Component(s)

processor/k8sattributes

What happened?

The k8sattributes processor currently assumes that the host.name resource attribute always contains a pod IP.

When an upstream component reports a textual hostname instead—say, metrics-indexer—the processor still tries to look up a pod using that string and ends up returning metadata for whichever pod last registered with that hostname. The effect is that spans or metrics from one workload can suddenly carry labels from an unrelated workload, such as k8s_statefulset_name="metrics-indexer" appearing on traffic originating from another service, which both confuses debugging and inflates label cardinality.

Collector version

76b4c60

Environment information

No response

OpenTelemetry Collector configuration

Log output

Additional context

No response

Tip

_{React with 👍 to help prioritize this issue. Please use comments to provide useful context, avoiding +1 or me too, to help us triage it. Learn more here.}