While implementing anti-affinity rules to improve infrastructure reliability (preventing connection drops and reducing potential downtimes), our team redistributed ingress pods across different nodes. During this process, a misconfiguration occurred in Helm charts: we did not explicitly specify the ingress controller version.
As a result, Helm automatically upgraded the ingress controller to the latest available version, which introduced breaking changes incompatible with our current configuration. This led to the ingress controller being unable to accept the configuration and caused the plugin to become temporarily unavailable.
We acknowledge the impact this may have caused and sincerely apologize for the inconvenience. Transparency is important to us, and we are taking steps to ensure such misconfigurations do not happen again.