Distributed test issues

I am running gatling tests from 8 aws nodes in parallel for 25K (3125 users * 8 nodes)concurrent users in total. The test fails with the below errors.
But the same test( ie same load) runs fine when the load is generated for 2 or 3 gatling nodes(12500 users *2 nodes)
Any clues what could be the possible limitations here. Is it due to network or load balancer or any gatling settings .
My target load is 100K concurrent users, so I definitely require 8 nodes to run tests without exhausting the ports and resources

* status.find.is(200), but actually found 502
* i.n.c.ConnectTimeoutException: connection timed out

Distributed load testing is only officially supported in Gatling Enterprise.
I’m afraid we from Gatling can only help customers on this.

I have 8 aws nodes on which I have installed gatling and start the tests in parallel from all nodes. So it can be considered as seperate tests in different nodes but same application url. My application is run as docker containers in the same aws cluster. so it cannot be accessed outside the cluster.
What could be the possible reason why for the same load its running fine from 2 gatling nodes but fail with 8 gatling nodes?

You can try enable http2 so Gatling tool simulate behavior like okhttp client for communication between 2 services


I also had encountered similar error and error has been resolved using http2