/usr/local/bin/shfs using a lot of CPU

[905jay]

Hello everyone, another post looking for help on another issue. 

It may very well be a bad configuration on my part, and if so I will gladly take some advice on where I may be going wrong. 

 

I noticed about a week ago (roughly) that my load average on the server was quite high, and the CPU was constantly in red on the dashboard. 

htop shows many processes with the command below:

/usr/local/bin/shfs /mnt/user -disks 31 -o default_permissions,allow_other,noatime -o remember=330

 

I have read that this is normal to have these processes as it's what unraid uses to facilitate shares, however I think the amount of CPU they are using is a little unreasonable.

 

I can't seem to pin down any particular actions I may have taken that would result in the high CPU I'm experiencing right now, but I have been playing around quite a but replacing NPM with traefik, crowdsec and authelia. 

 

Can someone help me understand why there are so many of these /usr/local/bin/shfs commands running, eating so much CPU cycles?

Did I mis-configure something along the way perhaps? 

 

Server Details:

Ryzen 7 3700x

Asus Prime x570 Pro

80GB DDR4 Memory

LSI HBA (IT Mode) with 6 Seagate Ironwolf 8TB ZFS Disks (2 parity, 4 array)

1 Samsung 980 SSD Cache

Nvidia 1070ti GPU for transcoding 

 

unraid-diagnostics-20230828-2303.zip

[JorgeB]

You can try changing everything you can to user disk shares instead, or the new exclusive shares.

[905jay]

4 hours ago, JorgeB said:

You can try changing everything you can to user disk shares instead, or the new exclusive shares.

 

Thanks @JorgeB for the info. Would this be something that I may have changed somehow, and your're recommending that I change back to `user disk shares` or exclusive shares?

 

I'm not familiar with either but let me google and see what I find. 

Is there a benefit /drawback to doing either approach?

[JorgeB]

Benefit is performance since it bypasses FUSE, drawback is that it can only be used with pools or a single array disk for each app.

[905jay]

@JorgeB does this correlate to my other post about the painfully slow transfers in any way?
 

Slow ZFS Array Transfers

[JorgeB]

This is a separate issue.