Slow transfer speed on my QNAP TS-239 PRO

[topfuel]

Hello everybody.
I found an old QNAP TS-239 PRO sitting around so I decided to put into use again. Firmware is updated to latest version.

My issue is file transfer speed is very slow when copying from the NAS, around 6,3 MB/s over both SMB on windows 10 and NFS on linux. Odd thing is, copying to the NAS is around 37 MB/s. Why this huge difference in copying to and from the drive, where can I start searching for the problem?

The drives are one 2TB seagate Ironwolf pro (ST2000NE0025) and one 4TB Seagate NAS drive (ST4000VN000). Configured as JBOD.

Now only thing I did when installing is that it complained about cylinders being set wrong so I looked them up on Seagates website and set them to 16,383 as specifications say, using fdisk. Could this be it? I thought setting cylinders was not needed since LBA but silly me went and did it anyway.

 

[topfuel]

Realized only one of the dirves was on the official supported list, so I swapped them for 2 supported drives. Now its 6 MB/s both ways, raid 0 setup. Any ideas?

 

[dosborne]

What is the connection from the NAS and client device? What are you using to measure the speed?

Keep in mind the specs for that model. You shouldn't set your expectations too high.

 

[topfuel]

What is the connection from the NAS and client device? What are you using to measure the speed?

Keep in mind the specs for that model. You shouldn't set your expectations too high.

It's connected with a cat5e cable to a router and wifi to my laptop at the moment. To measure speed I just used scp in linux and looked what it was done. Transferred a few gigabytes in large files. On windows I just used explorer.

Yeah it's possible its actually the hardware that restricts it, having updated it to newest software on ancient hardware. I was hoping to just share folders on the network on it.

 

[dosborne]

Sounds like you are trying to measure the transfer speed over WiFi? If over ethernet, what are the connection speeds? Need to know what your infrastructure is expected to handle.

In any case, I'd suggest using a benchmark tool to measure actual performance. Also, there is almost always a huge difference in transferring many small files vs 1 large file.

I would not expect the drive itself to be the bottleneck.

Since copying from the NAS is slower, it could also be something as simple as antivirus scanning as it writes to the target. Hopefully, a benchmark utility would also remove that possibility from the equation?.

 

[degrub]

try iperf to measure bandwidth available. also check your lan connection lights are all showing 1000 Mbit/s
Do your testing over lan entirely at first and then move to add in wireless.

 

[topfuel]

Thank you all for the replies.

Yes I was at wifi at the time, but later connected a cable to the switch (a cheap 8 port A-link, gigabit) between the router and and got the same. All connections supposed to be gigabit. Before you mentioned it I did not think of even looking at led if it's 100 or 1000 Mbit/s because I assumed why would it not be gigabit? Sure enough, for some reason it connects at 100 Mbit/s from my thinkpad to the A-link. I'ts a T14 so it definitely has 1000 Mbit/s. I connected to the router instead (Ubiquiti EdgeRouter X) and voila I do get 1000 Mbit/s from there and now copying from the NAS is at full speed. Btw, the copying from the nas being slower I could not replicate again, so I strongly suspect the A-link being unreliable with my thinkpad. All is fine if I throw the A-link out of the equation.

What is curious is that my desktop is connected to the same A-link that only gave 100mbit to my laptop, and it is and always has been at gigabit speeds between the desktop and other computers connected to the A-link. I tested and confirmed it now too. Could the A-link just wrongfully identify the laptops ethernet port as 100 Mbit/s ? Also one of the ports on the A-link totally refuse to work with my thinkpad, but seems to work ok on the desktop. Need a new switch anyway so I won't bother messing around with it anymore in this scenario.

Thanks for mentionining iperf, it will make things easier in the future!

Edit: all cables cat5e too

 

[L&LD]

What is 'A-link'?

There are many instances of mismatched interoperability between equipment, the cable(s) used, and link speeds.

A different cable between the problem equipment may 'fix' the issue, or not. Or, a simple/cheap unmanaged switch may do the same thing too (between the problem devices).

 

[dosborne]

As L&LD mentioned, the cable is often the problem. Sometimes it's as simple as a dirty contact too.