Closed ysksuzuki closed 1 year ago
The fix I proposed[1] has been applied to -net tree[2] and should also get backported to stable kernels (stable was CCed on it).
[1] https://patchwork.kernel.org/project/netdevbpf/patch/20220425103703.3067292-1-razor@blackwall.org/ [2] https://git.kernel.org/pub/scm/linux/kernel/git/netdev/net.git/commit/?id=acb16b395c3f3d7502443e0c799c2b42df645642
Thanks @NikAleksandrov, awesome work!
This fix landed in linux 5.15.38 which is in flatcar stable 3139.2.2.
Thank you again @NikAleksandrov for fixing this issue.
Description
All worker nodes got stuck in rebooting due to kernel panics when we run cilium v1.11.3 with loadBalancer.acceleration native setting on flatcar stable 3139.2.0.
Impact
We can't use Cilium LoadBalancer & NodePort XDP Acceleration feature. Enabling this feature is greatly beneficial for us since the XDP-based load balancer performs significantly well.
Environment and steps to reproduce
Expected behavior
Cilium with loadBalancer.acceleration set to native start up fine.