Recently, I observed a significant spike in RAM usage for the dataset-operator, increasing from under 200 MB to approximately 14.2 GB within roughly 30 minutes. This behavior suggests a potential memory leak in the application and should likely be investigated separately.
While looking into this, I also noticed that resource requests and limits are not defined in the Helm chart. Adding resources.requests and resources.limits would help Kubernetes better manage the operator’s resource consumption and allow it to be restarted automatically without negatively impacting other workloads in the cluster.
Although the memory spike itself may be an application-level issue, I believe defining resource constraints in the Helm chart is a valuable improvement regardless. I’d like to open a PR to add these configurations to the chart.
Recently, I observed a significant spike in RAM usage for the dataset-operator, increasing from under 200 MB to approximately 14.2 GB within roughly 30 minutes. This behavior suggests a potential memory leak in the application and should likely be investigated separately.
While looking into this, I also noticed that resource requests and limits are not defined in the Helm chart. Adding resources.requests and resources.limits would help Kubernetes better manage the operator’s resource consumption and allow it to be restarted automatically without negatively impacting other workloads in the cluster.
Although the memory spike itself may be an application-level issue, I believe defining resource constraints in the Helm chart is a valuable improvement regardless. I’d like to open a PR to add these configurations to the chart.