External_labels not being honoured

[johnmalone]

Hi,

I tried out 0.17.0rc2 and discovered that our external_labels settings were no longer appearing in our metrics.

Top of prometheus.yml:

global:
  scrape_interval: 60s
  evaluation_interval: 60s
  scrape_timeout: 30s
  external_labels:
    env: "dev"
    region: "eu-west-1"

We've been on an old version for a while so I had to update "labels" to external_labels above and I've tried 17.0rc1 and 16.2 and my custom labels don't seem to appear anywhere. Is there something else I need to do to get my labels back?

Thanks

John

[johnmalone]

I've been looking in the graph console expecting to see my above external_labels. In prod, running our old version, I see:

prometheus_local_storage_memory_chunks{env="production",instance="localhost:3000",job="prometheus",region="eu-west-1"}

While on my upgraded version I see:

prometheus_local_storage_memory_chunks{instance="localhost:3000",job="prometheus"}

[fabxc]

This was an intentional change in behavior – to make that clear the option was named to external_labels.
The labels will be applied to time series that are leaving the scope of your Prometheus server. This means specifically alerts and federation.

One of the motivations was that when attaching labels when storing ingested time series, a change to those labels would break all time series unnecessarily.

[johnmalone]

Thanks for the information.

John

[lock]

This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.