I have observed in my dev environment, that the memory actually used by Loki chunks cache can reach 4GiB; starting low, but slowly incrementing with usage. The `resources.requests.memory` for this component is set in our values to `500Mi` -- which is too low. We can, and should, leverage the automatic computation done by `loki` chart to automatically align memory requests and limits with the maximum use that memcached can do. Corresponding code snippet from the chart: ``` {{- /* Calculate requested memory as round(allocatedMemory * 1.2). But with integer built-in operators. */}} {{- $requestMemory := div (add (mul .allocatedMemory 12) 5) 10 }} limits: memory: {{ $requestMemory }}Mi requests: cpu: 500m memory: {{ $requestMemory }}Mi {{- end }} ... args: - -m {{ .allocatedMemory }} ```