Prometheus metrics inconsistently named when using prefix

KumoMTA Integrations
1

Hi,

I’ve setup prometheus via a servicemonitor and I’m scraping the kumomta pods with the params set like:

endpoints:
  - params:
      prefix:
      - kumomta_
    path: /metrics
    targetPort: 8000

which works to nicely namespace the metrics, however I noticed that although most are correctly namespaced when the prefix is added, some appear to be namespaced in the HELP and TYPE comments of the scrape output but the actual metric is not prefixed, eg:

# HELP kumomta_total_readyq_runs total number of times a readyq maintainer was run
# TYPE kumomta_total_readyq_runs counter
kumomta_total_readyq_runs 17
# HELP kumomta_connection_count number of active connections
# TYPE kumomta_connection_count gauge
connection_count{service="smtp_client"} 0
connection_count{service="esmtp_listener"} 0
connection_count{service="lua"} 0

Is this by design or a bug?

Thanks!
Luke

3

it’s a bug! Thanks for reporting this! Fixed in metrics: fix prefix not being applied to metrics with multiple labels · KumoCorp/kumomta@f0c2b49 · GitHub

4

Awesome, thanks! Does a new docker image get built on every GH commit/merge?

5

yes, all our “dev” release binaries (deb, rpm, containers) get built from main on every push to main

6

Thank you, how stable are the ‘dev’ releases? is it advisable to wait for a new ‘official’ release?

7

Generally pretty stable. We’re actually going to tag the next stable release next week.