Why does the API Proxy not retry Service Callouts that fail on the transport layer?

Not applicable

I am using a Service Callout with "Connection: keep-alive" and occasionally get an unexpected End of File (one in thousands of requests). Most HTTP Clients tend to retry requests automatically when the failure is at the transport level. Is there a way to get this behavior? Am I stuck retrying on my own at a higher level?

0 2 466
2 REPLIES 2

@Royal Chan ,

How about attaching the same policy one more time with a policy condition that verifies earlier service callout failed or not. Does that work ?

It certainly works, but it's not a behavior that I expected. Rather, it seems like a small gap in implementation: https://tools.ietf.org/html/rfc7230#section-6.3.1

Most HTTP Clients and Intermediate services like proxies will "almost" handle these sorts of failures for you by retrying once. Given the frequency of the issue it seems like I'm not even getting that by default, and am curious why.