Problems with the new NVME S.M.A.R.T. test

TrueNAS General
, , ,
1

I have a single NVME disk on which my boot pool is located. The drive mode is: KINGSTON SNV2S500G and so far I’ve had no issues with it. S.M.A.R.T. is enabled. I upgraded to ElectricEel-24.10.0.2 today and I noticed that I now have the option to perform S.M.A.R.T. test on the NVME from the GUI.

I started a SHORT test ~30 minutes ago which still hasn’t finished, which is quite strange IMO. The GUI console outputs the following message:

Manual S.M.A.R.T. Test
smart.test.progress 0.00%
nvme0n1
smartctl 7.4 2023-08-01 r5530 [x86_64-linux-6.6.44-production+truenas] (local build)
Copyright (C) 2002-23, Bruce Allen, Christian Franke, www.smartmontools.org

Read Self-test Log failed: Invalid Field in Command (0x002)

Any idea what may cause this or are these tests being outputed to a logfile?

Thanks.

Edit: Someone already found and posted the same issue here: NVMe SMART test (long) fails with error: Read Self-test Log failed: Invalid Field in Command (0x002) - #2 by oxyde

TL;DR - we still don’t have NVME S.M.A.R.T. tests via the GUI.

2

Save version, i can run smart from GUI

image
image833×426 12.3 KB

image
image706×427 11.4 KB

But look what is happening.

image
image1602×539 35.6 KB

Will be hard to monitor NVME disks like this

1 Like
3

I made a comment on the Announcement page for this version of EE which states that the NVME testing still does not work. It is a known problem. I was even asked to submit a bug report however it would be cancelled until smartmontools 7.5 arrives, they know about the problem and want to only use smartmontools to run the tests. At least we got iX to add the NVMe to the list. There are other software changes needed and smartmontools 7.5 needs to be released to fix smartd (what is causing the Invalid Field in Command).

But if you have some questions, please ask.

If you want to run a SMART test on your nvme drives in SCALE, the command is nvme device-self-test /dev/nvme0 -s 1 and this will run a Short test on nvme0. To run a Long test change “1” to “2”. You can setup a CRON Job to run these commands periodically as you see fit, or you can run my little script called Multi-Report.

1 Like
4

Thanks for the info!

It’s still interesting however how @oxyde managed to scan their NVME from the GUI… I guess it works to some degree.

5

That is a false indication, the test was never actually run.

6

Well damn. Isn’t it worrisome that the middleware lies about something as critical as a S.M.A.R.T. check?

7

Yeah, really strange that in our case the outcome is so different @scotrod .
BTW we will for sure be able to handle somehow with joe’s precious help (and like happen to us, probably will be usefull for a lot of other folks).

8

Another question came to my mind. Does the ‘failed’ S.M.A.R.T. attempt hurts the NVMEs in any way. Currently I have cron jobs that perform both LONG and SHORT (not in the same days) tests weekly for ALL disks. Should I exclude the NVME from this list until ix implement the fix for smartmontools?

9

don’t think it can hurt somehow the disk, but for sure is a good point: why let the system perform an action that will throw an error for sure (or like happen to me, start without throwing anything but still not perform the test).
I think i will disable it :thinking: just for be 100% sure
BTW did you start using the Joe’s multi report script?