__replica__ misbehaving?

Hi everyone,

I’ve got an odd situation on my hands.

I’m running the Agent in Static mode on Kubernetes, making use of both additionalScrapeConfigs on the MetricsInstance, as well as Integrations.

All the metrics coming in from the additionalScrapeConfigs seem to be duplicated on the __replica__ label, e.g. up{__replica__=""} (not visible), as well as up{__replica__="replica-0"}.

From what I gathered on Architecture | Grafana Agent documentation, __replica__ is added by the agent, but I’m not using HA Deduplication at this stage, and want to squash this label.

I’ve tried to relabel metrics on the scrape configs, but seems to be happening further up the chain.

Any ideas on how I can squash it?

Cheers :slight_smile:

I managed to figure it out :slight_smile:

Simply needed to add replicaExternalLabelName: "" to the GrafanaAgent spec, under metrics:. Et voila!

I hope it helps someone else down the line!