Fix CNPG postgres metrics exporter (PodMonitor deprecation) #154

New issue

Open

opened 2026-03-24 20:07:05 +00:00 by forgejo_admin · 0 comments

forgejo_admin commented

2026-03-24 20:07:05 +00:00

Contributor

Type

Feature

Lineage

plan-pal-e-platform -> Phase 16 -> 16e investigation

Repo

forgejo_admin/pal-e-platform

User Story

As a platform operator
I want CNPG postgres metrics scraped by Prometheus
So that the TargetDown alert clears and postgres observability is restored

Context

CNPG Cluster pal-e-postgres has enablePodMonitor: true but the auto-generated PodMonitor is being dropped by Prometheus (807 dropped targets). Port 9187 IS listening and serving valid Prometheus metrics over HTTP. enablePodMonitor is deprecated in CNPG 1.28 and will be removed.

Instance image: ghcr.io/cloudnative-pg/postgresql:17.4-1. Operator version: 1.28.1. Impact: Permanent TargetDown alert for postgres/pal-e-postgres.

Fix: Replace with manual PodMonitor kubernetes_manifest resource with proper selector labels. The metrics endpoint serves HTTP (not HTTPS), so no TLS config needed in the PodMonitor.

File Targets

Files to modify:

terraform/main.tf -- add manual PodMonitor resource, remove enablePodMonitor: true from CNPG cluster config

Acceptance Criteria

Manual PodMonitor resource created with correct selector labels
enablePodMonitor: true removed from CNPG cluster config
Prometheus scrapes CNPG metrics successfully (no TargetDown alert)
Metrics visible in Grafana postgres dashboard

Test Expectations

tofu plan shows PodMonitor creation and enablePodMonitor removal
tofu validate passes
Run command: tofu plan -lock=false

Constraints

PodMonitor must use HTTP scheme (not HTTPS) for port 9187
Selector labels must match CNPG pod labels exactly
Follow existing terraform patterns for kubernetes_manifest resources

Checklist

PR opened
Tests pass
No unrelated changes

pal-e-platform -- parent project
phase-platform-16-alert-tuning -- source investigation

### Type Feature ### Lineage `plan-pal-e-platform` -> Phase 16 -> 16e investigation ### Repo `forgejo_admin/pal-e-platform` ### User Story As a platform operator I want CNPG postgres metrics scraped by Prometheus So that the TargetDown alert clears and postgres observability is restored ### Context CNPG Cluster `pal-e-postgres` has `enablePodMonitor: true` but the auto-generated PodMonitor is being dropped by Prometheus (807 dropped targets). Port 9187 IS listening and serving valid Prometheus metrics over HTTP. `enablePodMonitor` is deprecated in CNPG 1.28 and will be removed. Instance image: `ghcr.io/cloudnative-pg/postgresql:17.4-1`. Operator version: 1.28.1. Impact: Permanent TargetDown alert for `postgres/pal-e-postgres`. Fix: Replace with manual `PodMonitor` kubernetes_manifest resource with proper selector labels. The metrics endpoint serves HTTP (not HTTPS), so no TLS config needed in the PodMonitor. ### File Targets Files to modify: - `terraform/main.tf` -- add manual PodMonitor resource, remove `enablePodMonitor: true` from CNPG cluster config ### Acceptance Criteria - [ ] Manual PodMonitor resource created with correct selector labels - [ ] `enablePodMonitor: true` removed from CNPG cluster config - [ ] Prometheus scrapes CNPG metrics successfully (no TargetDown alert) - [ ] Metrics visible in Grafana postgres dashboard ### Test Expectations - [ ] `tofu plan` shows PodMonitor creation and enablePodMonitor removal - [ ] `tofu validate` passes - Run command: `tofu plan -lock=false` ### Constraints - PodMonitor must use HTTP scheme (not HTTPS) for port 9187 - Selector labels must match CNPG pod labels exactly - Follow existing terraform patterns for kubernetes_manifest resources ### Checklist - [ ] PR opened - [ ] Tests pass - [ ] No unrelated changes ### Related - `pal-e-platform` -- parent project - `phase-platform-16-alert-tuning` -- source investigation