Product/components used and version/fix level are you on:

API Gateway

Detailed explanation of the problem:

Is there a way to configure server timeouts in milliseconds? The UI offers the option only in seconds.

https://higherlogicdownload.s3.amazonaws.com/IMWUC/webMethods/Files/b363a03e64f793678cf3dfde299905d0f38c08b6.png 1.5x, https://higherlogicdownload.s3.amazonaws.com/IMWUC/webMethods/Files/b363a03e64f793678cf3dfde299905d0f38c08b6.png 2x" data-dominant-color="FAFBFB">

image949×713 30.7 KB

Error messages / full error message screenshot / log fileL

Is your question related to the free trial, or to a production (customer) instance?

I am trying out the free trial version right now.

Have you installed all the latest fixes for the products and systems you are using?

Yes, I downloaded the 10.15 free trial docker image for our evaluation.

Hello @devashi.tandon ,

At the moment it is only in seconds , any usecase which you feel is is need of milliseconds?

Regards,
Bharath

Hi @Bharath_Meka1
Timeout in seconds for server is very high, especially when the servers are residing in the same internal network. When API-GW is running in the same network as the backend server or authentication/log server, we would have preferred a granularity in milliseconds.

Hello @devashi.tandon ,

Read Timeout is configured in seconds(max) after which a socket read attempt will get timed out. So it is better to be in seconds as part of the configurations ?
Any performance challenges are you incurring into if yes please mention?

Regards,
Bharath

Hi @Bharath_Meka1

From my observations, I see that API-GW opens a new socket connection for every log that it sends to the external endpoint server. After receiving a response from the endpoint server, it sends a FIN and closes the connection. When a new socket connection is created every time, a read timeout in seconds may be ok to have. However, since a new socket connection is created for every log sent to endpoint server it would have its own overhead, and can affect performance at high rate of traffic. If the same socket connections are reused for sending multiple logs, a read timeout in milliseconds would make more sense.

I have a follow up question:

1. Is there a configuration to change this behavior of API-GW? Meaning, can we configure the gateway to reuse socket connections instead of creating new ones every time?