Hello,
we are using DICOM-RST inside a Kubernetes cluster and observed that memory usage is steadily growing over time:
We observed similar behaviour with other multithreaded, tokio based web-services we deployed. According to our research this is related to the glibc memory allocator, which doesn't behave very well in multithreaded, long-running scenarios.
We solved this by switching to mimalloc, maybe we can investigate what would be a good solution?
Hello,
we are using DICOM-RST inside a Kubernetes cluster and observed that memory usage is steadily growing over time:
We observed similar behaviour with other multithreaded, tokio based web-services we deployed. According to our research this is related to the glibc memory allocator, which doesn't behave very well in multithreaded, long-running scenarios.
We solved this by switching to mimalloc, maybe we can investigate what would be a good solution?