Slow rebuild of /dev/md322

I’m performing a one-by-one drive upgrade on my TS-832PX, and I’ve replaced the 1st 8TB drive with a new 16TB drive. mdstat says that md1 is “resync=DELAYED”, and it’s currently rebuilding /dev/md322. The problem is that the speed is REALLY slow. Like, 2048K - 6192K per second. Is this normal? The last time I did an upgrade was right after I got it and swapped out some 2TB drives for the 8TB drive’s I’ve been using for the past 4 years…and I don’t remember it taking so long for the main data array to be rebuilt.

I DID get an error about not being able to access the device, but mdadm –examine /dev/sdi3 says that the superblock is clean, and that the array is rebuilding.

Is it safe to cancel out of the replace disks one-by-one dialogue to check disk health?

Ok, I’m just impatient. md1 is now rebuilding at something like 148MB/s.

I’m just confused about md322. The mdstat shows that there are 2 members, and 6 spares. So why didn’t the system start rebuilding the array on the next available spare disk the moment I pulled the drive out of the slot?

I can only assume that the slow speed was because the 5th partition is at the end of the drive, and the heads are having to transit across all cylinders?

Maybe swap was in use that time

https://forum.qnap.com/viewtopic.php?t=171841

…most likely:
Filename Type Size Used Priority
/dev/md321 partition 6702592 517696 -1
/dev/md256 partition 530048 0 -2
/dev/md322 partition 6702592 0 -3
/share/CACHEDEV1_DATA/.swap/qnap_swap file 16777152 0 -4

You’d think it would prioritize the SSD swap space… :confused:

What is your RAID resync priority set to?

It’s set to Resync First.

Odd, /proc/sys/dev/raid/speed_limit_min is set to the default of 1000.