pg_cnpg_dr_readiness — Measure DR Readiness Before It's Too Late

What Does This Tool Do?

pg_cnpg_dr_readiness calculates a weighted DR readiness score from 0 to 100 that measures the actual disaster recovery capability of a CNPG cluster. The tool queries the Kubernetes API for both primary and replica cluster status — across different kubecontexts — and evaluates five critical dimensions:

• Replica Health (30 points) — Are all replicas in the target cluster healthy and in streaming state?
• WAL Archive Lag (20 points) — Is the last WAL archive less than 5 minutes old?
• Replication Lag (20 points) — Is replication lag under 10 MB?
• Object Store Accessible (15 points) — Is the configured object store (MinIO, S3) reachable?
• Instance Count (15 points) — Is at least one instance running in the DR cluster?

The result is a clear status: READY (score ≥ 80), AT-RISK (score 50–79), or NOT-READY (score < 50).

How Is It Called?

Configuration

DR tools are configuration-driven. Instead of a direct database connection, system and environment are read from the DR config — allowing multiple cluster pairs to be managed:

Claude CLI (Natural Language)

MCP Protocol (Programmatic)

What Happens Under the Hood?

This tool executes no SQL. Instead, it uses the Kubernetes CustomObjectsApi to query CNPG cluster status from two different kubecontexts:

1. Load DR config — Reads the master reference configuration (system + environment) to determine primary and replica cluster coordinates
2. Query primary cluster — kubectl get cluster/db-primary -n app -o json in context k8s-primary
3. Query replica cluster — kubectl get cluster/db-replica -n app -o json in context k8s-replica
4. Check WAL archive — Reads .status.lastSuccessfulArchive from the CNPG cluster object
5. Test object store — Verifies reachability of the configured S3/MinIO endpoint

The weighted formula allows partial scores: if replication lag is 4.2 MB (under the 10 MB threshold but not zero), it awards 15 out of 20 possible points instead of the full 20.

Lifecycle Gates

The tool respects the tenant lifecycle:

• provisioning — Read-only access (score is calculated but no actions taken)
• active — Full access, all checks active
• maintenance — Blocked unless override is set
• decommissioned — Fully blocked

Output in Detail

• score — Weighted total score from 0 to 100
• status — READY (≥ 80), AT-RISK (50–79), or NOT-READY (< 50)
• checks — Array with 5 entries, each containing:
  • name — Check identifier (replica_health, wal_archive_lag, etc.)
  • weight — Maximum points for this check
  • score — Actual score achieved
  • pass — Boolean indicating whether the check passed
  • detail — Human-readable status text

What Does the Test Suite Validate?

The test suite validates pg_cnpg_dr_readiness against a lab environment with a primary cluster (2 instances) and a replica cluster (1 instance) on separate k3s clusters, with MinIO as WAL archive:

Test Data Scenario

The lab environment consists of:

• k8s-primary — k3s cluster with CNPG cluster db-primary (2 instances: 1 primary + 1 replica)
• k8s-replica — Separate k3s cluster with CNPG cluster db-replica (1 instance, WAL replay)
• MinIO — Object store for WAL archiving (S3-compatible, port 9000)
• Two kubecontexts — Simulates a real multi-cluster DR topology

Tests validate that:

1. The result contains a score (number) and status (string)
2. The checks array has exactly 5 entries (one per dimension)
3. With healthy clusters the status returns READY
4. With a missing replica the status switches to NOT-READY

Summary

pg_cnpg_dr_readiness is the central tool for DR monitoring in CNPG environments. It replaces manual kubectl checks across multiple clusters with a single weighted score that immediately shows whether a DR failover would be technically feasible.

Combined with pg_cnpg_switchover (→ actual failover) and pg_failover_readiness (→ replication status), it forms the core of the /pg-dr workflow skill.