Hi, first time poster

I setup an automated media server using unraid, arrs, jellyfin, jellyseerr etc,

the array consists of 2 8tb hdds connected to the host machine via usb (A laptop), there is no parity setup for the array (nothing irreplaceable on the drives)
downloads are written to a 1tb partition on a sata ssd while in progress and then written to the array upon completion, handled via qbittorrent
there is a 1tb ssd cache drive connected via USB, used for /system and /appdata

Functionality of all containers is fine, they all play well together, the only issue is that when downloads complete, moving them from the ssd to the array is giving me trouble

I understand that usb is not ideal for this application, but for the time being I am working with what I've got, for usb 3.0 I would expect tx speed of 30mbs minimum, but what seems to happen is that the speed of the tx starts are something like 500mbs, then slows to a crawl of 500kbs with high IOWAIT

I am unsure of what exactly is happening here, I have been troubleshooting for a little while, a throttled rsync tx will go for a while at 30mbs steady, but eventually ~10gb in will drop to the 500kbs rate,

From my reading it may be due to page caching of the writes from the partition to the array, which overloads the usb controller and leads to the dramatic slowdown, but I am unsure if that is the exact issue

Any insight into this issue and what might be done to resolve it are appreciated

media-diagnostics-20250406-2336.zip

Edited April 7, 20251 yr by gen_random_string

Both disks are SMR, but that still lower performance than expected, try writing to disk2 to see if it’s the same

I am a little taken aback,  with disk 2 there is that initial spike to 500mbs but it levels out and holds steady at approx 160mbs (tested and held steady throughout a 35gb tx), its especially surprising as I thought disk 2 was plugged into the usb 2.0 controller port, while disk one had a usb 3.0 port, does this indicate that there may be something wrong with disk1?

Note, both disks make up the array, but this is the first data written to disk2, disk1 was filling first

Edited April 7, 20251 yr by gen_random_string

1 hour ago, gen_random_string said:

does this indicate that there may be something wrong with disk1?

Most likely, and that specific disk model has been an issue before.

So the issue is more or less solved, I can only get that 160mbs consistent tx speed when only one drive is plugged in to that side of the laptop, both disk1 and 2 can achieve it, so the disks were ruled out, tried to replicate this on the other side with what should be a usb 3.0 port with everything else unplugged from that side, however I still only got 30mbs, and it would probably drop as well, but I didn't let it run long enough to find out

The good news is I can write to either disk 1 or 2 at an acceptable speed, and I tested reading from a drive connected to a port when it was only able to achieve 30mbs, and that doesn't drop to 500kbs in the same way writing to it would

It's not ideal, but the first 8tb drive isn't nearly full, so I will continue to write to one drive, I plan to upgrade to a more suitable host machine before its full, but even if I don't, filling one drive and then having it only be read from by jellyfin etc on a slow port while the other drive is filled at 150mbs on a good port is fine for me

 

I am unsure if I can mark this thread as solved or how to do so, any help with that is also appreciated

Edited April 7, 20251 yr by gen_random_string

10 hours ago, gen_random_string said:

tried to replicate this on the other side with what should be a usb 3.0 port with everything else unplugged from that side, however I still only got 30mbs

Could be a driver issue, post the output from

lsusb -vt

 

/: Bus 001.Port 001: Dev 001, Class=root_hub, Driver=xhci_hcd/16p, 480M
ID 1d6b:0002 Linux Foundation 2.0 root hub
Port 001: Dev 002, If O, Class=Hub, Driver=hub/4p, 480M
(D Obda:5411 Realtek Semiconductor Corp. RTS5411 Hub
port 004: Dev 003, If 0, Class=Mass Storage, Driver-usb-storage
480M
ID Obc2:2032 Seagate RSS LLC
Port 005: Dev 004, If O Class=Video, Driver=[none], 480M
ID 0408:a061 Quanta Computer, Inc
_ Port 005: Dev 004, If 1, Class=Video, Driver=[none], 480M
ID 0408:a061 Quanta Computer, Inc
I_ Port 014: Dev 005, If
C
Class-Wireless, Driver=btusb, 12M
ID 8087:0aaa Intel Corp. Bluetooth 9460/9560 Jefferson Peak (JfP)
|_ Port 014: Dev 005, If 1
Class=Wireless, Driver=btusb, 12M
[D 8087:0aaa Intel Corp. Bluetooth 9460/9560 Jefferson Peak (JfP.
/:
BUS 002 Port 001: Dev 001, Class=root.
hub,
Driver=xhci_hcd/8p, 10000M
ID 1d6b:0003 Linux Foundation 3.0 root hub
|_ Port 001: Dev 002, If 0 Class=Hub, Driver=hub/4p, 5000M
ID Obda:0411 Realtek Semiconductor Corp. Hub
Port 001: Dev 004 If
C class=Mass Storage Driver-usb-storage, 5000M
ID 0781:5597 SanDisk Corp
> Port 004: Dev 005 If 0
Class=Mass Storage Driver-usb-storage, 5000M
ID Obc2:2032 Seagate RSS LLC
I_ Port 002: Dev 003, If 0,
Class=Mass Storage, Driver=uas, 5000M
ID 04e8-:61f5 Samsung Electronics Co., Ltd Portable SSD T5

Seagate is not using the new uas driver, so that's not the issue, I'm afraid I don't have more ideas.

It's alright, performance is now steady 150mbs write isn't the best, but it's far better then it was when I started the thread, so I'll leave it be until I upgrade