(SOLVED) WD drives are disproportionately slow in drivepool

[dcvr]

I currently have 2 separate data pools configured like this:

 

pool1 (jellyfin media) housed in terra-master D4-320 via usb:

(ssd cache)

seagate ironwolf pro 12tb

seagate ironwolf pro 16tb

wd red pro 16tb

 

pool 2 (mirrored archive) housed in a second terra-master D4-320 via usb:

seagate barracuda pro 6tb

wd enterprise-grade "white label" 14tb (shucked from elements)

seagate barracuda 24tb

 

i have tested each individual drive in crystaldiskmark and nothing out of the ordinary in terms of speed.

what i have noticed is on the first pool, the rebalancing speed on both seagate drives move files at 250+ MB/s, while the wd drive caps at ~20 MB/s

on the 2nd pool, the rebalancing speed of the barracuda pro is ~240 MB/s, the barracuda 24tb is ~220, and the wd drive is capped at ~20 MB/s.

 

this cant just be a coincidence for the 2 wd drives here, since i have been following this "issue" for a while now and in every single one of these similar these cases the problem happens on wd drives:

https://community.covecube.com/index.php?/topic/5375-why-is-balancing-so-slow-is-it-just-me-or-like-this-for-everyone/

https://www.reddit.com/r/DataHoarder/comments/ji7qoh/stablebit_drivepool_is_this_slow_or_typical/

https://community.covecube.com/index.php?/topic/28020-drive-pool-balancing-speed/

https://community.covecube.com/index.php?/topic/5752-extremely-slow-balance-speeds/

 

now with that said, i doubt every single user with wd drives experiences this issue or else it would have been a much bigger issue since the past 6+ years it has been happening. instead i think this could be some kind of bug that only affects some wd drives specifically.

 

Edit: additionally, i can confirm that this bug only happens during rebalance writes. the read speed is unaffected (i tested by copying a file residing on the wd disk from the drivepool, not the individual drive, and it reads at 280 MB/s)

Edit 2: just to be a bit more thorough, i placed 14GB worth of files on the wd drive manually. then ran rebalancer so that it reads those files from wd drive and writes them to seagates. the wd drive reads at 250 MB/s and seagates writing at 250 MB/s. so yes, this is a bug that only affects write during rebalnce on wd drives.

[dcvr]

UPDATE: I think I finally solved the issue.

First, I installed WD Kitfox and disabled APM on both drives through the software interface (not sure if this was contributing to the problem but i dont need or want this anyway)

Next, i noticed that write-cache was not enabled on either of the WD drives (was defaulted on "quick removal"). there is a setting to change the "write cache removal policy" in Kitfox, but it doesnt seem to work for me.

Instead i went into windows device manager, rolled out the disk drives category and double click on each affected drive > policy > "better performance" > write-cache enabled [x]

 

this did the trick, instantly all my drives are firing full speed 250+ MB/s

 

now, even tho it works and all, none of this makes any sense as to why it even works. how is it that if "drive cache enabled" was all that it took, windows explorer was still able to do full speed file transfers without it, while drivepool was not. 🤔 the world may never know

 

anyway, marking this as solved...

[InnocantX]

Thank you for posting this, because I was wondering, why my Drivepool ist that slow since I added 2 WD Red Pro 16 TB to my Drivepool. I started a Test with my Lyrion Music Server with a complete Scan of 6000 music files. With the Drivepool including the WD Red Pro 16TB Drive it took appr. 4 minutes. Without the WD Red Pro Drive in the same pool the scan lasted about 15 seconds. Very strange.