I've looked for documentation on this so if there is some and I've just missed it I'd appreciate if someone can point me in the right direction.  With regard to disk failure scenarios, my interest is in what happens when disks fail beyond parity. 

 

So take the following scenario:

 

An array consisting of 8 data drives and 2 parity drives has 2 drive failures at once (lets assume the drives actually did fail and it's not a cabling issue, etc.).  You replace those 2 disks with new ones and begin a rebuild and a 3rd disk fails.  The array obviously can't be rebuilt now and I assume the data on those 3 failed drives is now gone forever (not including backups of course). 

 

My question is...Now what?  Your data is intact on the remaining 5 data drives but how does one go about building a new config with that data in place?  Is it even possible or do you have to copy all the data off those disks individually in a different system, then place the disks back into the current system with the (now) 3 replaced data drives to start from scratch with a fresh blank array?

 

I realize this isn't the likeliest of scenarios but I am curious none the less.

Each disk is an independent filesystem, so each disk can be read independently without the others. You can easily create a new configuration that only includes the good disks by going to Tools - New Config and it will rebuild parity based on the new configuration and the files on the good disks will be fine.

 

But, if you ever do get into a scenario like you describe, please ask on the forum for help. It might not be as bad as you think unless all the failed drives are truly unresponsive.

Each disk is an independent filesystem, so each disk can be read independently without the others. You can easily create a new configuration that only includes the good disks by going to Tools - New Config and it will rebuild parity based on the new configuration and the files on the good disks will be fine.

 

But, if you ever do get into a scenario like you describe, please ask on the forum for help. It might not be as bad as you think unless all the failed drives are truly unresponsive.

 

Yea my first thought having that many disk failures would be there is some underlying issue causing this that's not related to the disks themselves but I was just curious.  I was just curious and now that you mention the New Config tool I do remember using that in the past so thanks for confirming that process.

Each disk is an independent filesystem, so each disk can be read independently without the others. You can easily create a new configuration that only includes the good disks by going to Tools - New Config and it will rebuild parity based on the new configuration and the files on the good disks will be fine.

 

But, if you ever do get into a scenario like you describe, please ask on the forum for help. It might not be as bad as you think unless all the failed drives are truly unresponsive.

 

Yea my first thought having that many disk failures would be there is some underlying issue causing this that's not related to the disks themselves but I was just curious.  I was just curious and now that you mention the New Config tool I do remember using that in the past so thanks for confirming that process.

also, the reality is that in most cases those 'failed' disks are not totally unreadable.  Since each disk is a discrete file system this means recovery tools can still get most (if not all) data off them.

I put together a spreadsheet a while back which discusses  the probability of a disk failures.  You might want to read through it.  Here is the link:

 

        https://lime-technology.com/forum/index.php?topic=52229.msg501182#msg501182

 

If you are really risk-adverse and concerned about losing data, setup the Notifications option on your server and monitor what is happening.  As soon as the first disk fails, replace it immediately!  That way, two disks would have to fail during the rebuilt process!  And the probability of that happening is extremely  small.  If that level of risk is of concern to you, you should have at least two off-site backups of your server's data. 

I put together a spreadsheet a while back which discusses  the probability of a disk failures.  You might want to read through it.  Here is the link:

 

        https://lime-technology.com/forum/index.php?topic=52229.msg501182#msg501182

 

If you are really risk-adverse and concerned about losing data, setup the Notifications option on your server and monitor what is happening.  As soon as the first disk fails, replace it immediately!  That way, two disks would have to fail during the rebuilt process!  And the probability of that happening is extremely  small.  If that level of risk is of concern to you, you should have at least two off-site backups of your server's data.

 

Not sure if you read the full OP but I'm not at all concerned about losing data or disks failing.  This was just an "I'm curious what the process is" question for those very rare situations.