Spring cloud gateway - Connection prematurely closed DURING response

I am using Spring cloud gateway with eureka regisry and using open feign client to communicate between 2 services .
Cloud gateway is working fine with normal API's. It is throwing below error while running a Spring batch application:

reactor.netty.http.client.PrematureCloseException: Connection prematurely closed DURING response
Suppressed: reactor.core.publisher.FluxOnAssembly$OnAssemblyException:
Error has been observed at the following site(s):
*__checkpoint ⇢ org.springframework.cloud.gateway.filter.WeightCalculatorWebFilter [DefaultWebFilterChain]
*__checkpoint ⇢ HTTP POST "" [ExceptionHandlingWebHandler]
Original Stack Trace:

The error comes whenever it reaches a count 100000 records of ingestion (around 30 secs). Ingestion is happening through chunk and i have total 7.9 million records to be ingested.

Setting the configuration as below:

spring:
application:
name: *****
cloud:
openfeign:
httpclient:
connection-timeout: 10000000
client:
config:
default:
connectTimeout: 10800000
readTimeout: 50000000

When debug, i am getting the error because one of microservices called through Openfigen client is closing the connection as below:

DEBUG [http-nio-8051-exec-9] c.s.i.g.client.feign.ESFeignClient - [] connection: close

the routing service is having below response entity with response 200
INFO [http-nio-8051-exec-7] - response <200 OK OK,com.scb.icore.gbingestion.dto.ApiResponse@98f78b1,[connection:"close", content-type:"application/json;charset=UTF-8", date:"Wed, 26 Jun 2024 05:05:33 GMT", transfer-encoding:"chunked", vary:"Access-Control-Request-Headers", "Access-Control-Request-Method", "Origin"]>

Please help on this.

Start by reproducing the Spring Cloud Gateway request through Eureka and OpenFeign with the Spring Batch workload, using the reported timeouts and chunk size that reaches about 100,000 records. Inspect the Reactor Netty PrematureCloseException and the routing service's HTTP 200 response with connection close; done means identifying the component that closes the connection and verifying the cause against the reported workload.