On Nov 10, 2017, at 10:01 AM, Phil Turmel [off-list ref] wrote:

Hi Jun-Kai,

{Convention on kernel.org is to trim replies and avoid top-posting.}

On 11/10/2017 12:25 PM, Jun-Kai Teoh wrote:

quoted

I was reshaping/growing my array when it died. Computer came back up
but array wouldn't assemble.

I looked around and saw some commands on how to assemble without the
one drive that looked like it was kicked off sometime ago that I did
not realize (I used --assemble --force --verbose)

I kept getting the error that there were 6 drives, 1 rebuilding, and
not enough to get the array up. I unplugged the one drive that had the
incorrect superblock and ran "mdadm -A -R".

It ran and looked like it was resuming reshaping but it got really
slow - like 7 years slow.

I killed the process, rebooted the machine, and it just auto
reassembled the array and automounted it on boot.

Am in the process of backup right now.

And yes, the drive/array is mounted.

Ok.  I reviewed the other thread.  Consider creating a new array from
scratch after you complete your backups, using consistent partitioning
across all devices.  I would put LVM on top and leave part of it
unallocated (for emergencies), but that's just my preference.

Phil

My apologies all! I did not know this.

I'm backing up as much as I can, but I don't think I'll be able to back up everything. I can back up very little of the data, so I'm trying to prioritize carefully right now.

All devices are just 4TB drives, I didn't partition them separately for the array - or am I misunderstanding what you mean? <-- This is really common, all of you are far more knowledgeable than I am and I only understand some of the terms sometimes.