How do I replace a 2TB non-failing disk (online, no errors) with a 4TB disk.
When I try to “offline” I get an error posted previously “[EZFS_NOREPLICAS] cannot offline … no valid replicas”
4 bays max. No free bay. All being used @61%
This is similar to a post Dec 2024 by davistw which involves it seems only one disk and solved by winnielinne in post #15
Thank you.
My system:
OS: TrueNAS Scale - ElectricEel-24.10.2
Motherboard: HP ProLiant MicroServer Gen8 System Board- 724495
CPU: Intel Xeon CPU E3-1230 V2 @ 3.30GHz
RAM: 16GB (2 x 8GB) PC3-12800E DDR3 UDIMM
Hard Drives: 4 drives Max no spare bay(1 x 2TB + 3 x 4TB = 14TB) Max 16TB
Used Space 61%
Boot Disk: Sandisk USB 32GB stick
Network controller: HPE Ethernet 1Gb 2-port 332i Adapter
It sounds like a pool with four vdevs, each vdev a single drive.
You can indeed not offline anything here. However.
If you can temporarily power and connect another drive, you can complete the mirror of the 2TB drive with another 4TB (attach 4TB to 2TB), wait for resilver, then offline the 2TB, then change the mirror back to a single drive vdev (detach offlined 2TB)
This setup has zero redundancy and I trust you are AOK with total data loss (and restore from backup) if one of the drives goes.
If you in future decide you want single redundancy and are OK with 12TB total storage:
Verify, and verify again, that your backup is solid
What about permissions and metadata? Make sure you have that marked. Ditto shares and such.
Destroy the current pool
Create a new pool, this time with just one vdev, 4 x 4TB raidz1. Capacity of one drive is parity, leaving you with 12TB.
Restore backup
Restore permissions and shares as needed
However, if this is used for VMs, databases, iSCSI etc (block storage basically, not file and media storage), then raidz1 ain’t a great idea. In that case if you want redundancy, plan for a bigger case, and make each single vdev a mirror … eight 4TB drives, four of which are the redundant storage.
If you are 100% certain that there isn’t a PCIe slot somewhere to stick an HBA into even temporarily, then you get to test your backup
The only way to change out a drive in a pool without redundancy, if you cannot create that redundancy for the duration of the swap, is to destroy the pool, make a new one with the new drives, and restore all data, permissions, and shares.
Do you want to share the use case - type of shares, is it media and files, is it VMs or block storage?
Is having no redundancy and relying on your backup a choice, or did it just happen that way and you’re now surprised you have no redundancy?
This may be an opportunity to build with at least single drive redundancy. You still need the backup - and it’d make future drive replacements faster and easier.
We’re making a couple assumptions here, based on your question and the use of the moniker “NTFS” where it didn’t belong, that the setup you have wasn’t a fully informed choice.
That may not be so. If it was a choice to run without redundancy, set us straight.
You might be able to remove a VDEV. It is very risky considering your pool and VDEV layout, from what we can see. You are really better off backing up your data elsewhere and creating a better pool and layout situation.
EDIT
At this point, you should check all your HD to make sure they don’t use SMR tech. You can post the drive models, if you want help.
If you are using SMB sharing, maybe you can back up to usb drives attached to another computer.
You have to look up each HD drive model and make sure they are not listed as using SMR, CMR conventional Magnetic Recording, is okay. SMR has been used in some NAS drives but is common in regular consumer drives.
SMR and ZFS / TrueNAS don’t work well together.
If you are backing up from TrueNAS using SMB and a large HD on Windows, that would work for all the data inside the SMB share.
There has been a lot of stuff posted here but actually very little hard data. @madin3 can you please open a shell and run the following commands and post the results for each command in a separate <> box:
This will give us hard facts about your drives and pools which will allow us to give you precise advice (and even commands to run) to help you achieve your goal.
(I think it possible that you can achieve this without needing to back up your data - but it will depend on what the above commands show us.)
