This message was deleted.
# k3s
a
This message was deleted.
c
Were you expecting that they would be sent round-robin? I don’t believe that’s promised anywhere, is it?
We use kube-proxy to handle the actual routing between the backend service and the pods, I believe it is just random.
m
So problem I am facing is, When request comes specifically(from syslog-ng/tcp) then all request going to 1 executor.
that overloads 1 node
I am using LB type service
c
Servicelb just uses iptables rules to forward traffic to the in-cluster service endpoint, which is handled by kube-proxy. It does not have any specific behavior around traffic distribution across pods.
🙌 1
m
Make sense..