Enabling ekco-enable-internal-load-balancer on all nodes uses all the CPU


Setup/steps to reproduce:

os version: ubuntu22.04
haproxy version: haproxy:2.9.1-alpine3.19
k8s version: v1.25.12

I have an embedded cluster of 5 nodes (3 primary, control-plane; 2 secondary, without control plane).

I have installed the 1st node with the following options:
curl -sSL https://kurl.sh/<channel_name> | sudo bash -s ha ekco-enable-internal-load-balancer

While joining the rest of the nodes with the commands generated by the installer on the 1st node, I have appended the ha and ekco-enable-internal-load-balancer options to each join command as well.


Every node has a running instance of haproxy pod as expected, however, on every node the CPU usage of haproxy pods rise soon after start to “as high as possible” pushing the node to use 100% CPU.

The haproxy pods’ logs themselves don’t show any particular error or information that would help resolving the issue.

Any help would be appreciated! :slight_smile:


The issue was a bug in the v0.28.2 ekco addon version. v0.28.4 fixes the issue.

Thank you @Balint_Csonka for the topic and update on where this is now fixed :raised_hands: