Dynatrace CSI-Driver asking more request than limit

[Deepak_Shingote]

Hi All,

We have noticed that during the integration of Kubernetes in our DT environment, the CSI Driver is consistently configured with CPU requests that exceed its limits, leading to continuous CPU throttling. This has been observed in multiple tenants. Could someone help us understand the reasons behind the default high CPU request settings for the CSI Driver? What factors contribute to this configuration, and are there any recommended adjustments to mitigate the throttling issues?

Regards,
Deepak

[radek_jasinski]

Hi @Deepak_Shingote 

If your cluster is already saturated or under heavy load, the provisioner will be starved for CPU, causing throttling and slower pod init. For small clusters you can safely trim CPU to 100–200m. For larger clusters consider maintaining closer to the default.

https://docs.dynatrace.com/docs/ingest-from/setup-on-k8s/guides/deployment-and-configuration/resourc...

Regrads,

Radek

[Deepak_Shingote]

Hi Radek,

Thanks for the help. But, my concern here is why DT is changing the request and limit ratio by itself without any manual changes in Yaml file? Also, why request is 5x than limit for CSI-Driver?

Regards,
Deepak 

[radek_jasinski]

Hi,

To the best of my knowledge, Dynatrace does not modify the requests or limits values in YAML files on its own. However, this can happen if default values from the helm chart are applied during installation or upgrade, or if Kubernetes mechanisms like LimitRanges or mutating webhooks are in place and automatically adjust container resource configurations.

Radek