Quick pop back here to comment on what I’ve found recently, having just built a 1TB disk with these instructions. I used the systemrescueCD linux liveCD (actually, a DVD…), loaded into RAM, and running under an x-windows shell (startx
).
- GNU parted now uses IEC units. So ‘M’ is 1,000,000. This means that if you enter exactly what fox described using a GNU parted more recent than v2.4, you won’t get partition sizes and positions that you are expecting; fox expected ‘M’ to be old skool ‘MB’, meaning 2^20. I couldn’t run the dd for partition 5, because it was too small (99.5M, with partition 6 being 101M).
So, you can either use ‘MiB’ instead of ‘M’, or, when you start parted, set the default unit with unit MiB
, then you can just enter numbers (within parted):
unit MiB
mklabel gpt
mkpart primary 528 2576
mkpart primary 2576 4624
mkpart primary 16 528
mkpart primary 4828 100%
mkpart primary 4624 4724
mkpart primary 4724 4824
mkpart primary 4824 4826
mkpart primary 4826 4828
set 1 raid on
set 2 raid on
print
Having done this, print
reported partition sizes exactly as expected, and all the dd
commands worked.
I suspect the OP’s problem with running out of space was not being careful with the start point of partition 4; I suspect he entered 4628 rather than 4828. The numbered partitions are not created sequentially on the disk.
-
I’m pretty sure fox has a typo for the partition used for the swap space; he had /dev/sda3
, but all his other partitions are on /dev/sdb
, so I reckon it should be:
Format data partition:
mkfs -t ext4 /dev/sdb4
mkswap /dev/sdb3
-
I had trouble with the RAID stuff. I had to reboot after making the partitions, before performing the mdadm
commands. Even then, the --assemble
command (-A) reported an error. I carried on regardless, and the --create
seemed to run okay.
mdadm --assemble /dev/md0 /dev/sdb1 /dev/sdb2
gave error:
“mdadm: no recogniseable superblock on /dev/sdb1”
“mdadm: /dev/sdb1 has no superblock - assembly aborted”
It looks like this is because fox runs the --assemble
command before he runs the --create
command (which creates the superblocks assemble is looking for), This doesn’t look right, according to the documentation for mdadm
:
https://raid.wiki.kernel.org/index.php/RAID_setup
An earlier version of the instructions had the mdadm
command sequence reversed (i.e. ‘correct’).
I now have a 1TB MyCloud, up and running after the first attempt at mating the new HDD. Now that I’ve tested the process, I need to look out for an 8TB Red.
[edit: couple of follow-up glitches, although they’re fairly routine.
When setting it up after first boot, the last thing I did was turn off remote access. It didn’t seem to want to come back from that, so I hit refresh. That seemed to do it. Then the Dashboard home page entries started disappearing. oh-oh, what’s going on? Then I remembered: IP address. Checked the router, and, yes, it had changed. Pop the new one into the browser, and all is well. Set fixed at the router.
Proves that the Dashboard leaves a ‘ghost in the machine’; lights are on, but nobody’s home.
Then I upgraded the firmware, which should sort out any issues with the RAID partitions. That took an age to return after completing, so again, I refreshed the page, and the new white UI appeared. The login page popped up, and then went blank. Refresh; same thing.
Oh yes: always do a 4-second reset after a firmware upgrade. That sorted it.]