Can I use transformation type sample for dynamic sampling

[shamj]

I have csv file that has schema for service and sampled_rate.
While using transformation of type sample (below config) : the output shows sampled_rate as 1 , not the one in csv file. Can I use key as rate ?
dynamic_sample:
type: sample
inputs:
- add_sample_rate
group_by: "{{ .service }}"
rate: 1
sample_rate_key: .sample_rate

[shamj]

Could someone please advise if vector supports dynamic sampling?

[pront]

Hi @shamj, yes, you can create groups based on event values with this setting: https://vector.dev/docs/reference/configuration/transforms/sample/#group_by

[shamj]

Thank you @pront . Let's say I have 20 services or more each one needs to be sampled at different rates. I have added those rations and corresponding services in csv, but I am reading service and rate as service and sample_rate , on testing below, it gives me error : ERROR vector::cli: Configuration error. error=invalid type: string "{{ .sample_rate }}", expected u64
type: sample
inputs:

• add_sample_rate
  group_by: "{{ .service }}"
  rate: "{{ .sample_rate }}"
  sample_rate_key: .sample_rate

But when I hardcode the value for rate, it works fine. How can I sample it based on .sample_rate.

[pront]

This is not supported at the moment. You can the type on each field. The rate field only accepts unints. You can create a feature request for this.

We recently added the ability to have templates for unsigned numbers: #22243