Description
Component(s)
No response
What happened?
Describe the bug
After upgrading the OpenTelemetry Operator and Collector to the following versions:
image:
repository: 'ghcr.io/open-telemetry/opentelemetry-operator/opentelemetry-operator'
tag: '0.125.0'
collectorImage:
repository: 'ghcr.io/open-telemetry/opentelemetry-collector-releases/opentelemetry-collector-contrib'
tag: '0.126.0'
The exported_job label in metrics (e.g., target_info) started including the namespace as part of the value, changing from just the job name to the format namespace/job_name.
Previously, the label had the value:
exported_job="APPREDACTED"
Now it appears as both:
exported_job="APPREDACTED"
AND exported_job="production/APPREDACTED"
Where production is the namespace of the resource.
This change is causing duplicated time series and confusion in dashboards and alerts that rely on the exported_job label.
Steps to reproduce
Steps to reproduce
- Deploy the OpenTelemetry Operator and Collector using the versions mentioned above.
- Deploy an Instrumentation resource in the production namespace.
- Observe the target_info metric in Prometheus.
- Note that the exported_job label appears with two different values:
APPREDACTED
AND production/APPREDACTED
What did you expect to see?
What did you expect to see?
I expected to see the exported_job label value remain consistent as before, containing only the job name: APPREDACTED, regardless of the namespace.
What did you see instead?
The exported_job label now has two different formats:
APPREDACTED
AND production/APPREDACTED
This new behavior introduces duplicates and inconsistencies in metrics.
Collector version
0.126.0
Environment information
INF Target K8s version is 1.31.7-eks-bcf3d70
OpenTelemetry Collector configuration
telemetry:
metrics:
level: normal
readers:
- pull:
exporter:
prometheus:
host: '[${env:MY_POD_IP}]'
port: 8888
Log output
Additional context
No response