Summary
Add an optional LGTM stack (Loki, Grafana, Tempo, Mimir) as subchart dependencies to the rstudio-connect chart, preconfigured to receive telemetry from Connect via OpenTelemetry.
This would give operators a path to full observability when deploying Connect on Kubernetes.
Motivation
Connect 2026.03.0 deprecated the legacy Metrics.Enabled / Graphite-based pipeline in favor of OpenTelemetry (see #820). With that transition, Connect can now emit metrics, traces, and logs over standard OTel protocols — but users are left to wire up their own backend.
The LGTM stack is the natural Grafana-ecosystem target for OTel data:
| Signal |
Destination |
| Metrics |
Mimir (or Prometheus) |
| Logs |
Loki |
| Traces |
Tempo |
| Dashboards |
Grafana |
Bundling these as opt-in subcharts with sane defaults would make Connect observability accessible out of the box for teams that don't already have a monitoring stack.
Proposed approach
-
Add optional subchart dependencies in Chart.yaml (all disabled by default):
-
Preconfigure Connect OTel settings in values.yaml when the subcharts are enabled — pointing Connect's OTel exporter endpoints at the in-cluster Mimir, Tempo, and Loki services.
-
Ship a starter Grafana dashboard (as a ConfigMap) covering the key Connect OTel metrics exposed since 2026.03.0.
-
Document a values.yaml snippet that enables the full stack in one block, suitable for local dev/demo use.
Related
Acceptance criteria
Summary
Add an optional LGTM stack (Loki, Grafana, Tempo, Mimir) as subchart dependencies to the
rstudio-connectchart, preconfigured to receive telemetry from Connect via OpenTelemetry.This would give operators a path to full observability when deploying Connect on Kubernetes.
Motivation
Connect 2026.03.0 deprecated the legacy
Metrics.Enabled/ Graphite-based pipeline in favor of OpenTelemetry (see #820). With that transition, Connect can now emit metrics, traces, and logs over standard OTel protocols — but users are left to wire up their own backend.The LGTM stack is the natural Grafana-ecosystem target for OTel data:
Bundling these as opt-in subcharts with sane defaults would make Connect observability accessible out of the box for teams that don't already have a monitoring stack.
Proposed approach
Add optional subchart dependencies in
Chart.yaml(all disabled by default):grafana/grafanagrafana/lokigrafana/tempografana/mimir-distributed(orprometheus-community/prometheusas a lighter alternative)Preconfigure Connect OTel settings in
values.yamlwhen the subcharts are enabled — pointing Connect's OTel exporter endpoints at the in-cluster Mimir, Tempo, and Loki services.Ship a starter Grafana dashboard (as a ConfigMap) covering the key Connect OTel metrics exposed since 2026.03.0.
Document a
values.yamlsnippet that enables the full stack in one block, suitable for local dev/demo use.Related
Metrics.Enabledsetting and is a prerequisite for this workAcceptance criteria