are experiencing time-out issues that cause the Drobo to see them as faulty.

Headed on to Seagate support forums and I saw that there is a much newer firmware available for my 2 identical Seagate drives (one of them is the reported failed one). Updated the firmware to both the failed and the healthy drive.

Removed all drives from Drobo and added just the failed one in and reset the Drobo. It “saw” the drive as a healthy one again.

Removed the drive and installed the old 3 healthy ones. Drobo booted fine all files were there. Then I added the previously reported “faulty” drive in the 4th spare bay of Drobo.

Capacity increased and everything is Green and OK now.

Should I be worried that the drive might fail sometime soon?

If that is the case will I HAVE TO replace it or can I use the rest of the 3 always healthy drives? Free space is more than enough even with the 3 drives in there.

Best regards,
George

My opinion is that you would be better off with three healthy drives, than with three healthy plus one possibly bad one.

If for some reason, any reason, one drive or track or sector fails, and you need to restore it, that is NOT the time to hope that one drive is still healthy.

At least that is my feeling for a backup machine, as opposed to a primary drive with some other backup mechanism.

Now, if you had a Drobo Pro with dual drive redundancy, I might breathe a little easier. Then you would be able to survive two failures.

But most of the people on this list are always experimenting and trying new things, including new firmware, etc. That’s a sufficiently good reason to pick up some cheap drives on sale, plus any questionable ones, and use them for test/experimentation, or even another offline backup, until you can afford a second fully-outfitted Drobo.

My 2 cents, anyway.

That for sure is the safe attitude, but it can be an expensive one.
The problem from my point of view is that Drobo does not give ANY reason (except for specialists in the encrypted log) about the cause of a supposed “bad” drive.
If the cause is some time-out similar to the WD TLER off issue (see bhiga links above), then the Drive has no real issue, it is just an artifact of desynchronized disk<->Drobo recovery mechanisms.
What most people like with the Drobo is its very Apple-like minimalist and easy to use interface.
What is missing is an “explain” utility in case of problems, and even better, some preventive maintenance utility you could run periodically, and before any upgrade/rebuilt.

Suite B you were right.

The same drive reported as failed… failed again. The good thing is that after removing it Drobo did not ask me to put a new one there instead it started re-laying out data to the rest 3 drives (there is more than enough free space though).

Lesson learned. Once a drive is marked as faulty just never put it in the Drobo again.

Geeji: Double Agree!

geeji,

I’ve talked to Tech Support and know that a new version of the Drobo Dashboard is in the works, and that they have gotten the message about needing to improve the feedback in such cases. I know we all wish that such improvements can be made instantly, but…

In the meantime, I strongly suggest that anyone who experiences a reported failure immediately get the Diagnostic file for that drive, and send it to Tech Support.

I can’t blame them for encrypting the files, and we probably wouldn’t know what to look for in any case. And I can’t hurry up Product Development, especially if they are rewriting the entire Mac, so that it runs as a daemon. But in the meantime, they certainly CAN, and I believe WILL provide adequate tech support if someone experiences a reported failure.

I do agree in general about the preventative maintenance routine, but other than running the manufacturer’s diagnostics (and those haven’t proven to be very effective), I don’t know what else could be done.

I would like to see the periodic sweep that the DroboPro uses incorporated into the Drobo V2. That scans every sector at least once every 100 days, including mirror or stripe sectors, which I believe is very important.

Failing that, I believe that there are third party utilities that will nondestructively read (and/or copy) every sector on the disk, and presumably/hopefully correct any errors. If someone can recommend such a package for both the Mac and Windows, that would be very helpful.

[quote=“Suite_B, post:29, topic:467”]I’ve talked to Tech Support and know that a new version of the Drobo Dashboard is in the works, and that they have gotten the message about needing to improve the feedback in such cases.

I would like to see the periodic sweep that the DroboPro uses incorporated into the Drobo V2. That scans every sector at least once every 100 days, including mirror or stripe sectors, which I believe is very important.[/quote]
Thanks Suite B, that’s great news !.
I was not aware there was in the DroboPro this background scan utility, but for sure, that is exactly what I would like to get on Drobo too. Since the cost for DRI would be basically zero (I assume no porting effort is required), I hope they will not take lessons from Apple and deliberately cripple Drobo to justify the huge price increase of Drobo Pro.

[quote=“Suite_B, post:29, topic:467”]I’ve talked to Tech Support and know that a new version of the Drobo Dashboard is in the works, and that they have gotten the message about needing to improve the feedback in such cases.
.[/quote]

woo hoo!

maybe thats what all the VC funding is for (see my general post about the news)

Choosing to post this here, since we are discussing new drive failed and perceived failures.

My pro is:
4 x 1.5TB
4 x 2.0TB

and im running low on space, so im going to be upgrading.

im thinking it would be a good idea to put my “new” 2TB drives under some stress/testing for a week before plopping them in.

Does anyone know of any good software/have any ideas?

failing that i can easily just copy a few tb of data on and off them, but i was wondering if there was anything specifically built for stressing out HDDs. (windows please!)

@docchris – look for a Windows port of “dd”, its an old Unix command that you can use to write to drives at a block level. Its great for hacking, forensics, and cloning. Its part of the cygwin package.

It sure looks like a good idea, but if you look into the Forum for catastrophic failures where people lost all their Drobo data, those occurred because an error on one of the OLD disks, during the too lengthy rebuild of a new one.
But with Drobo Pro, you are allowed 2 errors, so that problem may be less fatal…

only allowed two errors if you have dual disk redundancy on - im sure some people dont

Plus that exactly why i dont want my new disk to fail during a rebuild, because it would leave my “at risk” period significantly longer!

Docchris, I think you have a good idea, but I’m not sure how practical it is.

As I’m sure you know, most electronic components have a U-shaped failure curve, with a certain amount of infant mortality, a lengthy period of good performance, followed by old-age failures. The question is how long does it take for the infant mortality problems to reveal themselves? Sure, if it is DOA, that’s easy to spot, but if there is a leaky capacitor, or a spindle bearing that only got 80% or the right amount of lubricant on it, how are you going to know – those things might not fail or a month or two? In other words, I think it is unlikely that a single day’s worth of testing would be likely to reveal anything, and perhaps not even a week.

Knowing you from your posts over the last year, I’m pretty confident that you have all of the data on your DroboPro backed up somewhere else as well (hopefully off-site), so that you aren’t at risk of a single catastrophic failure. If not, that’s something that you might want to address, urgently!

Regrettably, in my case I didn’t think through these issues well enough, and wasted time upgrading my DroboPro multiple times. (Yes, I had an existing backup. Or two. Or three!)

Finally, I did what I should have done in the beginning, and that was to remove all four 1 TB drives from a Drobo V2 and put them aside, then put four 2 TB drives in the V2 (my ArchiveDrobo), then copy all of the data from the DroboPro to the expanded V2 (much faster than rebuilding it).

Then I took the old 1 TB drives and added them to the DroboPro to increase the capacity that I would loose by turning on dual disk redundancy. After adding the first one, I turned on dual drive redundancy, and then added the remaining three all at once, so that it could rebuild quickly.

Shortly thereafter, I got an error message concerning one of the original 1TB drives, now about 18 months old. I sent the Diagnostic log into Tech Support, and they confirmed that one drive had taken too long to respond, and that another drive was “being watched” and looked like it might be about to fail. I therefore bought two more 2 TB drives, and replaced the two suspect 1 TB drives, one at a time (in order to maintain the dual drive redundancy).

Since then, I’ve taken those two WD drives, plus a 1TB Hitachi drive that I salvaged from my TimeCapsule after the power supply failed, and I’ve put them in a Drobo that I can play with. I’m using it for Time Machine for my MacMini, but there really isn’t any data on the Mac Mini, so if those drives were to fail it wouldn’t be catastrophic.

So at the moment, I have four 2TB drives in the V2, with 3.49 TB out of 5.42TB available. I’m using ChronoSync to backup my Mac Pro to that drive.

On the DroboPro, I have six 1TB drives plus two 2TB drives, with 4.16TB used out of 5.38 TB available with dual disk redundancy. I’m using ChronoSync to backup the Mac Pro to the DroboPro as well, and also have a Time Machine volume on that drive.

Both the DroboPro and my ArchiveDrobo contain multiple versions of all of my data, snapshots ken roughly six months apart. One of these days, I will have to get serious about data deduplication!

Finally, I am also backing up the Mac Pro to a Drobo V2 connected to my Mac Mini, using CrashPlan+ over the Internet, just to be sure it works. The advantage of CrashPlan+ is that the data on the disk is encrypted, which is importance since the Mac Mini is exposed to the Internet.

Once everything is synchronized, I will be able to access data from my office (via XP), or when I’m on a trip (via my MacBook), and no other solution I have found allows that on a cross-platform basis. However, because I am deliberately doing this over the Internet, I am limited to my uplink speed of 2 mbps, and I’m averaging about 1.5 Mbps. And at that rate, 20GB files take a long time!

Hi SuiteB

well my irreplaceable data is about 100GB, so its on two blu-rays (DL-RW) at my mothers house :slight_smile:

and on two computers, in parts of the house rooms (live synchronisations of each other)

and one of those computer backs up to drobo nightly (so i have previous versions/in case anything happens to the live sycned copies)

the rest is just media (lots and lots of media…) which can be re-acquired (most of it is just blu-rays and dvds and cd’s the originals of which are currently gathering dust hidden around the house) and the cost of “backing it up” (which is essentially going to be a second 'pro) outweighs the work required to just acquire it in case of catastrophe affecting my 'pro.

and i agree, a day isnt long enough, i was thinking a week of hardcore stress would give me a bit more confidence in a new drive

I know that obviously the OEM stress tests the drive before shipping - but I don’t know how long this process is.

Incidentally (and a little off topic) i was reading an interesting article the other day which suggests that just like you can now get hard disks with hardware encryption right on the logic board, Hitachi are considering adding de-dupe to the actual HDD itself, in hardware, which sounds like an interesting proposition!

I would love to know how many of these drives were recent WD models.

Maybe next year?

Does the FS do this?

Does the FS do this?[/quote]
The Drobo S does, so its derivative FS likely too.

Browsing the forums I got the impression those 2 models they have more differences than meets the eye.

[quote=“skywalka, post:39, topic:467”]Browsing the forums I got the impression those 2 models they have more differences than meets the eye.
[/quote]
The so-called Drobo-FS “self-healing technology” is your “periodic sweep”.
And I do agree it is unfair for Data Robotics not to offer it on Drobo-v2 too, since it is pure software and could easily be part of a software upgrade…