Files
NexaPG/docs/security/secret-management.md
nessi 597579376f
Some checks are pending
Container CVE Scan (development) / Scan backend/frontend images for CVEs (push) Waiting to run
Migration Safety / Alembic upgrade/downgrade safety (push) Successful in 2m43s
PostgreSQL Compatibility Matrix / PG14 smoke (push) Successful in 8s
PostgreSQL Compatibility Matrix / PG15 smoke (push) Successful in 8s
PostgreSQL Compatibility Matrix / PG16 smoke (push) Successful in 8s
PostgreSQL Compatibility Matrix / PG17 smoke (push) Successful in 7s
PostgreSQL Compatibility Matrix / PG18 smoke (push) Successful in 7s
Proxy Profile Validation / validate (push) Successful in 3s
Python Dependency Security / pip-audit (block high/critical) (push) Successful in 26s
Docker Publish (Release) / Build and Push Docker Images (release) Successful in 1m41s
[NX-204 Issue] Add secret management guidelines and enhance security notes
Introduced a comprehensive guide for secure production secret handling (`docs/security/secret-management.md`). Updated `.env.example` files with clearer comments on best practices, emphasizing not hardcoding secrets and implementing rotation strategies. Enhanced README with a new section linking to the secret management documentation.
2026-02-15 12:29:40 +01:00

75 lines
2.5 KiB
Markdown

# Secret Management (Production)
This guide defines secure handling for NexaPG secrets in production deployments.
## In Scope Secrets
- `JWT_SECRET_KEY`
- `ENCRYPTION_KEY`
- `DB_PASSWORD`
- SMTP credentials (configured in Admin Settings, encrypted at rest)
## Do / Don't
## Do
- Use an external secret source (Vault, cloud secret manager, orchestrator secrets, or CI/CD secret injection).
- Keep secrets out of Git history and out of image layers.
- Use strong random values:
- JWT secret: at least 32+ bytes random
- Fernet key: generated via `Fernet.generate_key()`
- Restrict access to runtime secrets (least privilege).
- Rotate secrets on schedule and on incident.
- Store production `.env` with strict permissions if file-based injection is used:
- owner-only read/write (e.g., `chmod 600 .env`)
- Audit who can read/update secrets in your deployment platform.
## Don't
- Do **not** hardcode secrets in source code.
- Do **not** commit `.env` with real values.
- Do **not** bake production secrets into Dockerfiles or image build args.
- Do **not** share secrets in tickets, chat logs, or CI console output.
- Do **not** reuse the same secrets between environments.
## Recommended Secret Providers
Pick one of these models:
1. Platform/Cloud secrets
- AWS Secrets Manager
- Azure Key Vault
- Google Secret Manager
2. HashiCorp Vault
3. CI/CD secret injection
- Inject as runtime env vars during deployment
4. Docker/Kubernetes secrets
- Prefer secret mounts or orchestrator-native secret stores
If you use plain `.env` files, treat them as sensitive artifacts and protect at OS and backup level.
## Rotation Basics
Minimum baseline:
1. `JWT_SECRET_KEY`
- Rotate on schedule (e.g., quarterly) and immediately after compromise.
- Expect existing sessions/tokens to become invalid after rotation.
2. `ENCRYPTION_KEY`
- Rotate with planned maintenance.
- Re-encrypt stored encrypted values (target passwords, SMTP password) during key transition.
3. `DB_PASSWORD`
- Rotate service account credentials regularly.
- Apply password changes in DB and deployment config atomically.
4. SMTP credentials
- Use dedicated sender account/app password.
- Rotate regularly and after provider-side security alerts.
## Operational Checklist
- [ ] No production secret in repository files.
- [ ] No production secret in container image metadata or build args.
- [ ] Runtime secret source documented for your environment.
- [ ] Secret rotation owner and schedule defined.
- [ ] Incident runbook includes emergency rotation steps.