Image push delay

[hyeyoung-leee]

Hi,
We're experiencing image push delays in our Kubernetes-based Harbor deployment, and during the investigation, we found the following error logs.
Harbor Core and Registry communicate directly via Kubernetes services.

Registry log:
net/http: HTTP/1.x transport connection broken: http: ContentLength=20 with Body length 0

Core log:
http: proxy error: context canceled

Could you help us understand:

• What might be causing this behavior in such push delay scenarios?
• Are there any recommended timeout configurations or tuning parameters in Core/Registry to prevent this?
• Are there any known issues or best practices in recent Harbor versions related to this?

Any advice or insights would be greatly appreciated.

Thanks.