From da09dab91bdab490e49b28e1c964cf64eb9ea755 Mon Sep 17 00:00:00 2001 From: Jeff Ortel Date: Fri, 17 Nov 2023 07:58:42 -0600 Subject: [PATCH] :bug: Addon memory request and limit match. (#281) Deploying the pod with request = limit should be a good (incremental) step towards preventing OOM kills. My memory profiling suggests that 4Gi is a good number for our TestApp. Although, I can run analysis on the TestApp using 2Gi/2Gi just fine (though memory peaks at 1.9Gi) on my minikube but Pranav reports that he cannot. We should still follow up with passing -Xmx=80% of limit to the Java provider. IMHO, 4Gi is much higher than I would have expected for the analyzer. This means that when running on most personal, single node clusters (like minikube), we will only be able to analyze a few applications in parallel. Signed-off-by: Jeff Ortel --- roles/tackle/defaults/main.yml | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/roles/tackle/defaults/main.yml b/roles/tackle/defaults/main.yml index 5b45204c..385465da 100644 --- a/roles/tackle/defaults/main.yml +++ b/roles/tackle/defaults/main.yml @@ -163,7 +163,7 @@ analyzer_service_name: "{{ app_name }}-{{ analyzer_name }}-{{ analyzer_component analyzer_container_limits_cpu: "1" analyzer_container_limits_memory: "4Gi" analyzer_container_requests_cpu: "1" -analyzer_container_requests_memory: "2Gi" +analyzer_container_requests_memory: "4Gi" cache_name: "cache" cache_data_volume_size: "100Gi"