I have loads of old 1 tb disks and I can see that one gotten some error recently.

 

Is there a way I can remove that disk and still have parity (have loads of free space) is there a way to opt that content to the other disks automatically?

Someone will need to confirm, but I think the process is to:

1) Move the data on the "problem disks" to another drive assuming it's still operating.

2) Perform a new config, but leave out the disk you no longer want to use.

 

It would be nice if there was a "remove drive" feature instead of building a new config, but it isn't a big deal as long as you assign the disks correctly again.

You can check out this thread.

It will show you how to do this without losing parity..  I think I did this a loooog time ago and it works.  I'm not sure if it still applies today.. but I think it might..

 

Jim

But don't forget to move the files off the problem drive first before you start filling it with zeros.

 

Someone else can chime in if you don't want to go the dd route with the "mdcmd set invalidslot" process..

 

I'm sure there is an official way..  But you'll have to rebuild parity.

 

Jim

But don't forget to move the files off the problem drive first before you start filling it with zeros.

 

Someone else can chime in if you don't want to go the dd route with the "mdcmd set invalidslot" process..

 

I'm sure there is an official way..  But you'll have to rebuild parity.

 

Jim

 

"Is there a way I can remove that disk and still have parity"

 

I read this ^ as, not wanting to just pull a disk and have the parity drive serving up data (at risk of data loss if a drive dies)... not having a concern about actually rebuilding the parity.

You can check out this thread.

It will show you how to do this without losing parity..  I think I did this a loooog time ago and it works.  I'm not sure if it still applies today.. but I think it might..

 

Jim

It still applies, but the risk of data loss through a mis-typed command is HUGE compared to just pulling the drive and rebuilding parity from the remaining drives, assuming the array will currently run a full parity check without errors. If you are VERY careful, and VERY sure of your ability to understand what a command does and not just type blindly, then by all means, move the data off the drive and zero it out.

 

Much better in my opinion to copy the data to other drives, remove the desired drive and rebuild parity from the remaining drives.

 

To the OP, there is no good automatic way to empty a drive of data, you will need to enable disk shares and rebalance manually, or do it from the console, either local or telnet.

Thanks for the advices! it would have been cool if you could force the data on the parity disk (from broken disks to the array).

Thanks for the advices! it would have been cool if you could force the data on the parity disk (from broken disks to the array).

There is no data on the parity disk. It is only used in a mathematical equation with all the good disks to recreate the contents of 1 failed drive. If you have 2 bad disks, no recovery of those 2 disks is possible using parity. That's why it's so important to immediately correct any disk errors by replacing a failing drive. As long as all your other disks are ok, you can remove 1 disk, start the array in degraded mode, and copy the data off of the parity simulated disk to one of the good disks, then rebuild parity with the remaining drives to once again be protected from failure of 1 of the remaining disks.

I personally pull one drive and let it build. Swap out the next drive and let it built. Continue until all my drives are swapped out or upgraded to larger.

It would be nice if there was a "remove drive" feature instead of building a new config, but it isn't a big deal as long as you assign the disks correctly again.

 

+1

Thanks for the advices! it would have been cool if you could force the data on the parity disk (from broken disks to the array).

There is no data on the parity disk. It is only used in a mathematical equation with all the good disks to recreate the contents of 1 failed drive. If you have 2 bad disks, no recovery of those 2 disks is possible using parity. That's why it's so important to immediately correct any disk errors by replacing a failing drive. As long as all your other disks are ok, you can remove 1 disk, start the array in degraded mode, and copy the data off of the parity simulated disk to one of the good disks, then rebuild parity with the remaining drives to once again be protected from failure of 1 of the remaining disks.

 

Thanks for clearyfing. But I wrote disks when I ment disk.