I have and Aoostart with 4 drives. All the drives are identical size 12 Tb Seagate drives form GoHardDrives. One of them started throwing SMART errors. I ended up buying 5 more of the exact 12Tb Seagate hard drives. One to replace the one in the TrueNas and the others to put in a Synology I have. The TrueNas is saying that the new drive is too small to add the pool. I tried all the things in another post but a combination of it did not work and I am over my head. Here is the output of lsblk -b. sda is the drive I am trying to replace. I does look like the new drives are smaller than the existing one. I have tried three of the new drives and they all seem smaller than the old ones.
truenas_admin@TrueNAS[~]$ sudo lsblk -b
[sudo] password for truenas_admin:
NAME MAJ:MIN RM SIZE RO TYPE MOUNTPOINTS
loop1 7:1 0 382722048 1 loop
sda 8:0 0 12000138625024 0 disk <---
└─sda1 8:1 0 12000136527872 0 part <---
sdb 8:16 0 12001339219968 0 disk
└─sdb1 8:17 0 12001337147392 0 part
sdc 8:32 0 12001339219968 0 disk
└─sdc1 8:33 0 12001337147392 0 part
sdd 8:48 0 12001339219968 0 disk
└─sdd1 8:49 0 12001337147392 0 part
nvme1n1 259:0 0 256060514304 0 disk
├─nvme1n1p1 259:3 0 1048576 0 part
├─nvme1n1p2 259:4 0 536870912 0 part
└─nvme1n1p3 259:5 0 255520480768 0 part
nvme0n1 259:1 0 1000204886016 0 disk
└─nvme0n1p1 259:6 0 1000203091968 0 part
nvme2n1 259:2 0 1000204886016 0 disk
└─nvme2n1p1 259:7 0 1000203091968 0 part
Sorry about not putting it in the box. The offending drive is sdc.
truenas_admin@TrueNAS[~]$ lsblk -bo NAME,MODEL,ROTA,PTTYPE,TYPE,START,SIZE,PARTTYPENAME,PARTUUID
NAME MODEL ROTA PTTYPE TYPE START SIZE PARTTYPENAME PARTUUID
loop1 0 loop 382722048
sda ST12000NM0127 1 gpt disk 12001339219968
└─sda1 1 gpt part 2048 12001337147392 Solaris /usr & Apple ZFS 3fb77293-dbfc-42a7-808c-0ccb2850384d
sdb ST12000NM0127 1 gpt disk 12001339219968
└─sdb1 1 gpt part 2048 12001337147392 Solaris /usr & Apple ZFS 6eb8f3b4-2c12-424b-9601-6d79a2de1b88
sdc ST12000NM0127 1 gpt disk 12000138625024
└─sdc1 1 gpt part 2048 12000136527872 Solaris /usr & Apple ZFS 1a67906d-2b5e-482c-a1ef-e4d94df1e82d
sdd ST12000NM0127 1 gpt disk 12001339219968
└─sdd1 1 gpt part 2048 12001337147392 Solaris /usr & Apple ZFS 1eb89960-5213-478f-8be1-1ec570e6a343
nvme1n1 PC SN740 NVMe WD 256GB 0 gpt disk 256060514304
├─nvme1n1p1 0 gpt part 4096 1048576 BIOS boot 6d36a785-f75d-404a-938a-283e1395e349
├─nvme1n1p2 0 gpt part 6144 536870912 EFI System 987b1fbc-4359-4fd3-8529-ea64fd88f9e5
└─nvme1n1p3 0 gpt part 1054720 255520480768 Solaris /usr & Apple ZFS cae466c4-4184-49b6-be25-76b199b5c4b2
nvme2n1 WD_BLACK SN770 1TB 0 gpt disk 1000204886016
└─nvme2n1p1 0 gpt part 2048 1000203091968 Solaris /usr & Apple ZFS 626123a1-96d6-47d2-a5b9-c42fe0546092
nvme0n1 WD_BLACK SN770 1TB 0 gpt disk 1000204886016
└─nvme0n1p1 0 gpt part 2048 1000203091968 Solaris /usr & Apple ZFS acfe9a1f-0932-4668-a572-588e86e00365
I find it very odd indeed that two disks with identical model numbers should have different sizes. Whilst they both technically meet the 12TB (12 * 10^12) definition, I would expect a change in engineering to result in a (perhaps minor) change of model number.
It is, IMO, perfectly reasonable for users to expect to be able to resilver an existing disk to a new disk with exactly the same model number, and clearly this is not possible when the drives are slightly smaller. And this is likely equally true for hardware RAID as it is for ZFS mirrors or RAIDZ.
P.S. I cannot find any other mention of this on the internet. Are you 100% certain that both old and new drives are genuine ST12000NM0127? Do the disks’ physical labels seem identical in both design and quality?
I am not able to post photos. Maybe I am too new? The ones I bought were all purchased from GoHardDrives. Four in December, five last week. They are all the V7 like your photo shows. Exactly like your photo. The only difference is some are FW:003 and some are FW:004 The ones from December were about 12 TB the ones from last week are like 1.2 GB smaller. They all look identical, I mean I have only bought a dozen or two hard drives, so not an expert in telling a counterfeit from a real one. The stickers all seem to be place properly. The printing on them seems legitimate. Yeah I don’t know.
You know I started on TrueNAS with HexOS when they had their sale on licenses and that is what is showing SMART errors. TrueNas itself seems fine. I never should have just did TrueNAS from the beginning.
I guess in the future the best bet is to hook a drive up to a computer and run shell commands to see how big it actually is. Not sure what you can do when you order to make sure they are all the same size. As you wrote, the expectation is the same model number will avoid any problems.
This is why I think it’s shortsighted to remove the 2-GiB “swap” partition when creating or adding drives to a vdev. Even if you never use swap, it provides an essential “buffer” to mitigate against this very issue that the OP is experiencing.
I know SCALE completely disabled swap, even on the boot-pool. Did they also remove the default 2-GiB partition buffer as well?
Expand point #1 to read and understand my concerns about switching from Core to SCALE about this very issue.
Let’s say it was just above the minimum spec? The new disk would still be ~1-GiB smaller in capacity. It still cannot replace a failed disk in the current vdev.
The same problem persists.
A 2-GiB buffer partition at the start of every disk would have prevented this.
When the user is hit with the “not large enough” error, they would simply (temporarily) uncheck or disable the 2-GiB buffer for new/replacement drives, then try again, in which it will succeed.
They can then once again re-enable the 2-GiB option in the GUI.
Since this has been apparently removed in SCALE (or is no longer a default?), you get a nasty surprise when trying to replace a failed disk.
EDIT: @Protopia, you calculated based on the partition size, not the disk size.
The disk itself is 12000138625024 bytes, not 12000136527872 bytes. So the spec sheet and labeling for the drive is correct.
It is possible that the host protected area is larger on new, smaller disk. SATA standard allows changing the apparent size of a disk to be smaller than the maximum. Not sure if any of the extra HPA space is available as spare sectors.
See the manual page for hdparm and option -N to see if there is any HPA space.
If their is, it may be an indication that the drive is used, and not brand new.
I have a Synology with four 2 TB drives in it, so I am planning on using the too small 12 TB drives in there. I have two old QNAPs that I am decommissioning. Plus I have a DAS box attached to an AirPort Extreme I am using for a Time Machine backup. So in the end I have several boxes of discs where I can put in larger discs, So I don’t necessarily need to return them.
But yeah, I started in the NAS/Homelab thing with these QNAPS I got on sale from NewEgg. Then I read bout Synology so I got one of them. Then I wanted to learn something a little more complicated, so I got the Aoostar WTR which has a Ryzen and 4 drive bays to put TrueNAS on.
The biggest thing is I have is JellyFin and Audiobookshelf on the TrueNAS. Admittedly the Synology is easier to put stuff on because there is a website from Marius Bogdan Lixandru who has screenshots for every step to do. The documentation for TrueNAS makes a lot of assumptions that you are very experienced so I end up lost on how to do things. But I am plugging away and trying to learn.
In the end I am trying to self host as much as I can and de-google a bunch.
You are NOT alone in finding that TrueNAS and ZFS have a steepish learning curve.
Uncle Fester’s Basic TrueNAS Configuration Guide is an attempt to try to fill this gap - the original Core based guide is showing its age, and it still has some huge gaps, but there is also some good stuff there too.
It is a wiki, so if you are willing to contribute that would be welcome. You will need to ask @Dan for contributor’s access.
They already checked. It’s an exact match of the guaranteed sectors.
However, between revisions (of the same Seagate model HDD), the newer ones are about 1.2 GB smaller in capacity, even though they are sold as “12-TB”.
For 12-TB drives, a difference of 1.2 GB is nothing for casual users. But when it comes to managing arrays / ZFS vdevs, it can prevent you from replacing a failed drive.