Is there an existing issue for this?
Feature description
The AWS/EC2 namespace contains metrics scoped to VPCs rather than EC2 instances, such as NetworkAddressUsage, which uses the Per-VPC-Metrics dimension (e.g. Per-VPC-Metrics=vpc-0abc123ef).
yet-another-cloudwatch-exporter only discovers ec2:instance resources for AWS/EC2 jobs. When CloudWatch returned a metric with a Per-VPC-Metrics dimension, the associator found no matching resource and emitted the metric as a global metric with name="global" and empty tag labels.
We would like for tags that are attached to the VPC to be surfaced in the resulting metrics, similar to how tags on EC2 instances are surfaced on EC2 related metrics. This helps us to meet prometheus labelling requirements enforced by the observability platform.
What might the configuration look like?
apiVersion: v1alpha1
sts-region: eu-central-1
nilToZero: true
discovery:
exportedTagsOnMetrics:
AWS/EC2:
- my-tag
jobs:
- type: AWS/EC2
regions:
- eu-central-1
metrics:
- name: NetworkAddressUsage
statistics: [Average]
period: 60
length: 600
Anything else?
I have already opened a PR with our suggested change in it
Is there an existing issue for this?
Feature description
The
AWS/EC2namespace contains metrics scoped to VPCs rather than EC2 instances, such asNetworkAddressUsage, which uses thePer-VPC-Metricsdimension (e.g.Per-VPC-Metrics=vpc-0abc123ef).yet-another-cloudwatch-exporter only discovers
ec2:instanceresources forAWS/EC2jobs. When CloudWatch returned a metric with aPer-VPC-Metricsdimension, the associator found no matching resource and emitted the metric as a global metric with name="global" and empty tag labels.We would like for tags that are attached to the VPC to be surfaced in the resulting metrics, similar to how tags on EC2 instances are surfaced on EC2 related metrics. This helps us to meet prometheus labelling requirements enforced by the observability platform.
What might the configuration look like?
Anything else?
I have already opened a PR with our suggested change in it