Hi brains trust!

 

I've been using Unraid for a couple of years now and have recently upgraded my server to an i7-13700k (new motherboard, ram etc) and at the same time put in a new NVMe PCIe cache drive.  Previous cache drive was a regular sata SSD.  The upgrade process all went smoothly as did the transfer to the new cache.  I won't claim to be an expert, but with the help of guides and these forums I've been able to get everything working the way I wanted, at least up until this point!

 

Since recreating the dockers using the previous templates and starting them up, I've noticed that SABNZBD is not achieving even half the speed it was previously.  I have a 1000mbps connection which SABNZBD would normally saturate (100 MB/s) and as I type this it is hovering around 13 MB/s.

 

I need some help trying to diagnose the speed issue as I am not sure how to proceed.  I was suspecting it was not writing to the cache drive, but that doesn't appear to be the case and I am not sure how to best further diagnose. What else can I check or look into?

 

I have checked:

• Downloads share setup to Prefer Cache.
• binhex-sabnzbd configured to present /mnt/user/Downloads/ to container as /Downloads.
• SABNZBD configured to use /Downloads for temporary and completed downloads.
• Checked File Activity during a download and it appeared to only be on /mnt/cache/Downloads/
• Internet connection speed via the speedtest-tracker docker container (942.78 Mbps).

 

During the transfer from the old cache SSD to the new NVMe drive, I did change the cache configuration of a number of shares and I also noticed after recreating the dockers that a number of them didn't have their network type set.

 

Below is a screenshot of the SABNZBD speed test results. For some reason it lists the internet bandwidth as 12.9MB/S, which reflects the speeds I'm currently getting.

 

 

Below is the config of the docker container.

 

I am running Unraid Version 6.11.5 and have now attached the diagnostic dump.

tower-diagnostics-20221214-1128.zip

 

I appreciate any assistance you can provide!!

Edited December 14, 20223 yr by Loggy
Added diagnostic log

Attach Diagnostics to your NEXT post in this thread 

Now with diagnostics! 

Apologies!

tower-diagnostics-20221214-1128.zip

Does anyone have any thoughts?

 

I can only think that there is something wrong in my configuration resulting in the docker using the array for downloads and not the NVMe disk.

Try mapping to /mnt/cache/downloads, fuse will always have some extra overhead.

With the diagnostics being anonymized it's hard to exactly tell, but I'd get that the downloads are either going to D--a or  D-------s share which both have files on the array, which will significantly degrade your speeds...

 

a-----------p                     shareUseCache="no"      # Share exists on disk7
appdata                           shareUseCache="prefer"  # Share exists on cache
b-----s                           shareUseCache="no"      # Share exists on disk5, disk6
C--------------------------------pshareUseCache="no"      # Share exists on disk6
d--a                              shareUseCache="yes"     # Share exists on disk1, disk2, disk3, disk4, disk5, disk6, disk7
domains                           shareUseCache="prefer"  # Share exists on cache
D-------s                         shareUseCache="prefer"  # Share exists on cache, disk7
g-------e                         shareUseCache="yes"     # Share exists on disk1, disk6
H---------s                       shareUseCache="no"      # Share exists on disk2, disk3
isos                              shareUseCache="prefer"  # Share exists on cache
J----------e                      shareUseCache="no"      # Share exists on disk7
l-----t                           shareUseCache="no"      # Share does not exist
M---a                             shareUseCache="no"      # Share does not exist
m------s                          shareUseCache="no"      # Share exists on disk5, disk7
M-------t                         shareUseCache="no"      # Share exists on disk4
P--------------e                  shareUseCache="no"      # Share exists on disk3, disk5, disk6
P----s                            shareUseCache="no"      # Share exists on disk1, disk2, disk3, disk4, disk5
system                            shareUseCache="prefer"  # Share exists on cache
T-----------e                     shareUseCache="no"      # Share exists on disk3, disk4, disk5, disk6
u----------p                      shareUseCache="no"      # Share exists on cache
v-------p                         shareUseCache="no"      # Share exists on disk7
w---------k                       shareUseCache="no"      # Share exists on disk6, disk7

 

Ok, thanks for both the suggestions.

 

The downloads are all going to the share Downloads.  As you pointed out Squid, I had some data on disk 7.  So I moved all of it onto the cache and restarted SABNZBD.  Unfortunately no speed improvement.

 

I also remapped from /mnt/user/Downloads to /mnt/cache/Downloads.  Also unfortunately with no significant improvement in speed.

 

I'm really scratching my head as this should ensure it is going straight to the new NVMe cache.  Is it possible that Unraid isn't able to properly utilise the read-write speed of the new drive? Something hardware based / driver based?

Did you resolve this issue? I'm having the same problem 

@Loggy did you solve this?

Hi @equ1nt1s - I actually gave up on it at the point above, scratching my head, after following the guidance above to ensure all the data was on the correct disks etc.  However at some point later, it had resolved itself and I was again getting the full speed via SABNzbd. I can't explain why. Possibly the mover function still had some activity to complete that I had overlooked in terms of moving SABNzbd download data off the array and onto the NVMe cache.

Sorry I cant be more specific!