4KN drive issues

[foureight84]

I am hoping someone could help me understand what's going on. I have two HGST HUS726060ALN610 6TB 4kn drives and I'm trying to create a btrfs raid1 (in Ubuntu 22.04).

fdisk -l:

Disk /dev/sdc: 5.46 TiB, 6001175126016 bytes, 1465130646 sectors
Disk model: 726060ALN610
Units: sectors of 1 * 4096 = 4096 bytes
Sector size (logical/physical): 4096 bytes / 4096 bytes
I/O size (minimum/optimal): 4096 bytes / 4096 bytes


Disk /dev/sdd: 5.46 TiB, 6001175126016 bytes, 1465130646 sectors
Disk model: 726060ALN610
Units: sectors of 1 * 4096 = 4096 bytes
Sector size (logical/physical): 4096 bytes / 4096 bytes
I/O size (minimum/optimal): 4096 bytes / 4096 bytes

First I've noticed that mkfs.btrfs -d raid1 -m raid1 did not complete in over an hour, I didn't check for dmesg errors at the time but saw console output that the journalctl could not restart. I ended up doing a hard reset and wiped the drives. This time around I decided to create partitions on the drives first using cfdisk. It was while writing partition to one of the harddrives that started checking dmesg. This is the error I received:

[  897.888213] sd 2:0:0:0: [sdc] tag#0 CDB: Write(10) 2a 00 57 54 1e 91 00 00 01 00
[  897.888474] sd 2:0:0:0: [sdc] Unaligned partial completion (resid=3584, sector_sz=4096)

I've not worked with 4kn drives before but I was under the impression that they are fully supported since kernel 2.6.x? The other thing is that the drives are in a USB enclosure (SYBA SY-ENC50104) using the JMicron JMS567 chipset. I wasn't able to find any mentions of compatibility issues and the drives seem to work fine on Windows in the same enclosure.

 

[Sean Ho]

pretty sure that's a sector size mismatch. Is it in 512e mode? 4096 - 3584 = 512

 

[foureight84]

pretty sure that's a sector size mismatch. Is it in 512e mode? 4096 - 3584 = 512

It doesn't seem like it. fdisk shows that both logical and physical sectors are 4096.

This is the lsblk -t output:

NAME         ALIGNMENT MIN-IO OPT-IO PHY-SEC LOG-SEC ROTA SCHED       RQ-SIZE  RA WSAME
sda                  0    512      0     512     512    0 mq-deadline      64 128    0B
├─sda1               0    512      0     512     512    0 mq-deadline      64 128    0B
└─sda2               0    512      0     512     512    0 mq-deadline      64 128    0B
sdb                  0   4096      0    4096     512    1 mq-deadline      60 128   32M
sdc                  0   4096      0    4096    4096    1 mq-deadline       2 128    0B
sdd                  0   4096      0    4096    4096    1 mq-deadline       2 128    0B
mmcblk0              0    512      0     512     512    0 mq-deadline      64 128    0B
mmcblk0boot0         0    512      0     512     512    0 mq-deadline      64 128    0B
mmcblk0boot1         0    512      0     512     512    0 mq-deadline      64 128    0B

 

[CyklonDX]

you need to run your command with proper sector sizing.
mkfs.btrfs --sectorsize.

 

[foureight84]

you need to run your command with proper sector sizing.
mkfs.btrfs --sectorsize.

Let me give that a try. I thought that was autodetected

 

[foureight84]

you need to run your command with proper sector sizing.
mkfs.btrfs --sectorsize.

That doesn't work. dmesg error:

[ 1992.464436] systemd-journald[408]: /dev/kmsg buffer overrun, some messages lost.
[ 1992.464721] sd 2:0:0:0: [sdc] Unaligned partial completion (resid=25088, sector_sz=4096)
[ 1992.464724] sd 2:0:0:0: [sdc] tag#0 CDB: Write(10) 2a 00 00 00 00 f9 00 00 07 00

 

[i386]

I would say it's the sata to USB Controller

 

[foureight84]

I would say it's the sata to USB Controller

Hmm would it be more of an issue of the chipset support in linux? Since I am able to format the drive just fine on Windows to NTFS.

 

[CyklonDX]

Windows format is always using 512 sector format, and GPT hardly takes any advantage of 4Kn.

Can you see if this still happens when you use disk 512 sector, and then retry formatting in brtfs as 512? (just to exclude any other potential issues)

(might be related)

Linux-Kernel Archive: Re: sd 6:0:0:0: [sdb] Unaligned partial completion (resid=12, sector_sz=512)

 

[foureight84]

Windows format is always using 512 sector format, and GPT hardly takes any advantage of 4Kn.

Can you see if this still happens when you use disk 512 sector, and then retry formatting in brtfs as 512? (just to exclude any other potential issues)

(might be related)

Linux-Kernel Archive: Re: sd 6:0:0:0: [sdb] Unaligned partial completion (resid=12, sector_sz=512)

Can you swap 4kn to 512e when they're native 4kn? To my knowledge, these drives do not have 512e mode.

 

[foureight84]

I would say it's the sata to USB Controller

Looks like you're right. I am seeing various issues with JMS567 UAS support in Linux but no specific case of anyone trying 4Kn drives on this particular bridge controller.

 

[foureight84]

$ sudo sg_format --size=512 -v /dev/sdc
    HGST HUS  726060ALN610      0103   peripheral_type: disk [0x0]
      PROTECT=0
      Unit serial number: DD202206140045
      LU name: 3044202206140045
    mode sense(10) cdb: [5a 00 01 00 00 00 00 00 fc 00]
mode sense(10):
Fixed format, current; Sense key: Illegal Request
Additional sense: Invalid field in cdb
bad field in MODE SENSE (10) [mode_page 1 not supported?]

There's no way to change this drive to 512b. It looks like the issue might be with the SATA to USB bridge supporting 4Kn on linux.

 

[foureight84]

Alright looks like I have it working now. JMS567 definitely doesn't like 4Kn drives under Linux.

I was able to reformat the drives to 512b blocks using WDC Hugo 7.3.2 https://forums.servethehome.com/index.php?threads/hugo-for-linux.34320/#post-316072

hugo f -m HUS726060ALN610 -b 512 --fastformat

 

[Sean Ho]

Excellent, glad to hear you got it sorted!