Usability Request for Failed Drives

[jeffreywhunter]

I recently had my first drive failure.  I then pulled the old drive and inserted a new one and then (ERRANTLY) assigned it to the old drive slot (not realizing that the system would not automatically take me into a pre-clear process).  Since I had assigned the drive to the slot the system immediately assumed was ready to use and promply/automatically pulled the drive into the array and began to format it.

 

I now know that I should have pre-cleared the drive first, then assigned to array..."duh" yes, but noobs do what they do...

 

What added to the confusion was when I saw the message that displayed near the start button, it further affirmed my errant understanding.

 

Start will bring the array on-line, start Data-Rebuild, then expand the file system (if possible). 
[checkbox] Yes I want to do this

 

Since this was my first drive failure experienced with unRaid, when I saw this message, then I thought (as most probably would?).. "Of course, I want to start data-rebuild"...  unRaid then started the rebuild process which is NOT what I wanted (I wanted to Pre-Clear first).  Once the rebuild process was started on the drive, even though I stopped the rebuild, rebooted, pulled the drive and rebooted again, it still remembered the drive (which it should have!).  I solved the problem by inserting a second/different 3TB drive which was recognized by the system, I then ignored the Start/Data-Rebuild message and did a Pre-Clear.  All is well...  Now I understand that you just ignore the Data-Rebuild message and do the Pre-Clear...

 

What would be nice (for expanding usability and for newbies) is that when you insert a new drive and start What should display here is ANOTHER button below the START button that says, "Pre-Clear Disk" with a message confirming, "PRECLEAR will prepare the replacement drive you inserted into your system.  If you have ALREADY pre-cleared your drive, then click START.  Most people will want to Pre-Clear their disk before adding it to the array.  [Link to Read about Pre-Clearing a Drive] [Link to Read instructions for replacing a failed drive]"

 

I created a mock-up of what the Array Operation page could look like after you have inserted a new drive into the system.  I would also think that the system should be able to identify if the drive has already been precleared based on the drives configuration, correct?

 

Mock-up of Main>Array Operation data-rebuild screen with new preclear option:

 

http://my.jetscreenshot.com/12412/20151207-0dtz-87kb.jpg

 

What would be even better would be to have instructions available when a failed drive is detected in the system.  Anything in the way of information or guidance to help the unknowing how to proceed...

 

[bonienl]

Nice mock-up, but first pre-clear needs to become standard in unRAID...

[trurl]

You probably know this, but it is not technically necessary for a rebuild drive to be clear. Preclearing is just for testing in that case.

 

Preclear is not actually required by unRAID in any case, since unRAID will go offline and clear a drive if it is added to a new slot in a parity array. Preclear just does a better test and keeps unRAID from having to clear it.

[jeffreywhunter]

I knew that it was not necessary, but from what I could read on the forums, it appeared to be very highly desired, and to do it 2 or 3 passes.  So you are saying that unRaid will automatically preclear the drive, but that its not as thorough as the preclear_disk.sh...  Still seems like its necessary even though it takes a lot of time to do so...

[jeffreywhunter]

Nice mock-up, but first pre-clear needs to become standard in unRAID...

 

I hope that becomes the case.  Its these fine points that can go askew very easily and frustrate the uninitiated...

[trurl]

I knew that it was not necessary, but from what I could read on the forums, it appeared to be very highly desired, and to do it 2 or 3 passes.  So you are saying that unRaid will automatically preclear the drive, but that its not as thorough as the preclear_disk.sh...  Still seems like its necessary even though it takes a lot of time to do so...

unRAID will take the array offline while it clears a drive for a new slot so parity will remain valid. Preclear takes care of clearing in advance of adding it to a new slot, which is why it is called preclear. Either way will thoroughly clear a drive, but preclear does a lot more to test that the drive is readable and cleared.

[John_M]

I prefer the way unRAID currently behaves. In other words, if a disk fails and I replace it, I want unRAID to rebuild the array as quickly as possible. It's my choice whether I use an untested virgin disk as the replacement or one that I've already tested by pre-clearing it. I think the sensible approach is to keep a tested (ie. pre-cleared once, twice or how ever many times you feel comfortable with - personally I prefer not to flog them half to death before using them) spare disk to hand. If I had a disk failure but no ready tested spare, I'd use a virgin, run the short SMART self-test on it (typically taking 2 minutes), and then let unRAID do the rebuild, rather than leave it unprotected for the two days it takes to run one cycle of pre-clear.

[jonp]

I prefer the way unRAID currently behaves. In other words, if a disk fails and I replace it, I want unRAID to rebuild the array as quickly as possible. It's my choice whether I use an untested virgin disk as the replacement or one that I've already tested by pre-clearing it. I think the sensible approach is to keep a tested (ie. pre-cleared once, twice or how ever many times you feel comfortable with - personally I prefer not to flog them half to death before using them) spare disk to hand. If I had a disk failure but no ready tested spare, I'd use a virgin, run the short SMART self-test on it (typically taking 2 minutes), and then let unRAID do the rebuild, rather than leave it unprotected for the two days it takes to run one cycle of pre-clear.

And this is how it will stay. Pre clearing is something we will probably add to the webgui as a standard feature, but its not actually necessary to function in the system or rebuild a drive.  Clearing is only required when expanding an existing array, which happens automatically. Does any other NAS solution prompt the user to do something to test drives before inserting them to replace one that failed or even to add one in the first place?

 

Trust for your hardware is a personal decision if you ask me. Some folks want multiple clears run others don't care.  I don't see us forcing a hand here or adding extra options before a disk rebuild occurs. If you care enough about having your drives run through multiple pre-clears, that should have happened on a stand-by disk before replacing the one that failed. Once a disk fails, I would argue that taking the extra time to run multiple pre clears on the new disk while letting the array run in a degraded state is more risky than just rebuilding the data on that disk and moving on.

[jeffreywhunter]

I prefer the way unRAID currently behaves. In other words, if a disk fails and I replace it, I want unRAID to rebuild the array as quickly as possible. It's my choice whether I use an untested virgin disk as the replacement or one that I've already tested by pre-clearing it. I think the sensible approach is to keep a tested (ie. pre-cleared once, twice or how ever many times you feel comfortable with - personally I prefer not to flog them half to death before using them) spare disk to hand. If I had a disk failure but no ready tested spare, I'd use a virgin, run the short SMART self-test on it (typically taking 2 minutes), and then let unRAID do the rebuild, rather than leave it unprotected for the two days it takes to run one cycle of pre-clear.

And this is how it will stay. Pre clearing is something we will probably add to the webgui as a standard feature, but its not actually necessary to function in the system or rebuild a drive.  Clearing is only required when expanding an existing array, which happens automatically. Does any other NAS solution prompt the user to do something to test drives before inserting them to replace one that failed or even to add one in the first place?

 

Trust for your hardware is a personal decision if you ask me. Some folks want multiple clears run others don't care.  I don't see us forcing a hand here or adding extra options before a disk rebuild occurs. If you care enough about having your drives run through multiple pre-clears, that should have happened on a stand-by disk before replacing the one that failed. Once a disk fails, I would argue that taking the extra time to run multiple pre clears on the new disk while letting the array run in a degraded state is more risky than just rebuilding the data on that disk and moving on.

 

All great points, and the challenge is balancing between consumer and commercial grade answers.  One a commercial basis, having a stock of pre-cleared drives only makes sense, hard to justify the expense for consumers.  Its always hard to balance these issues - but your point about a "personal decision" is the right one.  So in the end, the system just needs to provide an easy way to recover a bad disk - which it does if you already understand the assumptions of having a precleared drive in waiting.  I did not understand that, and there was nothing available to me as a newbie going through a disk recovery for the first time to help me get that...ergo, the recommendation.  A few simple comments, help files or wizards would help the usability significantly for low effort..."all I'm sayin"