I've been consolidating 2 arrays into 1 by mounting the 2nd arrays drives as unassigned drives and then rcloning the data into the array. This starts off pretty good and I get ~200MB/s writes when using 16 transfers concurrently. These are all large media files so not a ton of itty bitty things happening. After about 10-15 minutes, the same transfer (sometimes on the same files depending on size) will drop to 30-50MB/s. The destination of the copy is all going to 1 drive on the array because of high-water, but it still started off fine at ~200MB/s and it just progressively gets slower till it stabilizes at 30-50MB/s. The destination of the copy is not going to any cache pool so its not like its filling cache and then switching to array. 

I also noticed this with my parity checks. At the beginning of the parity check, its flying at ~200+MB/s and then by the end the average is ~50MB/s and its taken ~3 days. I thought maybe it was drive bay or HBA cable, but I swapped drive bays today and still seeing the same behavior. 

 

Cache is already being used for vms and docker so its not like there are competing writes from those.
 

stylophora-diagnostics-20240108-1422.zip

Do you have “turbo Write” mode enabled?   If not then I expect the initial burst of speed happens while data is being cached to RAM and then the 30-50 MBps speed that continue is not atypical when writing to a parity protected array.  This section of the online documentation accessible via the ‘Manual’ link at the bottom of the GUI or the DOCS link at the top of each forum page describes what is going on.   The Unraid OS->Manual section in particular covers most features of the current Unraid release.