apology in advance if this is documented somewhere we haven

Question

apology in advance if this is documented somewhere we haven t found or answered already we ve been looking at memory for capacity planning and have a Helm deployed 8GB limited QuestDB instance running on one of our k8s clusters we recently began scraping metrics off of it I m trying to get to the bottom of the `questdb memory mem used` metric which occasionally sees excursions way beyond the resource limits does anyone have a good handle on what contributes to this metric and what we could be seeing

Hasitha Dharmasiri · Answer

for reference the `NATIVE ` tagged metrics seem much more sane in the same time period

Andrey Pechkurov · Answer

Hi Hasitha We have the list of Prometheus metrics documented here <https questdb io docs third party tools prometheus> `questdb memory mem used` includes all native memory allocations which may include virtual memory if it wasn t touched yet

Andrey Pechkurov · Answer

This metric includes mmapped files so that s why its value is that big

Hasitha Dharmasiri · Answer

hm I guess I d expect the sum of everything tagged NATIVE to equal it

Hasitha Dharmasiri · Answer

ah got it

Hasitha Dharmasiri · Answer

that would do it so accessing the larger tables would blow up the number I assume

Andrey Pechkurov · Answer

Yup that s right However mmapped memory is elastic and mostly virtual Only free memory is used for the page cache so it s fine if this metric has a large value

Miguel Arregui · Answer

distinct and join queries would make use of this metric

Hasitha Dharmasiri · Answer

yup thanks for the clarification

javier ramirez · Answer

Hashitha I find your question and the replies by Andrei and Miguel very relevant and it could help other users Do you mind if I post it as a Q amp A on the <https stackoverflow com questions tagged questdb|StackOverflow questdb tag>