How long do I have to replace a failed drive?

[paperclip]

I don't have a failed drive. I may never, but if I do, how long do I have to replace it? Is there a point where not replacing a problematic drive quickly will result in the parity drive not being able to reconstruct it with the eventual replacement?

 

I suppose I should have a spare drive laying around just in case unraid ever needs it, but I don't. If one of my drives goes bad I'll have to place an order for a replacement drive. That'll take about a week to arrive.

 

So during that time, do I stop using unraid? Do I have to keep it powered down until I get the drive replaced? Or is it ok to use unraid, as long as I don't write anything more to it? Or can I continue writing gigabytes of data to the good drives and the parity drive will still be able to reconstruct the bad drive a week later?

 

I understand the procedure of replacing a failed drive, I just haven't read much about what to do and what not do between the failed drive notice and actually obtaining a replacement.

 

Thanks.

[dogmanky]

if another drive goes bad or your parity becomes invalid... then you have lost your data.  I'm assuming this is the asnwer you are looking for.

 

[WeeboTech]

This all depends on how old the other drives are, or how many you purchased at one time.

 

If you have 10 drives and they were all purchased at the same time, do what it takes to get the drive replaced as soon as possible. Chances of multiple drive failures on large batch purchases are greater then if the purchases were staggered over time.

If you have some very old drives, I would suggest doing the same.

 

If you have purchased drives as you need them at least a few weeks apart from one another, then chances are good you can shop for the most cost effective replacement.

 

I would not write allot of data to the virtual/simulated drive. It will involve reading every single drive.

I would read as little as possible from the virtual/simulated drive for the same reason.

I would probably move whatever data was on the virtual drive, to any free space on other drives.

This way if a second drive does fail very soon after, you won't loose everything. (just the data on the two failed drives).

 

 

[dogmanky]

I think weebo is saying .. you should have a spare handy.  

 

I concur!  I definately will keep one handy just in case.  It's best practice in my line of work, and pretty much the same in my home unraid...considering why i have a server with parity to begin with.  

 

 

 

on a side note.. it if were me.. id take the unraid down until I had a replacement drive just in case... but that's just me and my OCD!! 

[paperclip]

if another drive goes bad or your parity becomes invalid... then you have lost your data.  I'm assuming this is the asnwer you are looking for.

 

Well, my question was more of if a drive goes bad and I don't have a replacement available any time soon, can I still read and write to the array while still preserving the parity data that'll be used for the eventual rebuild.

 

I would not write allot of data to the virtual/simulated drive. It will involve reading every single drive.

I would read as little as possible from the virtual/simulated drive for the same reason.

I would probably move whatever data was on the virtual drive, to any free space on other drives.

This way if a second drive does fail very soon after, you won't loose everything. (just the data on the two failed drives).

 

I think I have a clearer picture now. As I understand it, if a drive fails I can still read and write to the array until my replacement arrives. When my replacement drive is installed, the parity drive will be able to rebuild it in its entirety even though I've been actively reading and writing from the share in the meantime.

 

Now that it's clear that actively using unraid while a failed drive hasn't been dealt with can be done, but probably shouldn't be done.

 

So my next question is, at what point would I be pushing my luck?

 

Let me propose the following scenario:

- Drive 1: 2TB parity

- Drive 2-5: 2TB data

- One share across all data drives

- 7GB used, 3GB free space

 

Now, data drive #2 goes bad. I order a replacement 2TB drive that'll arrive in one week. During that time I write 500GB to the share. When I replace the failed drive #2 one week later, will unraid still be able to rebuild all its lost content even after I wrote 500GB of new data to the share? It appears the answer is yes?

 

I think weebo is saying .. you should have a spare handy.  

 

...

 

on a side note.. it if were me.. id take the unraid down until I had a replacement drive just in case... but that's just me and my OCD!!  

 

That's what I'd probably do too, but I'm just curious for how unraid would react in this situation.

[SSD]

Now, data drive #2 goes bad. I order a replacement 2TB drive that'll arrive in one week. During that time I write 500GB to the share. When I replace the failed drive #2 one week later, will unraid still be able to rebuild all its lost content even after I wrote 500GB of new data to the share? It appears the answer is yes?

 

unRAID theoretically could operate with one failed disk indefinitely so long as another disk does not fail.  But you are pushing your luck!

 

The point of getting unRAID in the first place is to have data redundancy so that if a drive fails you can rebuild it.  If you have a failed disk in your array, and then another disk in the array fails, then you will lose all of the data on both the failed disks. 

[WeeboTech]

will unraid still be able to rebuild all its lost content even after I wrote 500GB of new data to the share? It appears the answer is yes?

 

Yes.

 

on a side note.. it if were me.. id take the unraid down until I had a replacement drive just in case... but that's just me and my OCD!!  

 

That's what I'd probably do too, but I'm just curious for how unraid would react in this situation.

 

You don't have to take your unraid server down. I would hold back on large amount of writes and reads on the virtual/simulated drive.

I've gone for fairly long periods of loosing a drive. Mainly because I did not catch it and the server sits under my desk.

My drives are pretty recent and purchased over periods of weeks and months, so I do not worry.

I do try to keep a spare around just in case.

[guiri]

I have a spare in mine but isn't there some way to tell the unraid to start bitc*ing at you if a drive goes bad? I mean, e mail, beep or something?

 

George

[Joe L.]

I have a spare in mine but isn't there some way to tell the unraid to start bitc*ing at you if a drive goes bad? I mean, e mail, beep or something?

 

George

See this thread: http://lime-technology.com/forum/index.php?topic=6702.0