% QNAP_commission_disk(1) Version 1.0 | QNAPHomebrew admin
QNAP_commission_disk — Tries to keep HDD silent
QNAP_commission_disk [-b|--bootorder n ] [-v|--verbose] [-c|--check] ...drive
| QNAP_commission_disk [-h|--help]
Probably one of the first commands run in setting up QNAPHombrew.
It partitions the given disk to make it suitable for the QNAPHomebrew model**.
-h, --help
Prints brief usage information.
-v, --verbose
Make output more verbose.
-b, --bootorder
Determines the "preference" for this disk at boot (see model**)
-c, --check passes
The check flag to mkfs (can be specified twice, see mkfs(8) man page
<dev>
The device being reformatted. Can be of the form /dev/sdb or tray3 etc.
This is relatively destructive command. You are, in effect, handing the whole disk over to the QNAPHomebrew "world". To reduce the risk of accidental trashing an in-use disk it first checks the disk is "empty" . To make it empty you should simply add a GPT partition table (e.g. with gparted in another machine, but any method is acceptable) . There is no actual need for GPT but no other partitioning systems have not been tested.
You should use unique numbers for the bootorder, this is not checked or enforced but the choice of boot disk in that situation is undefined.
This command makes the filesystems:
- boot
- root
- var
- swap
- tmp
- home
- data
- rest
With the appropriate labels (e.g. label=home3) as required by the model**. You may well want to customise this script to choose values suitable for the odd disks you have. The partition with the label "dataX" will be of fixed size and exist in partition 9 (so /dev/sda9, /dev/sdc9 etc ..to be clear the filesystem on /dev/sdc9 may have the label data3 [3rd disk]) , so make sure that size fits on all the disks you plan to use it on (need not be all disks). Partition 9 is unusual in that it can be used as "yet another filesystem" (as-is) or replaced with a single RAID array "/dev/md0" ( see: QNAP_create_raid(8) and QNAP_recreate_raid(8)).
In the reference system. There were 4 HDD and a single (new) external SSD.
- TRAY1 => sda => SAMSUNG HD103UJ (1TB)
- TRAY2 => sdb => SAMSUNG HD154UI (1.5TB)
- TRAY3 => sdc => HGST HDN724040ALE640 (4TB)
- TRAY4 => sdd => SAMSUNG HD154UI (1.5TB)
- TRAY5 => sde => Kingston SSD (0.5TB)
These would have been commissioned as:
QNAP_commission_disk -b1 tray1
QNAP_commission_disk -b2 tray2
QNAP_commission_disk -b3 tray3
QNAP_commission_disk -b4 tray4
QNAP_commission_disk -b5 /dev/sde
This neat layout is very nice and helpfully it's actually what seems to happen when the QNAP is booted with all disks in place. However, be aware, if disks are added or removed post boot the numbering will differ. Also you MAY choose a different layout. However ease-of-use features like "TrayX" will assume this setup and examples in manuals also use it.
To explain the model, it is easiest to consider a concrete example (the system this was developed on). This is an "old" TS412 , which is an ARM based system with 4 "trays" holding disks + 2 eSATA sockets. The Trays hold an odd assortment of disks which were "retired" from another (QNAP) NAS for various reasons (too small, too hot, reporting SMART errors).
The use case is that this NAS holds infrequently accessed data" maybe backups or saved scans of documents or Photos. This system would normally sit with all the disks stopped and the Fan switched off. On my system this settles at around 37C and is, of course, silent.
The system should be able to boot with ANY SINGLE DISK remaining, so 3 of the 4 disks could fail and it would still boot. In the reference system I added a 5th disk, a small SSD plugged into the first eSATA port, this may sometimes be called Tray5.
The key to the resilience is an initrd(4) script called 'choose_root' this gets installed (by the Makefile) in /etc/initramfs-tools/scripts/local-top/. It looks for filesystems with labels like root1, root5 etc (single digit) and when it has checked them all it chooses the one with the highest number and that becomes the root filesystem. This in turn mounts label=homeX on /home etc, this is defined by your /etc/ftab. For example:
LABEL=root5 / ext4 errors=remount-ro 0 1
LABEL=boot5 /boot ext4 defaults 0 2
LABEL=home5 /home ext4 defaults 0 2
LABEL=tmp5 /tmp ext4 defaults 0 2
LABEL=var5 /var ext4 defaults 0 2
LABEL=swap5 none swap sw 0 0
Note, unless you use QNAP_create_raid(8) and/or QNAP_recreate_raid(8) there will be no RAID in this setup, so no data resilience.
Then next feature of this model is that the first few filesystems (root, boot, home,tmp, var) are all "almost" identical. This is achieved by use of QNAP_clone_disk(8) QNAP_clone_alldisk(8). This has a side-effect that files stored in /home will have many identical copies.
So the combination of choose_root script(8) QNAP_commission_disk(8) and QNAP_clone_disk(8) provide the "secret sauce" that allows the system to come up with only one working disk.
- /etc/fstab
: Needs custom setup using label= style syntax.
The "elephant in the room" is how do you get started. Well there are a number of strategies. The one I used was:
- Install Debian (bookworm)
- I started with stretch and upgraded
- see QNAPinstalldepends(8), to shrink initrd(4)
- read up on SABOTEUR, Mouiche and M11 layouts
- Read up on the "flash features"*** of QNAP Homebrew.
- Install 'choose_root'
- Use QNAP_commission_disk(8) to setup a new disk.
- Use QNAP_clone_disk(8) to clone the existing system onto the new disk
- Reboot, and check you come up with the new disk as / (root)
- Commission mores disks and clone to them all.
***My first TS412 Motherboard failed while reflashing, so I now boot via BOOTP and avoid reflashing the system.
See GitHub Issues: https://github.com/graemev/QNAPhomebrew/issues
Graeme Vetterlein graeme.debian@vetterlein.com
hdparm(8), QNAP_config_disk(8), sfdisk(8), fdisk(8), cfdisk(8), parted(8)