Hey everyone!
I'm facing some challenges with the API Gateway provided by the SoftwareAG team. I'm seeing unusually high memory consumption on my Elasticsearch pods, even though there is hardly any traffic coming through. Here's a snapshot of the memory usage:
- POD: apigateway-es-0, CPU: 11m, MEMORY: 30223Mi
- POD: apigateway-es-1, CPU: 14m, MEMORY: 30189Mi
- POD: apigateway-es-2, CPU: 7m, MEMORY: 30167Mi
- POD: apigateway-prd-0, CPU: 26m, MEMORY: 8089Mi
I've removed the memory limit, but upon restarting the pods, the memory usage skyrockets to over 30G. Can anyone help me figure out why there's this massive memory consumption and where it's coming from?
Thanks for any insights!
2 Answers
You should try setting the Xmx and Xms parameters in your Elasticsearch configuration to manage memory requests and limits more effectively. This can help prevent those memory spikes you’re seeing!
Have you checked your Elasticsearch configurations? You might want to limit memory there as well because just capping it in Kubernetes can lead to the pod being killed if it goes over the limit.
I did set a memory limit to 24GB, but the pod tends to automatically max out at around half of that memory when it’s ready. Even the support team seems puzzled about where the memory drain is coming from.
I’ve noticed those are usually set to 2GB for requests and a max of 4GB. But in my case, when I tried setting it to 20GB, it still consumed about 10GB automatically. It’s running as a stateful set, by the way.