On 2026-09-14 10:37, Andy Smith wrote:
Hi,
[Copying Gary because they don't seem to be subscribed to the list so I
do not know if they see replies that are only on the list.]
On Mon, Sep 14, 2026 at 07:40:22AM +0200, Michel Verdier wrote:
On 2026-09-13, Gary Dale wrote:
I can't fsck the partition obviously since it doesn't have a file system -
just the mdadm superblock, if zeroing it left anything.
Did you try to erase/recreate the partition, with fdisk or another tool ?
What would be the theory behind why that would fix anything? That's just
going to be a small amount of writes to the MBR or the GPT at the
start/end of the entire disk, not the area of the problem partition.
>From what I understand the issue is that mdadm reads the part of the
disk for this partition and gets an I/O error. Gary rebooted; I'm not
sure if it still happens after the reboot.
Gary said they did a SMART long self-test and it passed. That's
encouraging, but I have had the SMART self-tests pass even while there
were actually unreadable sectors on the drive. I don't know why that
happened, but it did.
I did suggest a read of the whole disk with "dd" which helps to rule out
issues.
If there is a repeatable bad sector or small number of bad sectors on
the drive, but the number of bad sectors isn't growing, I have had
success with using "hdparm" to force a write to the problem sector(s),
which either fixes the problem or make the drive remap to a spare
sector.
I/O errors are of course not always hardware problems. If things are
working after a reboot then the root cause might have been some software
thing and we may never know what it was.
Thanks,
Andy
1. I am subscribed to the list.
2. the problem has survived reboots.
3. can you explain your use of hdparm? It's not obvious to me how to
use it as you described.
Thanks.
