UFS gpt and whole disk mirrors?

spork

spork

In the pre-gpt days, this was possible and was a drop-dead simple way to create a mirror, even after the initial install.

My understanding is that gpt and gmirror battle for the last sector of the disk for metadata, and therefore the gmirror method became more complex - you basically have to create a mirror for each partition, which seems quite a bit more complex as it involves booting to alternate media.

While I was googling for more info about this, I did occasionally see some comments that full-disk mirroring with gpt would one day be implemented, but as of now, the Handbook doesn't mention any such thing. So is it still a wishlist item? Are there some unofficial workarounds out there?

Has everyone just moved on to zfs? :)

I have a home server that boots UFS, but then has a huge zfs pool. I originally went UFS for the boot drive on the whole KISS principle, but now that I'm finding $20 120GB TLC SSDs at retail outlets, I'm thinking everything I've got at home with a single boot drive should now get mirrored, it's just crazy not to.
 
I am using whole disk mirrors using GPT scheme. No problems here. There is a warning in the handbook but I have not had any issues.
Have removed drive and added new one. Rebuilt automatically in good time.
forums.freebsd.org

Install UEFI FreeBSD on gmirror

I wanted to post instruction for installing FreeBSD on a GEOM_MIRROR -aka- gmirror(8). This is an advanced topic so I assume you are capable of determining that your two chosen disks are empty. UFS RAID1 on FreeBSD is enabled with the geom_mirror module. I am using ada0 and ada1 as examples...
forums.freebsd.org forums.freebsd.org

I have even tried RAID10 instructions from DutchDaemon using GPT instead of MBR.

TrueNAS - Welcome to the Open Source Storage Era

TrueNAS is the World's #1 Open Source Storage. Based on OpenZFS, it is trusted by millions and deployed worldwide.
www.freebsdnews.com www.freebsdnews.com
Everything worked fine but I did not see double throughput I was hoping for. So that was just for testing.

My virt machine uses two Samsung PM953 M.2 NVMe in a geom_mirror for hosting my VM's.
Code: 
root@virt:~ # gmirror status
       Name    Status  Components
mirror/tank  COMPLETE  nvd0 (ACTIVE)
                       nvd1 (ACTIVE)
Code: 
root@virt:~ # gmirror list
Geom name: tank
State: COMPLETE
Components: 2
Balance: load
Slice: 4096
Flags: NONE
GenID: 0
SyncID: 27
ID: 2177820761
Type: AUTOMATIC
Providers:
1. Name: mirror/tank
   Mediasize: 960197123584 (894G)
   Sectorsize: 512
   Mode: r1w1e2
Consumers:
1. Name: nvd0
   Mediasize: 960197124096 (894G)
   Sectorsize: 512
   Mode: r1w1e1
   State: ACTIVE
   Priority: 0
   Flags: (null)
   GenID: 0
   SyncID: 27
   ID: 4078490585
2. Name: nvd1
   Mediasize: 960197124096 (894G)
   Sectorsize: 512
   Mode: r1w1e1
   State: ACTIVE
   Priority: 1
   Flags: (null)
   GenID: 0
   SyncID: 27
   ID: 3840500338

So unless you need to run gpart recover (which does knock out the last sector) all is well.
With a mirror I have not had any disk issues that needed recovery.

Because I am using my two M.2 drives on a Supermicro AOC-SLG3-2M2 card I need PCIe bifurcation.
As I tinker alot with my NVMe the second NVMe has gotten knocked offline many times when I switched around card slots.
(When you stick this card in a slot not designated as bifurcation only 1 NVMe device shows up)
The array rebuilds itself quite reliably once I notice the issue and fix the bios settings.

I have another machine where I use gmirrored Innodisk SATADOM for booting.
 
You must log in or register to reply here.
Back
Top