Kern.log is filling up with i2c errors

**rmelkert** · April 27th, 2021

I've been installing a new Ubuntu Server 20.04 system and all is going fine, except there is a (apparently non fatal) i2c error flooding the kern.log

Code:

Apr 27 05:17:37 srv01 kernel: [31398.980494] i2c i2c-3: sendbytes: NAK bailout.
Apr 27 05:17:57 srv01 kernel: [31419.460807] i2c i2c-3: sendbytes: NAK bailout.
Apr 27 05:20:11 srv01 kernel: [31552.579808] i2c i2c-3: sendbytes: NAK bailout.
Apr 27 05:20:31 srv01 kernel: [31573.059753] i2c i2c-3: sendbytes: NAK bailout.
Apr 27 05:21:33 srv01 kernel: [31634.499618] i2c i2c-3: sendbytes: NAK bailout.
Apr 27 05:22:03 srv01 kernel: [31665.219176] i2c i2c-3: sendbytes: NAK bailout.
Apr 27 05:22:34 srv01 kernel: [31695.939112] i2c i2c-3: sendbytes: NAK bailout.
Apr 27 05:23:05 srv01 kernel: [31726.659046] i2c i2c-3: sendbytes: NAK bailout.
Apr 27 05:23:35 srv01 kernel: [31757.378968] i2c i2c-3: sendbytes: NAK bailout.
Apr 27 05:24:57 srv01 kernel: [31839.298770] i2c i2c-3: sendbytes: NAK bailout.

i2cdetect says it's a radeon thing

Code:

i2c-3   i2c             Radeon i2c bit bus 0x90                 I2C adapter

I believe its a sensor, might be the fan? Thing is this is a fanless card (only really needed during installation as the system is headless most of the time).

It doesn't seem to do any harm but I hate it filling the log this way, so:

Can I disable this sensor somehow?

Any help/pointers would be appreciated.

**slickymaster** · April 27th, 2021

Thread moved to Server Platforms for a better fit

**ameinild** · April 27th, 2021

I gave an answer to this on AskUbuntu.com

https://askubuntu.com/questions/1324...o-much/1324384

Just replace the "contain" string with something else, for instance:

Code:

if $msg contains "i2c-3: sendbytes" then { stop }

**rmelkert** · April 29th, 2021

Originally Posted by ameinild

I gave an answer to this on AskUbuntu.com

Thanks.

That is certainty a last resort option, but I was hoping on a more elegant solution.

Also I read the i2c-3 label can change between boots.

Anyone knows of a (kernel parameter) setting to exclude the 0x90 radeon sensor.

Or a way to determine what is pulling that sensor and make it stop doing that.

I'm not running lm-sensors but I did notice the /etc/sensors3.conf. It, however, does not contain an obvious reference to the problem sensor.

**rmelkert** · May 19th, 2021

Just an update, for people having similar issues.

I still don't know what the core cause of this issue is, but it went away when I disconnected the monitor.

So it seems it has nothing to do with sensors at all, maybe it doesn't like the monitor switch or the card has some problems with the old 'vga' port.

Anyhow, as this is a headless system anyway it solves the issue for me.

**coolspot18** · January 7th, 2022

Originally Posted by rmelkert

Just an update, for people having similar issues.

I still don't know what the core cause of this issue is, but it went away when I disconnected the monitor.

So it seems it has nothing to do with sensors at all, maybe it doesn't like the monitor switch or the card has some problems with the old 'vga' port.

Anyhow, as this is a headless system anyway it solves the issue for me.

I built a new Ryzen server but used an old Radeon HD 2400 card for video. The server worked well with a VGA monitor but tonight I racked the server and connected it to my old VGA KVM switcher. It was after I hooked up my KVM that I noticed the NAK error in dmesg ... so I think you're right, it's something to do with the VGA port and the connected device.