I can't think of a way to easily monitor the times that the request takes to go between network components in this situation.
The browser's developer tools->network tab would monitor the overall length of time for the request including the time the IHS reports... so you could substract that but it wouldn't tell you which element was causing the problem
I am interested to know if anyone suggest a way.
I'm guessing that Wireshark wouldn't help unless it was placed in multiple places in the network path
------------------------------
Mark Robbins
Support Lead/Technical Design Authority / IBM Champion 2017 & 2018 & 2019 & 2020 & 2021
Vetasi Limited
https://www.linkedin.com/pulse/maximo-support-advice-from-non-ibm-engineer-article-mark-robbins/------------------------------
Original Message:
Sent: Wed November 15, 2023 07:19 AM
From: Eduardo Izquierdo Lázaro
Subject: Clarification about the response time measured by liberty in the access log for a HTTP POST request
It is synchronous call, looks like the problem is related to some network mediation devices (Load Balancer, Gateways) implemented in the Cloud.
------------------------------
Eduardo Izquierdo Lázaro
Automation Architect
DECIDE
Madrid
609893677
Original Message:
Sent: Tue November 14, 2023 12:24 PM
From: Kevin Grigorenko
Subject: Clarification about the response time measured by liberty in the access log for a HTTP POST request
%D is the total elapsed time including network response time, so what you're observing is surprising. What framework does the client use to make the call? Does it make the call using a synchronous or asynchronous call?
------------------------------
Kevin Grigorenko
Application Runtimes SWAT
IBM
Original Message:
Sent: Tue November 14, 2023 07:37 AM
From: Eduardo Izquierdo Lázaro
Subject: Clarification about the response time measured by liberty in the access log for a HTTP POST request
Good morning,
A client application is calling a service running on liberty server and is getting a timeout in some of the request. Timeout is set to 2sg. Note that client is running in AWS Cloud, liberty running on-premises and the request is a HTTP POST with a JSON payload, so network latency maybe relevant.
I've activated the access logs in the liberty profile including a %D in the log format to get the response times. Even in the worse case, when the JSON payload is very heavy, the response time logged by liberty profile is never higher than 800 msg.
¿Does the response time reported by liberty include the time of serializing the payload? In other words ¿is counting the time since the client opens the input connection to write the payload?
If this is not the reason,¿How can be possible that the time reported by liberty is so different to the time perceived by the client when the timeout expires?
------------------------------
Eduardo Izquierdo Lázaro
Automation Architect
DECIDE
Madrid
609893677
------------------------------