Gatling shows spikes in response time but New Relic and Grafana don't

Gatling version: 3.11.3 (must be up to date)
Gatling flavor: java kotlin scala javascript typescript
Gatling build tool: maven gradle sbt bundle npm

I read the guidelines and how to ask a question topics.
I provided a SSCCE (or at least, all information to help the community understand my topic)
I copied output I observe, and explain what I think should be.

Hi, I’m running concurrent 2000 Users test with a forever loop and some throttle to get my desire QPS, following closed workload model.
When coming to investigation, I noticed that both Grafana graph and New Relic do not show any request or transactions that exceeded 1 seconds request, even memory and CPU was not overused showed on both monitoring tools, but Gatling shows around 30~60 requests that has response time that is higher than 1 second.
Note that I can only use 1 pod for references, I had double checked to ensure all pods statistics are the same and system was stable during the test time



What could possibly happen during the test time that causing this difference ?

Your APM is blind on lots of things, including:

  • network latency between your clients/load generators and your edge server
  • everything that happens in your edge server (here, your ingress), including DNS, TCP and TLS.

Note: Gatling Enterprise would give you additional metric to investigate this:

  • number and duration of DNS resolutions,
  • number and duration of TCP connect,
  • state of TCP connections,
  • number and duration of TLS handshakes.

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.