澄潭
澄潭
Maybe we can use resource name as the cache key?
Or create a reference relationship between the resource name and the cache. Because we can only find out which modules are in use from the response of LDS.
> @johnlanni Did you encounter this situation in real cluster? In normal setting, Istiod will disconnect ADS connection for each 30 min, and reconnect. So, ECDS will be delivered again...
> > @zirain How about purging unused wasm binaries when a new ECDS comes, instead of triggering it by timer. Then, if there is no ECDS, no purging will happen....
@zirain @ingwonsong Thanks, I see.
BTW, the current implementation seems to push all resources in any ECDS response. https://github.com/istio/istio/blob/32afc27ea91ef68e4e5f0ee3dd47d0aedfba35f1/pilot/pkg/networking/core/v1alpha3/extension_config_builder.go#L15-L35
内部重定向是支持的:https://www.envoyproxy.io/docs/envoy/latest/api-v3/config/route/v3/route_components.proto#envoy-v3-api-msg-config-route-v3-internalredirectpolicy
这个已经有部分实现了，后面代码放上来，你可以在基础上再优化下哈
计划后续基于插件中访问redis的能力来扩展实现这一能力
插件sdk已经支持redis api，欢迎认领