Kernel-throttling stopped the websdr....

[Bas ON5HB]

The websdr was crashed again, but the kernel started giving errors like these in dmesg:

perf: interrupt took too long (6398 > 6345), lowering kernel.perf_event_max_sample_rate to 31250

Right after these messages it crashed.

It seems the Kernel of Linux doesn't like when a program hoggs the CPU too long so it removes attention for a bit, throttle down the number of CPU-slices it gets.

This is not good for realtime applications like hours.

I hope this solves it:

Code Select Expand

sysctl -w kernel.perf_cpu_time_max_percent=0
Just turn it off

To make it permanent, add this to /etc/sysctl.conf

Code Select Expand

kernel.perf_cpu_time_max_percent=0


sysctl -p to let it work instantly, no reboot needed but because you added it, it will stay after each reboot.

[Bas ON5HB]

And that didn't work.

But I left it in there, but upped this one again: fft_size=8388608

Maybe it's FFTW that does this.

As this time right before it crashed again the NIC decided to go away and come back right after.
Making the websdr crash.

After reading more logs....this happened...

Jun  1 13:38:17 Websdr avahi-daemon[543]: Registering new address record for fda7:ffbb:e137:0:48d9:ce50:449c:89db on eno1.*.
Jun  1 13:38:18 Websdr avahi-daemon[543]: Registering new address record for 2a02:578:440e:0:dbab:16f0:d925:fe7d on eno1.*.
Jun  1 13:38:19 Websdr avahi-daemon[543]: Withdrawing address record for fda7:ffbb:e137:0:8dac:49ca:df13:6af0 on eno1.
Jun  1 13:38:20 Websdr avahi-daemon[543]: Withdrawing address record for 2a02:578:440e:0:765:5b94:ed3:b114 on eno1.

It played with the NIC! Need to find out why it does this.

Here we go....time to stop this stupid service that I do not need at all.

Code Select Expand

root@Websdr:/var/log# systemctl status avahi*
● avahi-daemon.service - Avahi mDNS/DNS-SD Stack
     Loaded: loaded (/lib/systemd/system/avahi-daemon.service; enabled; vendor preset: enabled)
     Active: active (running) since Sun 2025-05-25 13:58:43 CEST; 1 week 1 day ago
TriggeredBy: ● avahi-daemon.socket
   Main PID: 543 (avahi-daemon)
     Status: "avahi-daemon 0.8 starting up."
      Tasks: 2 (limit: 18787)
     Memory: 1.5M
        CPU: 32.108s
     CGroup: /system.slice/avahi-daemon.service
             ├─543 "avahi-daemon: running [Websdr.local]"
             └─629 "avahi-daemon: chroot helper"

mei 31 13:38:28 Websdr avahi-daemon[543]: Registering new address record for 2a02:578:440e:0:765:5b94:ed3:b114 on eno1.*.
mei 31 13:38:29 Websdr avahi-daemon[543]: Withdrawing address record for 2a02:578:440e:0:42b0:f0b2:2878:745a on eno1.
jun 01 13:38:17 Websdr avahi-daemon[543]: Registering new address record for fda7:ffbb:e137:0:48d9:ce50:449c:89db on eno1.*.
jun 01 13:38:18 Websdr avahi-daemon[543]: Registering new address record for 2a02:578:440e:0:dbab:16f0:d925:fe7d on eno1.*.
jun 01 13:38:19 Websdr avahi-daemon[543]: Withdrawing address record for fda7:ffbb:e137:0:8dac:49ca:df13:6af0 on eno1.
jun 01 13:38:20 Websdr avahi-daemon[543]: Withdrawing address record for 2a02:578:440e:0:765:5b94:ed3:b114 on eno1.
jun 02 13:38:10 Websdr avahi-daemon[543]: Withdrawing address record for fda7:ffbb:e137:0:48d9:ce50:449c:89db on eno1.
jun 02 13:38:10 Websdr avahi-daemon[543]: Registering new address record for 2a02:578:440e:0:a6ec:9e52:671d:726f on eno1.*.
jun 02 13:38:10 Websdr avahi-daemon[543]: Registering new address record for fda7:ffbb:e137:0:14d:32b2:9887:4604 on eno1.*.
jun 02 13:38:11 Websdr avahi-daemon[543]: Withdrawing address record for 2a02:578:440e:0:dbab:16f0:d925:fe7d on eno1.
Quick solution:

systemctl stop avahi-daemon
systemctl disable avahi-daemon

Then it should not play with my NIC anymore.

[Bas ON5HB]

Still running, no more NIC reloads!

Avahi-daemon is definitly the problem of many crashed if you have a Networkprinter with Bonjour protocol in your network.