Hi everyone,

I'm new to Unraid and have been troubleshooting a performance issue for several hours. At this point I've ruled out many of the obvious causes and would appreciate another set of eyes.

Hardware

• UGREEN DXP4800 GT

• Unraid 7.3.2

• 2 × WD Red Plus 8 TB SATA

  • Disk 1

  • Parity

• Cache: NVMe SSD (Btrfs)

• Mac mini running macOS 26.5.2

• Router: FRITZ!Box 6670 Cable

• NAS connected via 1 Gbit Ethernet

• Mac connected via built-in Gigabit Ethernet

Symptoms

Copying a large file (67 GB) from the Mac to Unraid:

• Transfer starts around 150–170 MB/s

• Drops to 0 MB/s for roughly 10–12 seconds

• Resumes

• Repeats throughout the transfer

Average throughput is only 30–60 MB/s.

This happens:

• writing directly to the protected array

• writing to the cache SSD

• with Finder

• with cp

• with scp

So it doesn't appear to be Finder-specific or cache-specific.

What I've tested

Local disk performance

On the NAS:

dd if=/dev/zero of=/mnt/disk1/testfile bs=1M count=10000 status=progress

Result:

~200 MB/s sustained.

This suggests the disks and filesystem are capable of normal write speeds.

Network

ethtool:

• 1000 Mb/s

• Full Duplex

• Link detected: yes

Router also reports both Mac and NAS linked at 1 Gbit.

System utilization

During transfers:

• CPU low

• RAM low

• Network not saturated

• Dashboard sometimes shows disk writes dropping to zero during the stalls.

iostat

Disk wait times appear normal (roughly 10–20 ms). I don't see obvious multi-second I/O waits.

sar

Network throughput averages roughly 30–35 MB/s during transfers.

SMART

SMART appears healthy.

Question

Has anyone seen similar behavior on the DXP4800 GT or with Unraid 7.3.2?

Any advice would be greatly appreciated.

Run an iPerf3 test from your windows pc to the server and confirm it’s not the networking.

This is very normal. Same performance on my DXP4800 Plus.

I have a DXP2800 that I can fully saturate at least 1Gbps. Which I’ve done for a full duration of 9TB during a data backup session.

This was strait to a btrfs raid1/mirror pool of 2x 12TB HDDs

Unraid to unraid via restic over sftp

The one thing that can dramatically cut speeds is unraids FUSE layer, if you can bypass FUSE with an exclusive share it can greatly speed up copying. With that said FUSE seems to only be affected by many small files being read/wrote. A single large file shouldn’t really be an issue.

Edited July 25Jul 25 by MowMdown

10 hours ago, MowMdown said:

Run an iPerf3 test from your windows pc to the server and confirm it’s not the networking.

Did that. From the Mac side I got about ~1400 retransmissions in about a 10s interval.

7 hours ago, MowMdown said:

The one thing that can dramatically cut speeds is unraids FUSE layer, if you can bypass FUSE with an exclusive share it can greatly speed up copying. With that said FUSE seems to only be affected by many small files being read/wrote. A single large file shouldn’t really be an issue.

Hmm.. Thing is, even on single large files, there's no difference in behavior when writing directly to the array or ssd cache pool.

8 hours ago, Led57 said:

This is very normal. Same performance on my DXP4800 Plus.

Can't imagine this to be normal to be honest. Especially as it results in average writing speeds of ~30mb/s.. That doesn't seem right at all.

I've since created a Ticket with official support. I'll update this should we reach a fix for the problem. In the meantime I am grateful for any further ideas to fix this! :)

2 hours ago, LrdSpn said:

From the Mac side I got about ~1400 retransmissions in about a 10s interval.

That sounds like a bad cable or a bad NIC port somewhere in the chain. This wont be an unraid issue.

Edited July 25Jul 25 by MowMdown

At this point I agree, that this is likely not an Unraid issue. I've changed cables, switched LAN ports on my router and made sure that every interface is communicating with the correct speeds (at least on paper).
There seems to be something, most likely on the Mac side, limiting my connection to about 100mbit.

The cables are brand new. So unless 3 cables I tried were all bad/damaged, that's not it.

do both devices and the router show link speeds negotiating at full speed?

Found the issue. I was transferring data from an external SSD that was connected to my Mac.

Not sure whats happening there since with the speed of the SSDs, it shouldn't make a huge difference, if any.

Also not sure why I get these low iperf3 results with lots of retransmissions, but transfers look fine and fast now from my Macs internal SSD.

Quick update:

I ran into this issue a few times more and it turns out that this is related to the Mac waking up from sleep mode.
I don't know if it's because it can't recover the SMB/NFS state, TCP session, Ethernet interface or whatever else. But there's clearly something jammed after waking the Mac from sleep. I randomly noticed because I restarted my Mac at some point and that solved the issue. And the issue is reproducible.

So if anyone else runs into that at some point. Restart your Mac/avoid using sleep mode.