Multiple RetryGatewayFilterFactorys conflict on retry_iteration

[bpeterson3134]

Is your feature request related to a problem? Please describe. I have a need to configure multiple Retry filters in the default-filters for a few reasons.

1. I dont want to retry POST/DELETE operations if the first call times out, but I do for other operations
2. We have downstream systems that sometimes return 500 and on retry they succeed, so I want to retry on 500 response just once but on 503 or timeouts (in some cases) I want to retry multiple times.

The above is not possible in a clean solution as all the retry filters share the attribute retry_iteration so if I have enough retry filters configured, its possible for a retry filter to show that it exceeded max iterations when it did not. This can also make it so that a specific call is not retried as many times as it should be.

And the logs dont make it clear all the time which of the multiple retry filters is running:

04/16/25 09:05:42 d218c-5c79-4df9-b883-944e588d4 TRACE factory.RetryGatewayFilterFactory: Entering retry-filter 04/16/25 09:05:42 d218c-5c79-4df9-b883-944e588d4 TRACE factory.RetryGatewayFilterFactory: Entering retry-filter 04/16/25 09:05:42 d218c-5c79-4df9-b883-944e588d4 TRACE factory.RetryGatewayFilterFactory: Entering retry-filter 04/16/25 09:05:42 d218c-5c79-4df9-b883-944e588d4 TRACE factory.RetryGatewayFilterFactory: Entering retry-filter 04/16/25 09:05:42 d218c-5c79-4df9-b883-944e588d4 TRACE factory.RetryGatewayFilterFactory: Entering retry-filter

Describe the solution you'd like In my use-case, the retry filters cannot share retry_iteration so either a sequence (order of retry filters), a name/id for each configured Retry filter need to be appended to the attribute key used (ex: retry_iteration_1 or retry_iteration_<name/id>) so that each retry filter has its own value. The name/id solution could also help with the logs that get printed to understand which filters are executing because right now it looks like below:

04/16/25 09:31:23 b0064-3b42-45eb-a7d6-80e4e7354 TRACE factory.RetryGatewayFilterFactory: Entering retry-filter 04/16/25 09:31:23 b0064-3b42-45eb-a7d6-80e4e7354 TRACE factory.RetryGatewayFilterFactory: Entering retry-filter 04/16/25 09:31:23 b0064-3b42-45eb-a7d6-80e4e7354 TRACE factory.RetryGatewayFilterFactory: Entering retry-filter 04/16/25 09:31:23 b0064-3b42-45eb-a7d6-80e4e7354 TRACE factory.RetryGatewayFilterFactory: Entering retry-filter

This is what I want my retry filters to look like:

• name: Retry args: retries: 5 series: statuses: BAD_GATEWAY, SERVICE_UNAVAILABLE methods: - GET - PUT - PATCH - name: Retry args: retries: 5 series: statuses: BAD_GATEWAY, SERVICE_UNAVAILABLE methods: - POST - DELETE exceptions: - java.io.IOException # dont retry on timeouts ## Retry only once on 500 error - name: Retry args: retries: 1 series: statuses: INTERNAL_SERVER_ERROR methods: - GET - PUT - PATCH - name: Retry args: retries: 1 series: statuses: INTERNAL_SERVER_ERROR methods: - POST - DELETE exceptions: - java.io.IOException

Describe alternatives you've considered I dont want to configure the retry filter on each route

Additional context None

[bpeterson3134]

I guess this may be a duplicate of https://github.com/spring-cloud/spring-cloud-gateway/issues/2683 . Please review and feel free to close if so.