Simplify GitOps bootstrap workflow for complex customer deployments

[paulczar]

Summary

Reduce GitOps bootstrap complexity and failure modes observed in customer deployments (Transport for NSW, NBN).

Context

The team reported difficulties with GitOps bootstrapping in production customer contexts. Paul Foster's proposed mitigation — extract only essential variables from Terraform into a cluster-adjacent config file — is tracked in #27, but bootstrap script and Helm chart flow may need separate hardening.

Pain points to investigate:

• scripts/cluster/bootstrap-gitops.sh complexity and required env vars
• Customer-specific Helm values / Argo Application wiring
• Interaction between Terraform outputs and bootstrap-time configuration
• Failure recovery and idempotency

Acceptance Criteria

• Document current bootstrap flow and failure modes from TfNSW/NBN experiences
• Identify minimum inputs required at bootstrap vs. deferrable to GitOps
• Propose simplifications (align with Refactor Terraform variables into external configuration file #27 external config approach)
• Improve validation/error messages in bootstrap scripts
• Test simplified flow on at least one cluster profile (public, BYO VPC, egress-zero)
• Update docs/operations/bootstrap-gitops.md

Meeting reference

2026-07-03 — GitOps bootstrap difficulties (~00:06:43)

Related

• Refactor Terraform variables into external configuration file #27 — Refactor Terraform variables into external configuration file
• Implement dynamic IDP passwords via bootstrap scripts #29 — Dynamic IDP passwords during bootstrap