[aws-for-fluent-bit] Is this chart the recommended way to setup FluentBit on AWS EKS?

[trallnag]

Asking because the AWS documentation does not mention this chart anywhere.

[matti]

related aws/aws-for-fluent-bit#357

[mengmann]

We are not only dealing with an inconsistency in the way the FluentBit installation is supposed to be done according to the official AWS documentation (custom kubectl manifests vs. helm chart). Rather, we are also dealing with a difference in content. This can be seen just by looking at the ConfigMap for Cloudwatch Logs integration.

The referenced kubectl manifests (this and this) from AWS ensures that the logs are sorted into different LogGroups depending on their origin / meaning: application, host, dataplane. This is done via different include files inside the fluent-bit.conf. In the default version of the ConfigMap from the Helm chart, only a simple vanilla config is generated.

So why do we have different default settings here? How can the helm chart ConfigMap be overridden to look like the "official" variant regarding Cloudwatch Logs? This seems currently not possible because the current configmap template is very limited.

[joebowbeer]

FWIW, we've opted to use the fluent-bit helm chart from https://fluent.github.io/helm-charts

and then patch in the image provided by AWS: public.ecr.aws/aws-observability/aws-for-fluent-bit