gimlet-io
Huge memory footprint
Thank you for this great product.
We tested capacitor on one of our k8s clusters and figured out, that capacitor uses a lot of memory. Our cluster has regarding to flux the following basic data:
- sources: 1
- kustomizations: 2
- helm releases: 67
We are using capacitor v0.3.2 with flux v2.2.3 and kubernetes v1.27.5
Grafana gives us the following memory related graph:
The first and the last peak are single reload requests on the ui, the middle part are 3-times 3 parallel reload requests.
Normal behavior or bug?
We can certainly do better (https://github.com/gimlet-io/capacitor/issues/18) These numbers are not wanted for sure.
Measurement 1
- sources: 8
- kustomizations: 51
- helm releases: 21
Measurement 2
- sources: 3
- kustomizations: 243
- helm releases: 14
No conclusion yet. Slowness is an issue for these clusters. Next test: increase the number of helm charts and measure.
Maybe it's the number of dependencies. We have a lot of dependencies between the helm releases.