HDD read errors

[jensrobot]

Hello

I recently got some read errors (32 in total) on one of my drives. "kernel: md: disk6 read error, sector=1578945552" and so on.

I've run an extended test that passed, and SMART doesn't seem to report any errors. I'm currently doing a parity check too.

Any advice on how serious this is, and how to proceed would be appreciated

skabet-diagnostics-20211202-1819.zip

[JorgeB]

It's logged as a disk problem, but these can be intermittent, and since the extended SMART test passed the disk is OK for now, keep monitoring.

[jensrobot]

5 minutes ago, JorgeB said:

It's logged as a disk problem, but these can be intermittent, and since the extended SMART test passed the disk is OK for now, keep monitoring.

JorgeB to the rescue  

Thx for swift reply, ill keep monitoring

[jensrobot]

sadly my server has now turned itself off during parity check, and its starting with the array stopped. This happened 2½ months ago also during a parity check, but I thought it was fixed when I got a new PSU (Super Flower Leadex III 550W) about 2 months ago. I wanna try and change the sata cable for disk6, I think the problems started after I added this disk.

Is there any risk of data loss from the parity check crash in itself? I'm thinking its ok because it only reads data... I'm looking for a way to replicate this crash for troubleshooting purposes without risk of data loss

EDIT: Btw i started array and its seems ok - green lights

Edited December 3, 2021 by jensrobot

[jbquintal]

I had the exact same problem... drives were throwing random read errors (mostly when I try accessing a spun down drive).  Extended SMART tests didn't show any problem with the drives, and all the unreadable sectors were never the same. I tried swapping out two drives, two controllers, and another set of cables... it ended being a bad PSU and now one of my drives have an unmountable FS.

If the drives are good, it's probably the cable, controller, or in my case, the PSU.

[trurl]

6 hours ago, jensrobot said:

Is there any risk of data loss from the parity check crash in itself? I'm thinking its ok because it only reads data...

A non-correcting parity check only reads data. An unclean shutdown can sometimes result in parity sync errors though. And it's even possible a crash could result in incomplete writes to a data disk if anything was being written at the time.

[jensrobot]

Just a quick update.

I ran 3 parity checks, and no shutdowns! Sadly 168 read errors on disk6 again during 2nd pass

Smart still reports nothing wrong  

[JorgeB]

6 hours ago, jensrobot said:

Sadly 168 read errors on disk6 again during 2nd pass

You should replace it then, or at least swap both cables/slot with another one to rule that out and then replace if more errors.

[jensrobot]

21 hours ago, JorgeB said:

You should replace it then, or at least swap both cables/slot with another one to rule that out and then replace if more errors.

I'll switch ports next. The motherboard actually have 2 different sata controllers built in (there is 2 extra slots next to the group of i think 6 ports), and i just notice disk6 is the only one on the secondary controller. I'll report back