Latency spikes observed in frontend

[dilinade]

I've deployed the microservices-demo on GKE, and while querying a single product (/product/{ID}) using Locust, I'm encountering intermittent latency spikes. The request rate is approximately 10 RPS. The spikes seem to originate from three different places

• CurrencyService/GetSupportedCurrencies
• hipstershop.CurrencyService/Convert
• hipstershop.CartService/GetCart

I enabled tracing and obtained the following traces:

Normal Trace : RTT 20ms

Spike 1 : RTT 616ms

Spike 2 : RTT 333ms

Has anyone else encountered similar latency spikes? Are there any insights into why these spikes might occur, or is there any aspect that might have been overlooked in the deployment or configuration?

[bourgeoisor]

Hi @dilinade! I'm not able to reproduce this on my end. Are you still running into this?

[dilinade]

I believe the issue only arises when I set a CPU resource limit. I'm certain the usage doesn't exceed this limit; it's actually far below it, yet the problem still occurs.