PLEASE HELP - $18,000 POS UNRAID RIG

[plxmediasvr]

I need help majorly bad.

 

Im to the point where I am done with UNRAID and go back to plain old Windows. 

 

Unraid 6.8.3

 

Specs:

TRipper 3960x 24/48

256GB DDR4

Parity: 2- 16TB Seagate Exos

Array: 11- 16TB Seagate Exos

Cache: 4- 2TB Firecuda 520 Raid0 -8tb

GPU0: RTX 4000

GPU1: RADEON PRO W5700

 

 

I am trying to move files from one hard drive to the other and it works for like 10 to 25gb and then Krusader says "stalled".

I have even tried creating a VM with virtio everything enabled, with Redhat LAN showing 100gb/s and itll start at 250mb/s and then go back to 30mb/s occasionally it will go to 200 again and then stay at 30mb/s for like 80% of the time. To transfer 100gb it starts at saying 7 minutes and then sometime after 10gb to 25gb it slows down, and it changes the estimated time to 1 1/2 hours, and then shows stalled as the message in Krusader.

 

I have thus far also tried

All files/folders/shares/drives in question are set to root 777 just to rule that out

changed destination drive to use cache 1st, and mover to move. didnt change. Xfer still failed out after 25gb.

Tips and Tweaks Plugin set for Performance

Ryzen Enhancements Enabled

Changed the RAM from 10%/20% to 1%/2% since we are dealing with 256gb. Krusader stalled at 5gb transferred.

So then I changed it to 50%/80% and nothing changed but Krusader went back to stalling at 25gb.

So the RAM was put back to default 10%/20%

Global Share Direct IO enabled

Tunable scheduler to SFQ\BFQ as they are Exos Drives to rule out cache mq_deadline

Tunable MD Queue Limit to 85% so IO has the ability to nearly max the drive out.

Stripes set to 4096, but my understanding is thats only for Parity, not drive to drive xfer

TUNABLE set to RECONSTRUCT WRITE (TURBO MODE)

Changing Krusader to /mnt/cache/appdata along with all apps instead of mnt/user/appdata

Changed Docker to /mnt/cache/appdata instead of mnt/user/appdata

Stopped and disabled all other Docker Containers except for Krusader

Since I have cores to play with, and im sittting at 48 cores,

i can go against the standard and just pin docker cores without affecting cpu usage whilst guaranteeing cpu usage isnt the issue

Krusader pinned at 20 Cores / 48 Threads - and yes they are properly assigned (4/28, 5/29,-- ,-- ,-- 23/47 etc etc)

i even did the tunable test thing, even though that still seems to point to parity, and ran the generic, and my numbers did not change

from what the test stated if the numbers dont fluctuate then theres no need to run the full day or two test.

 

When I first started UNRAID, it took nearly 5 weeks to Xfer 80TB when in Windows it would have Taken only a week or so at 12 to 16tb per 24hours @250mb/s.

Now that the data is on the drives UNRAID itself works. Sonarr/Radarr/Sab/Plex/Handbrake all work. No issues with speed. Its only when I want to physically move the data around.

 

What am I missing here? how does this expensive of a rig have these kind of problems?

In Windows, on these exact same drives, i can push 250-260mb/s continuous transfer speeds.

 

Even my pcie 4.0 drives dont transfer at 4000mb/s the average i get is 600mb/s total (150mb/s x4) doing tasks. However that goes to 8000mb/s (2000x4) running SCRUB BTRFS

but 600mb/s daily use and 8000mb/s using SCRUB BTRFS is a far cry from 16,000MB/s for 4 raided gen4 nvme i should be getting

 

 

[falconexe]

1 hour ago, plxmediasvr said:

I need help majorly bad.

2 Words. UnBalance. I've moved literally hundreds of TB with it and have never had an issue. My largest combined transfer was re-dispersing 100TB, and it worked like a champ. It is just a GUI for CLI, but it is really great.

 

You may have some kind of hardware issue, but if you are getting stalls after 25GB, that is NOT normal. Try installing the unBalance plugin, head over to the webpage (YourIP:6238), and give it a shot. If you do, I am very curious if you are still having the same issue.

Edited August 28, 2020 by falconexe

[falconexe]

Sorry you are having so many issues. I also have a very expensive rig... I have seen some other posts about specific settings required for AMD ThreadRipper CPUs, and without those settings, you can get all kinds of freezing and intermittent lock-ups. Definitely search the forum, there is help out there. (EDIT: Looks like you did that "Ryzen Enhancements Enabled")

 

No need to throw away unRAID. I've been using it in a very technical/professional environment since 2014, and there has never been an issue I couldn't solve. Good Luck!

Edited August 28, 2020 by falconexe

[ChatNoir]

Hello, could you provide your diagnostics in your next post so the experts (not me ^^)  can have a looks and maybe provide some advices if they see things to change or try ?

Tools / Diagnostics.

[falconexe]

10 minutes ago, ChatNoir said:

Hello, could you provide your diagnostics in your next post

This 👍. Preferably during a Krusader freeze too.

Edited August 28, 2020 by falconexe