Roadmap for features like "Retries and Timeouts"

[Shuanglu]

There are some features like "Retries and Timeouts" for traffic management in other mesh like istio. I didn't find the corresponding features/API in nginx-service-mesh.
May I know if we have such features on the roadmap or any alternative approach to achieve in nginx-service-mesh?

[jbyers19]

@Shuanglu, would Circuit Breakers work for your needs?

[Shuanglu]

@Shuanglu, would Circuit Breakers work for your needs?

it looks like they are for different purpose https://istio.io/latest/docs/concepts/traffic-management/#retries ? pls correct me if I'm wrong. The scenario is when there is temporary networking failure and we expect the sidecar proxy can retry for several times behalf of the client app.

[brianehlert]

It appears that retries is for errors in the path between encrypting sidecars.
I would not expect this to be a problem within a K8s cluster, but rather when the mesh is extended to services outside or between clusters.
@Shuanglu can you help us understand your need / use case for this setting?

[Shuanglu]

It appears that retries is for errors in the path between encrypting sidecars. I would not expect this to be a problem within a K8s cluster, but rather when the mesh is extended to services outside or between clusters. @Shuanglu can you help us understand your need / use case for this setting?

The typical scenario is the trainsient networking failure mentioned in https://istio.io/latest/docs/concepts/traffic-management/#retries especially when there is no retry logic in application level

[f5-todd]

Retries and Timeouts have been added to our backlog of work. I will update this ticket when I have more details on when these features will be released.