You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Is your feature request related to a problem? Please describe.
When using two node pools in an Agones cluster, one for Agones and one for GameServers, allocation requests sometimes fail during the process of reducing nodes in the GameServers’ node pool, especially when reducing many nodes. I think it shouldn't be affected by changes in the GameServers’ node pool.
Hi @osterante , thanks for the contribution. Just a few questions, can you explain more on the problem? what metrics did you observed and why you think it's related to the health check? Did you test the fix in #4022 and verify it mitigates the issue?
Is your feature request related to a problem? Please describe.
When using two node pools in an Agones cluster, one for Agones and one for GameServers, allocation requests sometimes fail during the process of reducing nodes in the GameServers’ node pool, especially when reducing many nodes. I think it shouldn't be affected by changes in the GameServers’ node pool.
Describe the solution you'd like
Set the externalTrafficPolicy for the agones-allocator service to Local from Cluster(default).
https://cloud.google.com/kubernetes-engine/docs/concepts/service-load-balancer?hl=ja#health_check
Describe alternatives you've considered
A clear and concise description of any alternative solutions or features you've considered.
Additional context
Add any other context or screenshots about the feature request here.
The text was updated successfully, but these errors were encountered: