Problems swapping 3rd drive of Raid 1 mirror
1) I have a Raid 1, 3 disk drive set so that I can periodically swap 1 of the drives off site for safe keeping. It's obviously easy to pull the drive out of service. When I replace the drive from a previously swapped out drive I start getting all sorts of errors: Drives are too full to use a mirrors, and a few more. What are the steps for me to take after I've removed a drive before (and after) I insert a previously used member of the trio?
2) I have "Fast Mirror Rebuilds" checked and my 2T on a 16T set of 3 drives takes over 24 hours to rebuild from scratch. Given my problem in question 1 I've had to erase the drive to get it to be part of the mirror. Why does it take so long?
This is a good use case and should be straightforward.
I will assume you are maintaining the primary drive and not removing it.
Best practice is unmount the volume, wait 30 seconds for the disk cache's to flush, then pull the drive.
Put it in a safe place and it cannot be connected to another computer in the meantime.
Insert the new drive and it should auto rebuild.
Volume optimization should be set to workstation, or server, NOT any other setting.
Where is what you are doing, any different?
You should be able to pull a drive live, with no adverse consequences, assuming this is a Thunderbay and not USB.
the only "problem' is the volume is "dirty" if it needed to be used independently. But hot swapping is totally allowed.
Where are you getting a different experience?
Attach a support file so I can see what the errors were.
Here's the support file. Thanks! I had been removing the drive hot... and adding it's replacement hot. Maybe that's the problem?
@dhk Looks fine to me. there is no history because of the Big Sur upgrade.
Put a SoftRAID "disk Label" on the primary, so you know which it is and check it is still primary when removing a disk.
You should be able to pull it hot. I would pull the drive 1", wait a full minute to spin down, then remove.
Unmounting the volume is better, but not required.
Once you have a full rebuild, next rebuilds should be fast. You recently validated, so "fast mirror rebuilds" are active on the volume.
Be in touch if you get any errors and post another support file.
My problem wasn't in pulling the drive, but what I should do to place the new one in service. I've tried just "plugging it in" and run into problems where I'm told there is not enough room on the disk for example.
OK... Take the new disk and you must initialize it with SoftRAID. (I recommend a certify first)
Then you can "Add disk". I suspect you have a volume on the disk.
Once a disk is part of the volume, all you do is plug it in. rebuilds are automatic.
We'll give it a try and get back to you. Thanks.
OK, I unmounted the mirrored volume, pulled the drive I wish to swap and put it in the drive box, and started to certify it. It tells me it will take 100 hours, and has remained there for the past hour or two. (16T drive). Is this normal?
Drives are shipped untested. Certifu helps eliminate future failure and DOA drives from being put into service.
To do so, the platters must be written random data, then read back. 3 times. Hence it takes a long time.
But worth it.