My zfs pool shown the above condition after a long automatic scrub and the pool is not healthy. It reported one error in the last scan but also said the pool is “online, no errors”. I check each dish in the pool and they all seem healthy. Should I do another scrub? What else should I check?
Please advise,
Use the following command from the Unix Shell:
zpool status -v
This will list any issues with your pool.
Also, in general, when asking for help, it is always helpful to list complete hardware and how the storage devices are wired to the system board, (mother board SATA, SAS, etc…).
My system is based on a MSI PRO B550M-VC WIFI motherboard with 8 SATA connectors. I also added a M.2 SATA adapter with 6 SATA interfaces. Boot Pool is a mirror-0 (healthy). Pool 1 is a 1 x MIRROR | 2 wide | Mixed Capacity (healthy). Pool in question is a 1 x RAIDZ2 | 5 wide | 5.46 TiB. TrueNAS version 25.10.2.1.
zpool status -v shows the following
pool: Pool-M
state: ONLINE
status: One or more devices has experienced an error resulting in data
corruption. Applications may be affected.
action: Restore the file in question if possible. Otherwise restore the
entire pool from backup.
see: Message ID: ZFS-8000-8A — OpenZFS documentation
scan: scrub repaired 0B in 07:07:12 with 1 errors on Mon Mar 30 05:31:33 2026
config:
NAME STATE READ WRITE CKSUM
Pool-M ONLINE 0 0 0
raidz2-0 ONLINE 0 0 0
42fedc97-0910-4106-a13b-479d9bb59d6c ONLINE 0 0 4
96fbdf6b-7389-4add-9664-7649c28050d3 ONLINE 0 0 4
0cfa63e7-195e-4b99-9820-7a08b3788b23 ONLINE 0 0 4
4e689efa-bf00-4887-98d6-ddfd21aa6f60 ONLINE 0 0 4
024b418a-7884-441f-ae24-68bd1d58795c ONLINE 0 0 4
Not sure why all 5 disks show CHSUM error. After issuing sudo zpool status -v, one file shown permanent error. I guess that is why the checksum errors. I will try to restore that file, do another scrub to see. There is no way I can replaced all the disks in the pool.
Thank you for the help.
Is the pool with the errors on the motherboard connectors or on that m.2 card ?
4 on the motherboard and 1 on the m.2 card
I am in the process of doing smartctl on each disk.
Its not the disks. Checksum errors often stem from a faulty or overheating controller, bad cabling, bad PSU or even faulty RAM.
While your at it, get rid of that m.2 adapter. They are bad, just bad.
Call me confused…
In your postings above you said one file was corrupt.
Did you delete the offending file or restore it?
Did you run a zpool clear Pool-M to remove the recorded failures?
Did you run another zpool status -v Pool-M to verify the CKSUM errors were cleared?
Did you run zpool scrub Pool-M to run another scrub?
And finally, once the scrub was completed did you run zpool status -v Pool-M again to verify the errors were gone?
I have said this before, and I will say it again, I’m not the ZFS expert but it seems that all the drives had CKSUM errors because of the one corrupt file. I won’t pretend to explain why, it is just a guess on my part.
Next, if you continue to see CKSUM errors within the week or two, then I’d start to question my hardware.
If your Pool-M is connected to the NVMe to SATA adapter, that “might” be a problem but I would first just see how things go. File corruption across all the drives seems a bit off to me.
You might consider running MemTest86+ for 5 complete passes and then a CPU stress test such as Prime95 for 4 hours to ensure your system stability, at least what you can test.
Last thing, is your system on an UPS?
the motherboard is new. the m.2 is not. i did consider NOT using it. PCIe SATA adapter are cheap now. i will replace the m.2 shortly.
These are not much better. For the same price - around 30 USD - you can get a HBA. E.g a 9300-8i. Flash it to IT mode and strap a fan to it and you have a rock solid connection.
Have done any of this yet. I ran memtest86+ (10 times) when I replace the motherboard three weeks ago. I just did the smartctl on all disks on pool-m and they all passed. I am trying too see if I can just move the file off to a USB drive (no luck, file i/o error). I think I will replace the m.2 adapter first (or delete the file first, thinking). This is just a personal NAS, I am not that worry.
The file is corrupt and that is why. You should be able to delete it.
The motherboard is mATX with one X16 slot, the rest are X1. I can’t use a HBA.
Just deleted the file. I will replace the m.2 adapter first then do the commands in the following order.
Is there anything already in use for the x16 slot?.. I swear, if it is a gpu for transcoding, that you’d never notice the performance drop between x16 and x1 before you’d hit a different limiting factor (vram, artificial driver limits, etc) unless you’re running 6 different 4k hdr streams with color correction to 1080p.
Edit: if it is because they are physical x1 slots without an open back, there are open back pcie x1 risers for sale now, no need to dremel like I have done in the past
just an old Gforce GT1030 for console output to a monitor. Are you saying I can move the gpu to a X1 slot and free up the X16 slot? Interesting thing is the X1 slots are all full X16 connectors populated with only X1 pins. I can easily do the move, no dremel.
Found this in Amazon, LSI 9300-8i Controller HBA Card, LSI 3008 SATA SAS Host Bus Adapter. 2 * SFF-8643 Internal 12Gbps, PCI-e 3.0 x8. IT Mode Firmware. Non RAID; for $61. It has 4 SATA connectors which is more than enough. Any thought?
I’m 99% sure it would work. There are some possible concerns with wattage output from pcie x 1… but the 1030 is limited to what? 30 watts? But for display out of the TrueNAS terminal I don’t think you’ll run into issues.
You could likely find the same HBA for cheaper on ebay with the same chance of it being genuine - otherwise I have nothing against the choice. I guess amazon is less painful in case of returns needed. Slap a fan ontop of it when it is in your case.
If not mistaken it’d be 4 sata connections per SFF-8643, of which it has two - so you’ll get up to 8 additional sata port to play with!
Just make sure it comes with the SFF-8643 to sata wires
B550 does not provide as much. Spec sheets states that 4 ports are handled by an ASM1064, which is possibly OK-ish but old. The M.2 SATA adapter is possibly NOT OK if a port multiplier is involved; remove it.
if your mainboard can post without it… attach the gpu on need 
In the last 2 year that i run totally headless, i needed to put a vga only 2 times.
The chineses riser x1 to x16 (very fashionable on mining period) can also be usefull to have, so you don’t have to “free” the x16 slot anyway
That is why it is $61…. no cable.
Went thru my old inventory and found X4 PCIe (most likely PCIe 2.0) adapter with 6 SATA ports. I am going to use it to replace the m.2 adapter by moving the gpu to the x1 slot.
Continue HBA search, and you’re right, eBay return is painful.
Thank you, guys
Won’t lie - it’ll likely be on par with the m.2 port multiplier. Do try to grab an HBA, there are good sellers on eBay. Normally I’d recommend Art of Server, but seems they on vacation until April 5th.
How many drives are there?
I count 2 in a mirror, 5 in raidz1, and there’s possibly a SATA boot drive. That would be 8. Why the extra M.2 or PCIe x4 adapter?