D4-320 disconnecting in linux, and other problems

[kcbucker]

I have had nothing bit trouble with the D4-320 and Ubuntu 22.04 linux. First the drives were randomly disconnecting, I thought I mostly solved this through a combination of turning off drive spindown, turning off usb autosuspend, and writing to a file and running lsusb every minute. However this was not even working 100% of the time, and the drives would still disconnect once in a while. The final straw was that I run ZFS, and when a scrub would kick off the machine would become almost unusable, and after a while the scrub would basically stop and start only accessing one drive, then eventually all the drives in the enclosure would go offline. I took the two active ZFS drives and put them back in a simple 2 drive USB 3.0 dock and everything works fine, including the ZFS scrub. The ZFS scrub continues to access both drives during the process, which is expected. I can't figure out what the problem is with the d4-320, has anyone seen anything like this? I consider it unusable for linux and need to look for a different solution.

[sherbibv]

Hello,
I have the same issues. I get random disconnects while running proxmox 8.3.3
The divice is connected via original cable to my m920x. I also disabled usb autosuspend for the DAS.
I get the following logs when it happens:

When running

dmesg -T | grep -i usb

i get:

[Sat Feb 8 10:51:51 2025] Modules linked in: kvm_intel(+) sunrpc binfmt_misc i915(+) kvm irqbypass drm_buddy drm_display_helper rapl cec rc_core i2c_algo_bit video wmi vmgenid i2c_piix4 joydev input_leds mac_hid serio_raw sch_fq_codel dm_multipath efi_pstore nfnetlink dmi_sysfs qemu_fw_cfg ip_tables x_tables autofs4 btrfs blake2b_generic raid10 uas usb_storage raid456 async_raid6_recov async_memcpy async_pq async_xor async_tx xor raid6_pq libcrc32c raid1 raid0 crct10dif_pclmul crc32_pclmul hid_generic polyval_clmulni polyval_generic ghash_clmulni_intel usbhid sha256_ssse3 hid bochs sha1_ssse3 drm_vram_helper drm_ttm_helper xhci_pci psmouse xhci_pci_renesas ttm pata_acpi floppy aesni_intel crypto_simd cryptd
[Sat Feb 8 10:51:52 2025] Modules linked in: kvm_intel(+) sunrpc binfmt_misc i915(+) kvm irqbypass drm_buddy drm_display_helper rapl cec rc_core i2c_algo_bit video wmi vmgenid i2c_piix4 joydev input_leds mac_hid serio_raw sch_fq_codel dm_multipath efi_pstore nfnetlink dmi_sysfs qemu_fw_cfg ip_tables x_tables autofs4 btrfs blake2b_generic raid10 uas usb_storage raid456 async_raid6_recov async_memcpy async_pq async_xor async_tx xor raid6_pq libcrc32c raid1 raid0 crct10dif_pclmul crc32_pclmul hid_generic polyval_clmulni polyval_generic ghash_clmulni_intel usbhid sha256_ssse3 hid bochs sha1_ssse3 drm_vram_helper drm_ttm_helper xhci_pci psmouse xhci_pci_renesas ttm pata_acpi floppy aesni_intel crypto_simd cryptd
[Sun Feb 9 00:50:30 2025] usb 3-1: stat urb: no pending cmd for uas-tag 37059
[Sun Feb 9 00:51:35 2025] usb 3-1: USB disconnect, device number 2

And

dmesg

i get:

2025-02-09T00:45:10,071987+02:00 br-f6ee40e736af: port 5(veth5f67713) entered forwarding state
2025-02-09T00:45:10,096289+02:00 eth1: renamed from veth8919cee
2025-02-09T00:45:10,113305+02:00 br-502f0a0f9660: port 3(veth1c64b87) entered blocking state
2025-02-09T00:45:10,113323+02:00 br-502f0a0f9660: port 3(veth1c64b87) entered forwarding state
2025-02-09T00:50:30,506176+02:00 usb 3-1: stat urb: no pending cmd for uas-tag 37059
2025-02-09T00:51:34,437766+02:00 sd 3:0:0:0: [sdb] tag#10 uas_eh_abort_handler 0 uas-tag 1 inflight: CMD
2025-02-09T00:51:34,437778+02:00 sd 3:0:0:0: [sdb] tag#10 CDB: Synchronize Cache(10) 35 00 00 00 00 00 00 00 00 00
2025-02-09T00:51:34,451818+02:00 scsi host3: uas_eh_device_reset_handler start
2025-02-09T00:51:35,267562+02:00 scsi host3: uas_eh_device_reset_handler FAILED err -19
2025-02-09T00:51:35,267579+02:00 sd 3:0:0:0: Device offlined - not ready after error recovery
2025-02-09T00:51:35,267621+02:00 sd 3:0:0:0: [sdb] tag#10 FAILED Result: hostbyte=DID_TIME_OUT driverbyte=DRIVER_OK cmd_age=64s
2025-02-09T00:51:35,267626+02:00 sd 3:0:0:0: [sdb] tag#10 CDB: Synchronize Cache(10) 35 00 00 00 00 00 00 00 00 00
2025-02-09T00:51:35,267631+02:00 I/O error, dev sdb, sector 0 op 0x1:(WRITE) flags 0x800 phys_seg 0 prio class 0
2025-02-09T00:51:35,269200+02:00 usb 3-1: USB disconnect, device number 2
2025-02-09T00:51:35,269770+02:00 device offline error, dev sdb, sector 0 op 0x1:(WRITE) flags 0x800 phys_seg 0 prio class 0
2025-02-09T00:51:35,541753+02:00 EXT4-fs (sdb1): shut down requested (2)
2025-02-09T00:51:35,542642+02:00 Aborting journal on device sdb1-8.
2025-02-09T00:51:35,543484+02:00 device offline error, dev sdb, sector 11714955264 op 0x1:(WRITE) flags 0x9800 phys_seg 1 prio class 0
2025-02-09T00:51:35,543510+02:00 device offline error, dev sdb, sector 11714955264 op 0x1:(WRITE) flags 0x9800 phys_seg 1 prio class 0
2025-02-09T00:51:35,543531+02:00 Buffer I/O error on dev sdb1, logical block 1464369152, lost sync page write
2025-02-09T00:51:35,543553+02:00 JBD2: I/O error when updating journal superblock for sdb1-8.
2025-02-09T00:51:35,637807+02:00 sd 3:0:0:0: [sdb] Synchronizing SCSI cache
2025-02-09T00:51:35,757887+02:00 sd 3:0:0:0: [sdb] Synchronize Cache(10) failed: Result: hostbyte=DID_ERROR driverbyte=DRIVER_OK
2025-02-09T00:52:19,348522+02:00 EXT4-fs warning (device sdb1): htree_dirblock_to_tree inode #136839169: lblock 0: comm PMS LibUpdater: error -5 reading directory block
2025-02-09T00:52:19,387168+02:00 EXT4-fs warning (device sdb1): htree_dirblock_to_tree inode #36700161: lblock 0: comm PMS LibUpdater: error -5 reading directory blo

Have no solution for now, seems that the issue is the actual product wince I used other DAS devices before but did not exhibit this behaviour.

[TMeric]

Please check if the USB cable is damaged or loose, and ensure the USB port is functioning properly. You can try using a different USB port to rule out any potential port issues. Also, make sure you are using the device's original power supply to ensure stable power delivery.

[sherbibv]

Hello,
I've done the update on a windows machine.
I have reset my proxmox node (reinstalled os)
I have changed the cable to the original one. I am using the original power supply.
I do not have auto suspend on for the device and I still get the errors like this.
If i disconnect manually the usb when its connected and running fine i get these logs:

Code: Select all

Feb 20 09:03:58 pve-1 kernel: usb 1-5: USB disconnect, device number 5
Feb 20 09:03:58 pve-1 kernel: usb 2-4: USB disconnect, device number 6
Feb 20 09:03:58 pve-1 kernel: usb 2-4.1: USB disconnect, device number 8
Feb 20 09:03:58 pve-1 kernel: usb 2-4.2: USB disconnect, device number 7
Feb 20 09:03:59 pve-1 QEMU[3720841]: kvm: libusb_release_interface: -4 [NO_DEVICE]
Feb 20 09:04:00 pve-1 QEMU[5649]: kvm: libusb_release_interface: -4 [NO_DEVICE]
Feb 20 09:04:08 pve-1 pvestatd[1298]: proxmox-backup-client failed: Error: datastore '5tb-share' is not mounted

Seems reasonable, all 4 devices disconnect instantly, but when the problem occurs (this can happen after 2,3 days) it has a different behaviour, it disconnects usb devices one by one (this should not be the case if they are connected over the same USB port)

Code: Select all

Feb 20 00:18:25 pve-1 kernel: usb 2-4.2: reset SuperSpeed Plus Gen 2x1 USB device number 4 using xhci_hcd
Feb 20 00:20:00 pve-1 kernel: usb 2-4.2: reset SuperSpeed Plus Gen 2x1 USB device number 4 using xhci_hcd
Feb 20 00:22:03 pve-1 QEMU[5649]: kvm: libusb_release_interface: -4 [NO_DEVICE]
Feb 20 00:22:03 pve-1 kernel: usb 2-4.2: USB disconnect, device number 4
Feb 20 00:22:07 pve-1 pvestatd[1298]: proxmox-backup-client failed: Error: datastore '5tb-share' is not mounted
...
Feb 20 00:50:48 pve-1 pvestatd[1298]: proxmox-backup-client failed: Error: datastore '5tb-share' is not mounted
Feb 20 00:50:56 pve-1 QEMU[5163]: kvm: libusb_release_interface: -4 [NO_DEVICE]
Feb 20 00:50:56 pve-1 kernel: usb 2-4.1: USB disconnect, device number 3
Feb 20 00:50:57 pve-1 pvestatd[1298]: proxmox-backup-client failed: Error: datastore '5tb-share' is not mounted

As you can see, one device disconnects and after a while the other one does the same (i have 2 hdds in the 4 bay das).
As much as I want, I cannot reliably use it for my business.

[btschirren]

I have a D2-320, but had basically the same problems.

The solution is to switch from the new/fast "uas" driver to the old/slow "usb-storage" driver. This will reduce throughput by about 10x, but it will work. In my case, all errors went away and zfs started working as expected (just very slow).

You will need to add a kernel parameter to disable "uas". In my case, it was "usb-storage.quirks=1352:a006:u". The exact method to add this parameter will depend on your linux distro. And the values (1352:a006) are specific to the D2-320. Use "lsusb" to find your values.

Its unfortunate it doesn't work with the new "uas" driver. I guess you get what you pay for...