Hey all,

As per the title, I'm sure many of you have seen this issue and it's really getting to me. At the moment, I setup an array of 6 - 4tb HDDs, mix of Seagate Barracudas and WD reds. The Seagates twirl at 5400 rpm and the WDs around 5900. (ST4000VN008 and DN004, WD40EFAX-68JH4N0 and N1s)

I have a 8TB Barracuda parity drive.

I run a Xeon e5-2698 with 64GB Ram, a 512GB nvme, 2.5Gb networking card.

So, I'm trying to empty the nvme cash which had 320GB of data on it. I'm using Unbalanced and set the NVME to write to all disks. lt started off great with sustained writes at 170MB/s or so and would be done in an hour or 2. That was almost 7 hours ago. The speed slowed down to 35MB after about 1 hour then steadily kept slowing to currently 10MB/s and sometimes dips to 1 or 2 MB/s So currently it shows 20GB left and 32 minutes to go. Well, it's been going up and down so who knows how much time is left.

I'm just wondering why it would take so long to go from an NVME attached to the mobo to move over to the HDs. I purchased 6 new SATA 3 cables thinking that some cables may be old and defective but that didn't change anything.

There are no Sector Errors on any disk.

The drives are set to never spin down. (however i'm contemplating covering pin 3 on the drives)

I set the Tunable MD write to Reconstruct Write

I even disabled Parity for now.

My CPUs stay around 1% or 2% usage and will sometimes go to 100% on one or 2 cores but will last about 10 seconds.

I rebooted the Unraid server which helps but seems to reset the copy from Unbalanced so I have to start over.

But this happens when copying files from SMB or NFS shares. Starts at 98MB/s then slows and will stay at 30MB/s then, after an hour or so, will drop to 10 and below.

This is really frustrating and I still have about 12TB of data to move over. I have seen many posts about this issue and various answers that seem to help me on the short term. I though the nvme was the answer as it let me copy to over 130MB/s but then totally slow when the NVME got to 90% full. Then it looked like it bypassed the NVME and went straight to the array. The nvme would never dump it's content to the array. That's why I'm using to Unbalanced to empty it then I will remove the NVME as a cache and let it sit until I get a better result.

My shares are (well, were set) set to use the Cache then copy to the Array.

I am not sure if the array is limited to the speed of the Parity disk or not. All drives individually seem to work properly.

What am I missing.

Thanks all for any help.

Jacques

Edited November 6, 2025Nov 6 by Ayanefan

9 hours ago, Ayanefan said:

WD40EFAX

These are SMR, so that could be a problem.

I removed the cache drives last night and just let it go. This morning I started transferring 4TB of data over and it has been steady at 88ish MB/s . I did the same this afternoon with another 2TB. Same consistent speed. I don't know. Maybe it was fixing parity while I was moving the files from cache to drive? So many questions still remain. I was about to blame the X99 mobo and maybe some cheap electronics. At least it's consistent, just not super fast as I expected over a 2.5gbit line. It's averaging 680ish mebits/s . So today I will:

• Move my 8TB drive data over (maybe it will take more than a day)

• Add a 3TB spinner and a 2TB SSD drive to use as a Pool and not array (VMs and such)

• Add some more power to the Sata line.

• Order a CPU fan from Amazon to get the 2nd CPU connected.