How to check if HDD is failing

Posted by sdns575@reddit | linuxadmin | View on Reddit | 5 comments

Hi,

on my personal backup server (@home) I have an mdadm raid5 with 3x3TB wd red (I checked they are CMR).

One disk get detached from the array, I tried to read it but after some days it get detached again. I get error about speed level decrease from 6.0 gb/s to 3.0 gb/s

I checked smart logs and nothing is reported. I run badblocks to check if some block is gone but it is clean.

There is a way to check the connection port of the disk? I tried to change sata cable and sata port but it got the same message. At this point I don't know if is the motherboard sata controller or the disk itself.

I can attach the disk on another machine, but don't know what test runs to check this problem.

Any help is appreciated.

Thank you in advance

[-]

alpha417@reddit

I'm sure you have backups, right?

... right?

freightcar@reddit

Is the disk interface dying?

Sure sounds that way. Take a look at smartctl --all on that device, if it doesn't see issues, then I think you're on the right track.

sdns575@reddit (OP)

Hi and thank you for your answer.

smartctl does not report nothing bad.

There are other thing that I can check?

Seems to me you've already been pretty thorough, swapping cables and host port, even the host itself, testing it under load. The kernel messages you shared do point to a bus problem.

At this point I would just replace the disk, since it's part of a RAID5, if I understood you correctly.

If you really really want to keep the disk for whatever reason, there are companies that will replace the logic board of a disk for you, which might resolve the issue, but just buying a replacement disk will be a lot cheaper and easier, it seems to me.

Thank you for your advice