Help: Slow Docker performance and Slow Linux VM Performance

[Cap-_-merica]

I put this in general because i seem to have 2 issues but i believe they are all related to linux.

 

Basic Idea:

Docker pull and install are slow, performance is bad.

Linux VM took 40min to install 110 mb update. 

 

Long Version:

I had noticed that the docker containers performed slowly. The navigation on Nextcloud being slow was the first thing that showed up. I the issue cam back up when i installed the binhex Minecraft docker container. downloading and extracting seems to be slow as the docker container was pulled down. After installation the server Minecraft docker crashed when i had 3 player. I installed binhex's MineOS and similar problems took longer than usual to pull the container and extract. 

I deleted the docker image and reinstalled docker containers. Things still seems to be slow while pulling and extracting. Minecraft dockers still crashed. 

Next I installed the server on a windows 10 vm and that seemed to work better and not crash. 

I then installed the MineOS Turnkey as a vm which is based on Debian. Still had bad performance issues and crashed.

I then installed Ubuntu server 20.04 (8bg mem and 4 core 8 threads) as a vm, sent the sudo apt update and upgrade comands and it took over 40 min to download 110 Mb and install them, which in my opinion is slow.

 

Unraid Server:

Ryzen 2700 with Asus x370 prime pro and 32 GB ECC ddr4.

8 disk array with 1 tb HP ex920 nvme SSD as cache

All vms and docker are assigned to cache or set as prefer cache.

Dockers that are always running: Plex, Nextcloud, mariadb, OpenVPN, qbittorent

VM's: 2 Windows 10 (1 gpu passthrough, one on VNC), 1 Unbuntu server 20.04, Mac High Sierra.... all Vms rarely run, and one at a time.

 

Diag zip attached.

 

Thank You

tower-diagnostics-20201229-2055.zip

[Cap-_-merica]

So after some more testing I found out that my nvme drive was acting up. I would sometime benchmark fine at other times it has atrocious speeds, down to 40mb/s. I had backed up the cache drive and reformatted it from btrfs-encryted to xfs-encrypted. Still had the problems after I moved everything over. I decided to replace that drive with a sata SSD and since the performance of both dockers and vm's is back to normal. Smart data doesn't show any problems for the nvme drive. The weird part is that I tested the drive once again in the unassigned devices and it performed beautifully. 

 

Question is the drive bad (as in hardware?) or was it something to do with its place as cache drive and unraid?

How do you verify that the hardware nvme is bad besides smart data and simple benchmark?

 

Drive Benchmark is attached.

 

Smart Data:

SMART Data as of   Jan 3, 2021 2:08 PM

Temperature31 Celsius

Warning Comp. Temperature Time0

Critical Comp. Temperature Time0

Critical Warning0x00

Available Spare100%

Available Spare Threshold10%

Percentage Used0%

Data Units Read41,354,551 [21.1 TB]

Data Units Written397,036,618 [203 TB]

Host Read Commands150,229,416

Host Write Commands2,006,570,124

Controller Busy Time221,673

Power Cycles52

Power On Hours11,151

Unsafe Shutdowns9

Media and Data Integrity Errors0

Error Information Log Entries0

[LittleLama]

Hi @Cap-_-merica, it looks like I'm experiencing the same problem. Did you find anything else ?

[Cap-_-merica]

Hello @LittleLama... The problem was a bad nvme drive...while the drive worked somewhat it had slowed down to a crawl...I RMA'd the drive and replaced it and afterwards everything worked fine.

[LittleLama]

1 hour ago, Cap-_-merica said:

Hello @LittleLama... The problem was a bad nvme drive...while the drive worked somewhat it had slowed down to a crawl...I RMA'd the drive and replaced it and afterwards everything worked fine.

Thank you very much for your report.

[LittleLama]

The problem was the same for me. Juste replaced my two Samsung QVO (btrfs) by a single Samsung EVO (xfs) and everything returned fine.

Don't know if it is a hardware issue or a file system issue.

[Cap-_-merica]

9 hours ago, LittleLama said:

The problem was the same for me. Juste replaced my two Samsung QVO (btrfs) by a single Samsung EVO (xfs) and everything returned fine.

Don't know if it is a hardware issue or a file system issue.

I too thought that it could be a btrfs problem. I formated the drive in windows and still had a problem with it. The next SSD installed I used xfs and had no problems since.

Edited February 6 by Cap-_-merica