Skip to content

adr-0040-runtime-container-telemetry.md

Latest commit

 

History

History
62 lines (47 loc) · 3.67 KB

adr-0040-runtime-container-telemetry.md

File metadata and controls

62 lines (47 loc) · 3.67 KB

Workstream ADR 0040: Docker Runtime Container Telemetry

  • ADR: ADR 0040
  • Title: Docker runtime container telemetry
  • Status: live_applied
  • Branch: codex/adr-0040-runtime-container-telemetry
  • Worktree: ../proxmox-host_server-runtime-container-telemetry
  • Owner: codex
  • Depends On: ADR 0011
  • Conflicts With: none
  • Shared Surfaces: docker-runtime, playbooks/monitoring-stack.yml, roles/docker_runtime_observability, managed Grafana dashboards

Scope

  • collect Docker container metrics from docker-runtime through Telegraf's Docker input plugin
  • ship container telemetry into the existing InfluxDB bucket on monitoring
  • extend the managed LV3 docker-runtime Detail dashboard with container-level panels and a runtime snapshot
  • document the convergence and verification path for operators

Non-Goals

  • changing the Docker runtime VM software baseline from ADR 0023
  • changing public publication for runtime-hosted services
  • adding alert rules or notification routing for container health
  • changing protected integration files such as VERSION, changelog.md, README.md, or versions/stack.yaml on the workstream branch

Expected Repo Surfaces

  • playbooks/monitoring-stack.yml
  • roles/docker_runtime_observability/
  • roles/monitoring_vm/templates/_grafana_dashboard_macros.j2
  • roles/monitoring_vm/templates/lv3-vm-detail.json.j2
  • inventory/host_vars/proxmox-host.yml
  • docs/runbooks/monitoring-stack.md
  • docs/repository-map.md
  • docs/adr/0040-docker-runtime-container-telemetry-via-telegraf-docker-input.md
  • workstreams.yaml

Expected Live Surfaces

  • docker-runtime runs telegraf with Docker socket access
  • InfluxDB receives docker_container_* measurements from docker-runtime
  • LV3 docker-runtime Detail shows container-level runtime data

Verification

  • make syntax-check-monitoring
  • ansible -i /Users/live/Documents/GITHUB_PROJECTS/proxmox-host_server/inventory/hosts.yml docker-runtime -m shell -a 'systemctl is-active telegraf && id -nG telegraf' --private-key /Users/live/Documents/GITHUB_PROJECTS/proxmox-host_server/.local/ssh/hetzner_llm_agents_ed25519 -e proxmox_guest_ssh_connection_mode=proxmox_host_jump
  • ssh -i /Users/live/Documents/GITHUB_PROJECTS/proxmox-host_server/.local/ssh/hetzner_llm_agents_ed25519 -o IdentitiesOnly=yes -J ops@100.118.189.95 ops@10.10.10.40 'sudo influx query --host http://127.0.0.1:8086 --org lv3 --token "$(sudo cat /etc/lv3/monitoring/influxdb-operator.token)" '\''from(bucket: "proxmox") |> range(start: -15m) |> filter(fn: (r) => r.host == "docker-runtime" and (r._measurement == "docker_container_status" or r._measurement == "docker_container_cpu" or r._measurement == "docker_container_mem" or r._measurement == "docker_container_net" or r._measurement == "docker_container_health")) |> limit(n: 20)'\'''

Merge Criteria

  • runtime telemetry convergence is idempotent
  • the managed runtime detail dashboard is provisioned from repo state
  • the workstream registry and this document are current
  • protected integration files are reconciled only during integration on main

Notes For The Next Assistant

  • Live apply completed on 2026-03-22 through make converge-monitoring.
  • Verification confirmed telegraf is active on docker-runtime, InfluxDB is receiving docker_container_* rows for uptime-kuma, and the runtime detail dashboard now contains 16 panels.
  • The first live rerun failed because roles/monitoring_vm/tasks/main.yml mixed Restart Grafana and restart grafana; commit aad25fe fixed the handler names before the final idempotent rerun.