Hey all,
I'm setting up a testing environment to run the latest accel-ppp and debian code.
Installed my latest debian 11, kernel 5.10.0-10-amd64. When porting the old config I noticed RPS not working anymore:
/bin/echo fffe > /sys/class/net/eth5/queues/rx-0/rps_cpus
/bin/echo: write error: Value too large for defined data type
Apparently it's the kernel who is doing this.
Has anyone encountered this ?
accel-ppp + debian11 + rps
Re: accel-ppp + debian11 + rps
Hi @proca, do you have 16 CPU cores? Why do you define exactly fffe?
Re: accel-ppp + debian11 + rps
I have 12 cores and I want to not use the first core in the packet processing. Seems like in kernel 5 this cpu mask has changed ?
went back to 4.19 to get it working how i want
went back to 4.19 to get it working how i want
Re: accel-ppp + debian11 + rps
hi,
its working for me on fedora kernel 5.8
cat /sys/class/net/enp3s0f0/queues/rx-0/rps_cpus
ffffff
its working for me on fedora kernel 5.8
cat /sys/class/net/enp3s0f0/queues/rx-0/rps_cpus
ffffff