Hi,

 

I'm struggling with high iowait during times when I'm downloading with NZBget and then extracted files are moved with Radarr/Sonarr. When viewing Netdata the interrupts are somehow coming from eth0:

IPv4 outbound is practically at full capacity (its jumping up and down).

 

Now, I dont know what is supposedly going through eth0. I'm at a 500mbit connection, and this data is not going to my switch so it must be internal in the Unraid server somehow. Is data between Dockers moved through network?

See how sent and recieved/forwarded are following eachother?

 

My network settings in Unraid:

 

 

Any idea what is causing these interrupts/ipv4 outbound traffic and if I can do anything with it? Id really like to reduce the iowait 

 

Thanks!

Following up with some more info:

The problematic interrupts I'm seeing are all going to CPU0, which is basically drowning this core. I'm attaching my interrupts.

 

When I'm searching for this problem it seems to be related to Linux kernel version, and I know the version used by 6.3.5 which I'm running is not the best for my Proliant. I could try downgrading to version 6.2.4 which I know to work. How do I downgrade? Get another flash and set it up with 6.2.4 and then just switch to test it out during the trial?

Interrupts.xlsx

1 hour ago, Fredrick said:

How do I downgrade?

If you upgraded from 6.2.4 then those files should be in /boot/previous on your flash drive. All you would have to do is overwrite the 6.3.5 files with them. I'd make a backup of everything first though.

I didnt. 6.3.5 is my first install of Unraid.

 

I dont understand why data is being sent to the NICs, and I'm thinking this could be the cause of interrupts.

I get that the ipv4 reporting in netdata can be confusing here as it somehow also includes statistics coming from the docker-container-nics. 

 

Is this normal? As both containers (ie radarr/sonarr and nzbget) has access to /mnt/user/Downloads I don't understand why any data is going outbound ipv4.