Hard Drive SMART tests stuck at 90%

[minez]

Hi all,

I've just purchased 4 new hard drives (Seagate Ironwolf 10TB) and am going through the burn-in process as indicated in the resources section. On each of the 4 drives, I have run through a short SMART, Conveyance, extended SMART and full destructive badblocks burn in. All of these completed without error. To finish, I ran one more extended SMART test on each of the drives. 2 of the drives have completed the SMART test without error, but the other 2 drives are both stuck on 90% remaining. It has been approx. 24 hours since the test was initiated, so I would have expected all 4 drives to complete at similar times.

Just wondering if anyone has any suggestions on what should be done at this point? Should I continue to just wait and see if there is any progression on these drives, or force stop and restart the test? Anything I can check to see if there has been a fault?

 

[Chris Moore]

Just wondering if anyone has any suggestions on what should be done at this point? Should I continue to just wait and see if there is any progression on these drives, or force stop and restart the test? Anything I can check to see if there has been a fault?

If you interrupt the test it just records that the test was interrupted by host. If you let it run, it should finish, even if it takes a lot longer than it should. As long as it finishes, you should get some result to let you know what happened, but the long delay makes me suspect there is an error.
If you have two out of four drives with errors, it would be really bad statistically, because you shouldn't see that many bad drives. Infant mortality rates have historically been less than 5%.

I would wait if it was me.

 

[minez]

I would wait if it was me.

Happy to wait it out for a while longer - I'd be surprised if there were errors on 2 disks after completing all the previous tests successfully. Seems more like the test is stuck somehow. It's approximately an 840 minute test according to the estimates. If there isn't any progress in another 4 or 5 hours, I might abort and restart the test on the affected drives.

 

[Jailer]

Be patient. I had one of my schucked WD 8TB drives get stuck at 90% for 2 days and then it finally finished and hasn't been an issue since.

 

[minez]

Be patient.

No worries - I'll let it keep going. Will report back in a day or two, hopefully with a completed test!

 

[SaltyCoffee]

I know this is an old post, but if OP is still around, did they ever complete?

I have exactly the same issue with 4 Seagate Barracuda 5TB drives, except instead of 2/4 drives, I have all 4 drives getting stuck at 90% remaining repeatedly. And there doesn't seem to be any other %. Just goes straight to 90 and hangs there indefinitely (3 days was the longest I waited).

And this is after successfully running the short, conveyance, long (1st run), and badblocks tests with no issue, according to the sticky on this board. It's the final smartctl -t long test that is giving me hell. Should I just ignore it and proceed with provisioning my pool? This seems to be a very common issue across countless message boards with not 1 good answer. Top 2 are just-wait-it-out or RMA. I have very high confidence that I didn't just brick 4 drives with the badblocks test. Unless someone has had experience otherwise? I'm willing to listen.

EDIT: I forgot to mention that I am running these tests on a server doing literally nothing else but these tests. So the low priority thing that some give as the reason doesn't apply here. FWIW

Thanks for any $0.02

 

[winnielinnie]

This was a bug which I thought was fixed? Unless it's a similar-but-not-the-exact-same bug?

[NAS-114835] - iXsystems TrueNAS Jira

ixsystems.atlassian.net

Keep in mind that since the Jira migration, bug reports were copied over and broke the formatting of the reports and comments. (And even broke the attached images and screenshots.)

 

[SaltyCoffee]

Thanks. Although I think this is a similar, but different issue.

I'm curious though, even if I am not using the TN gui, do tests initiated from the CLI still use middlewared? I wonder if even with different details the underlying is still the same? Although, as you pointed out, this seems to be a resolved issue in the eyes of ix.

EDIT: Exploring your ticket, I ventured to one of the other linked tickets, which accurately described the issue I am experiencing. But iX's answer was ¯\_(ツ)_/¯ (translated from 'smart test is handled from hdd firmware'). I actually remember coming across this in my searching before, I guess I hadn't noticed it was in the iX support system. So, I guess that's that. It remains an internet mystery.

EDIT 2: I have noticed though that my most recent attempt is now at 80% remaining, so that is actual progress. Only took 4 tries but maybe this one will go through. Maybe that's the answer. If at first you don't succeed, try, try again.