Has unRaid ever saved your butt?--I'd like to hear about it.  e.g.: I'd really like to know how well the parity drive accurately reconstructs a failed drive. I'd like to know how well the unRaid server determines which drive is going bad in the array.  How it notifies you?--I don't recall seeing a place to enter in my email address for SMTP notifications; etc..

It has saved my bacon on MANY occasions. In fact, I have a failed drive right now, where bad sectors are accumulating.

 

There is no direct notification, unless you install the email notification system in UnMenu. You can set that up to email you with a daily (or period of your choice) all's well email, and also to email you in the event of a disk failure, parity check, or other events.

 

I'm looking at my WebUI page, and looking at the list of disks. Out of nine data disks, there are only three that haven't been replaced in the past two years. So I've had six disk failures of one sort or another. All but one rebuilt perfectly. The last one caused me to lose about 500 GB of data, which I'm in the process of rebuilding. Luckily, it's all just media stuff that I can re-rip. For my photos and other critical data I have a secondary back-up to the cloud.

 

UnRAID does pretty well in determining which disk is going bad, but it does a lot better if you're vigilant. For instance, I had a drive which was sending weird ReiserFS messages for awhile, but didn't notice until I had to use the console one day and it was sitting on the screen. I stopped the array and ran Reiserfsck, and it seemed to be okay, but I kept my eye on it, and sure enough a week later, that disk threw up a bunch of errors on a parity sync. It was sent back for RMA and has just been replaced.

 

Assuming your parity is valid, then UnRAID will reconstruct your data disk perfectly. It really does seem like magic, but you can't take it for granted.

I have had a data drive fail, at a later moment I have also had a parity drive fail.

 

In both circumstances the array and its data kept available, failure was noticed by means of the red ball.

 

I view unraid at two levels:

 

1) My level: I check for red balls daily (thats only one click away), and have also instigated email notifications on errors (thru SimpleFeatures)

 

2) Girlfriend level: Just works ... Even on failure the girlfriend level will not notice something wrong untill I tell her I need to spend money on another drive ;-)

I'm looking at my WebUI page, and looking at the list of disks. Out of nine data disks, there are only three that haven't been replaced in the past two years.

 

Wow that sounds really dismal--I've never heard of such failure rates before.  Now I am afraid to invest in hard drives for this lil nas project lol.  I was planning on going with up to 15 2tb drives lol.

I've never had a disk fail in anger in the 4/5 years I've used unraid. My method if usage has been carefull crafter such that they spend most of their time spun down.

 

However, when I was intially testing I built a small array and dumped a bunch of data on. Then purposely broke things to see how it would cope.

 

No problems and it behaved as expected. I would have confidence that if (when) I do have a drive failure the same will happen.

 

You could quite easily do the same test yourself for peace of mind.

Keep in mind that unRaid uses ALL the data area of EVERY member disk instead of just what is actively being used currently for data on the member drive for protection. A very large number of hard drives happily in service in windows machines have bad sectors in locations that just haven't been used yet, and until the bad area is actively hit, they will never show an error.

 

These two facts are the reason preclear is very important for a healthy unRaid box. You must be able to trust ALL the remaining drives to read correctly in their entirety to reconstruct a failed drive correctly. Monthly parity checks are good for preemptively finding a sketchy drive before you end up with a full failure on one drive, and a sketchy drive that almost works. NEVER put a marginal drive into an unRaid array thinking "it's all good, if it fails I can just reconstruct it". Your data depends on ALL your drives being perfect except 1. Knowingly using a bad or marginal drive is just asking for a 2 drive failure.

 

The good news is a drive that passes several preclear cycles is statistically very likely to live a long and healthy life.

i've had two instances of hardware failure where unRaid saved the day

 

http://lime-technology.com/forum/index.php?topic=18476.0

http://lime-technology.com/forum/index.php?topic=9662.0

Wow that sounds really dismal--I've never heard of such failure rates before.  Now I am afraid to invest in hard drives for this lil nas project lol.  I was planning on going with up to 15 2tb drives lol.

*shrug* Several of those drives were older ones that I reused. I also had several drives from Seagate's bad run.

 

Drives failing is why we all want data redundancy, right? I think I mentioned in another thread that I've had to do half a dozen RMAs in the past couple of months. Not all of those were in my UnRAID array. I also had to RMA a drive from my HTPC, and now I'm having to RMA the replacement they sent me, which is showing an increasing number of reallocated sectors. I have another drive that is also a replacement of a replacement. I just ran that second replacement through multiple pre-clears and it's coming back clean. Does that guarantee that it won't go bad? Of course not. That's the thing. Nothing guarantees that a drive won't go bad at any time. UnRAID does its job, and it does it well.

 

As a final note: I don't worry about the rash of failures I've had lately. Numbers are clumpy. While it may feel like someone's out to get me, I know that in reality, it's just the law of large numbers at work. For every guy who can say they've gone for years without a single drive failure, there's another poor shlub like me who has lots. Of course it is possible that I'm doing something very wrong. I would never rule out the possibility that I'm at fault. 

 

I've had at least 3 drives fail in service while in my unRAID array (in about 3 years):

 

1. a 1.5TB seagate that locked up due to their firmware bug of a few years ago

2. a 2TB western digital that would sometimes return different data for the reads of some sectors, the drive still seemed to work but once I figured out which drive was causing the problem I took it out of service

3. a 2TB western digital that recently failed

 

in all cases replacing the bad drive with a new one and rebuilding from parity worked without any issues.

 

In addition I have had either 2 or 3, 2TB western digital drives, fail to successfully preclear and I took these back to the dealer and got replacements for them.

 

I have switched to buying Hitachi drives.

 

Stephen

 

Thanks for all the replies; I'm really happy I went with unRaid. Seems like not many people are losing data and it's so scalable