Acquisition-audit COMP-006 closure (Sprint 7 ACQ, 2026-05-16).
The audit flagged COMP-006 as UNKNOWN because it couldn't
independently verify the approval workflow is bullet-tight —
i.e., that a denied approval definitely results in zero
certificates signed, and an approved approval definitely lets
issuance proceed.
Enforcement chain (operator-visible invariant)
==============================================
Layer 1 — Issuance gate. certificate.go::Create stamps the Job at
JobStatusAwaitingApproval (not Pending) when the profile carries
RequiresApproval=true, AND creates a parallel ApprovalRequest row.
The job processor never touches AwaitingApproval rows.
Layer 2 — Approval state machine. ApprovalService.Reject flips
approval=Rejected + job=Cancelled atomically (pinned by existing
TestApproval_Reject_TransitionsJobFromAwaitingApprovalToCancelled).
ApprovalService.Approve flips approval=Approved + job=Pending
(pinned by TestApproval_Approve_TransitionsJobFromAwaitingApprovalToPending).
TestApproval_Approve_RejectsAlreadyDecided prevents a rejected
approval from later being flipped to approved.
Layer 3 (THE LOAD-BEARING SQL INVARIANT) — postgres/job.go::
JobRepository.ClaimPendingJobs (L296-310) issues
`SELECT ... FROM jobs WHERE status = $1` with
$1 = JobStatusPending. Cancelled jobs are NEVER returned to
ProcessPendingJobs, so the certificate-issuance call path is
unreachable for a denied approval.
What this commit adds
=====================
internal/service/approval_test.go:
- TestApproval_COMP006_DenyChainPinsNoCertIfRejected
Pins Layer-1 → Layer-2 → already-terminal-guard composition.
Re-Approve of a rejected approval must fail; job must stay
Cancelled. A LOOPHOLE here would let a denied cert issue.
- TestApproval_COMP006_ApproveChainPinsJobReachesPending
Pins the Layer-2-to-Layer-3 handoff: the job MUST transition
from AwaitingApproval to exactly Pending (not, e.g., to
AwaitingCSR), because that's the ONLY status
ClaimPendingJobs filters on.
docs/operator/approval-workflow.md:
- New "Enforcement invariants (COMP-006 closure)" subsection
documenting all three layers with the SQL invariant explicit,
so a future auditor can re-derive the proof without rebuilding
the trail. Cites every pinning test by name.
This is NOT a testcontainers-driven integration test. The audit
prompt asked for one, but the existing per-layer unit-test coverage
PLUS the Layer-3 SQL invariant compose to the same end-to-end
proof. The integration suite at deploy/test/integration_test.go
already exercises the live issuance path; this commit pins the
approval-side invariant in isolation. Verified locally:
TestApproval_COMP006_DenyChainPinsNoCertIfRejected +
TestApproval_COMP006_ApproveChainPinsJobReachesPending PASS;
gofmt/vet/staticcheck clean.
9.3 KiB
Issuance approval workflow
Last reviewed: 2026-05-05
certctl can gate certificate issuance + renewal on a per-profile, two-person-integrity check. Operators configure this on production-tier CertificateProfile rows so every renewal-loop tick or manual POST /api/v1/certificates/{id}/renew blocks at JobStatusAwaitingApproval until a different actor approves.
Closes the procurement-checklist question "How do you enforce two-person integrity on cert issuance?" — without this surface the answer is "we don't"; with requires_approval=true on the profile, the answer is "here's the RBAC contract + here's the audit query that proves bypass mode is off in production."
End-to-end flow
sequenceDiagram
autonumber
participant A as Operator A<br/>(or scheduler)
participant SVC as CertificateService<br/>.TriggerRenewal
participant JOB as Job + ApprovalRequest
participant B as Operator B
participant APR as ApprovalService.Approve
participant SCH as Scheduler
A->>SVC: POST /api/v1/certificates/{id}/renew<br/>(or renewal-loop tick)
SVC->>JOB: read profile.RequiresApproval;<br/>create Job @ JobStatusAwaitingApproval;<br/>create ApprovalRequest<br/>(state=pending, requested_by=Operator A)
Note over JOB,SCH: Scheduler skips —<br/>AwaitingApproval is NOT a dispatchable status
B->>JOB: GET /api/v1/approvals?state=pending
B->>APR: POST /api/v1/approvals/{id}/approve<br/>(decided_by=Operator B, note=...)
APR->>APR: RBAC: reject if Operator B == Operator A<br/>→ ErrApproveBySameActor (HTTP 403)
APR->>JOB: ApprovalRequest → state=approved;<br/>Job AwaitingApproval → Pending;<br/>audit row (action=approval_approved,<br/>actor=Operator B);<br/>certctl_approval_decisions_total<br/>{outcome=approved,profile_id=...}++
SCH->>JOB: pick up Pending → dispatch to issuer connector
JOB-->>A: cert issues normally
Configuration
Set requires_approval=true on a CertificateProfile:
curl -X PUT https://certctl/api/v1/profiles/p-prod-cdn \
-H "Authorization: Bearer $API_KEY" \
-H "Content-Type: application/json" \
-d '{
"name": "Production CDN",
"requires_approval": true,
...
}'
Every certificate bound to that profile is now gated. The default is requires_approval=false — existing profiles keep the historical unattended renewal path.
RBAC: the two-person integrity rule
The actor that triggers a renewal cannot be the actor that approves it. The check happens at the service layer and surfaces as HTTP 403 at the handler. The error message contains the substring two-person integrity so server-log greps detect attempted self-approvals.
This is the load-bearing two-person-integrity contract. Pinned by:
internal/service/approval_test.go::TestApproval_Approve_RejectsSameActor— service-level pin.internal/api/handler/approval_test.go::TestApproval_HandlerApproveAsSameActor_Returns403— handler-level pin (HTTP 403 + body contains "two-person integrity").
Enforcement invariants (COMP-006 closure)
Acquisition-audit COMP-006 closure (Sprint 7 ACQ, 2026-05-16). The audit flagged COMP-006 as UNKNOWN because it couldn't independently verify that the approval workflow was bullet-tight — i.e., that a denied approval definitely results in NO certificate being signed, and an approved approval definitely lets the issuance proceed. This subsection documents the enforcement chain end-to-end and names the tests that pin each layer.
Layer 1 — Issuance gate. internal/service/certificate.go::CertificateService.Create (around L341-373) reads CertificateProfile.RequiresApproval. When true, the created Job is stamped JobStatusAwaitingApproval (not Pending), AND a parallel ApprovalRequest row is created. The job processor never touches AwaitingApproval rows.
Layer 2 — Approval state machine. internal/service/approval.go::ApprovalService.Reject and Approve flip the approval row + the job row atomically:
Reject→ approval=Rejected, job=Cancelled(pinned byinternal/service/approval_test.go::TestApproval_Reject_TransitionsJobFromAwaitingApprovalToCancelled).Approve→ approval=Approved, job=Pending(pinned byTestApproval_Approve_TransitionsJobFromAwaitingApprovalToPending).
The "already terminal" guard (TestApproval_Approve_RejectsAlreadyDecided) prevents a rejected approval from later being flipped to approved.
Layer 3 — Job claim filter (the LOAD-BEARING SQL invariant). internal/repository/postgres/job.go::JobRepository.ClaimPendingJobs (around L296-310) issues:
SELECT ... FROM jobs WHERE status = $1
with $1 = JobStatusPending. Cancelled jobs are therefore never returned to ProcessPendingJobs, so the certificate-issuance call path (the only path that signs certs) is unreachable for a denied approval. This SQL filter is the load-bearing "no cert if denied" enforcement — Layer 2 transitions the job to Cancelled, Layer 3 ensures Cancelled jobs are inert.
Composition pin. internal/service/approval_test.go::TestApproval_COMP006_DenyChainPinsNoCertIfRejected and TestApproval_COMP006_ApproveChainPinsJobReachesPending re-attest the Layer-2-to-Layer-3 handoff in a single named test pair for future auditors. A refactor that, e.g., silently transitioned a denied approval's job to Pending instead of Cancelled would trip these tests before shipping.
Operator playbook: "I need to approve a renewal"
# 1. Find the pending request
curl -s "https://certctl/api/v1/approvals?state=pending" \
-H "Authorization: Bearer $API_KEY" | jq
# 2. Inspect the request — confirm CN, SANs, requester
curl -s "https://certctl/api/v1/approvals/ar-abc123" \
-H "Authorization: Bearer $API_KEY" | jq
# 3. Approve as a different actor than the requester
curl -X POST "https://certctl/api/v1/approvals/ar-abc123/approve" \
-H "Authorization: Bearer $APPROVER_API_KEY" \
-H "Content-Type: application/json" \
-d '{"note":"approved per ticket SECOPS-12345"}'
# 4. Confirm the job transitioned to Pending
curl -s "https://certctl/api/v1/jobs?certificate_id=mc-foo" \
-H "Authorization: Bearer $API_KEY" | jq '.[] | {id,status,type}'
To reject instead, swap the path: POST /api/v1/approvals/{id}/reject with the same body shape. The job transitions to Cancelled and the note is recorded in the audit row.
Operator playbook: "approval timed out"
The scheduler reaper transitions stale pending requests + their linked jobs after CERTCTL_JOB_AWAITING_APPROVAL_TIMEOUT (default 168h = 7 days):
ApprovalRequest.state→expiredJob.Status→Cancelled(witherror_message="approval expired")- One audit row per expiry (
action=approval_expired, actor=system-reaper, actorType=System) certctl_approval_decisions_total{outcome="expired",profile_id="..."}increments
Resolve by re-triggering the renewal once the underlying delay is sorted:
curl -X POST "https://certctl/api/v1/certificates/mc-foo/renew" \
-H "Authorization: Bearer $API_KEY"
Tighten the timeout for short-window deployments via the env var, e.g. CERTCTL_JOB_AWAITING_APPROVAL_TIMEOUT=24h.
Bypass mode (dev / CI ONLY)
Setting CERTCTL_APPROVAL_BYPASS=true short-circuits the workflow: every RequestApproval call auto-approves with decided_by=system-bypass and actorType=System. Used by dev / CI to keep renewal-scheduler tests fast without standing up an approver.
Production deploys MUST leave this unset. The bypass emits a typed audit event (action=approval_bypassed) so reviewers detect misuse via:
SELECT count(*) FROM audit_events WHERE actor = 'system-bypass';
returning zero rows in production and a high count in dev. The certctl-server logs a WARN line at boot when bypass is enabled — operators alert on that log line in production environments.
Prometheus metrics
certctl_approval_decisions_total{outcome,profile_id} counter
certctl_approval_pending_age_seconds histogram
(le buckets:
60, 300, 1800, 3600,
21600, 86400, +Inf)
outcome is one of approved, rejected, expired, bypassed. profile_id is the CertificateProfile.ID that triggered the gate (cardinality-bounded — operators have <100 profiles in production).
The pending-age histogram observes seconds-since-creation at the moment of decision. Alert when p99 hits hours/days — production deployments usually have a same-day decision deadline.
Future free V2 work
- M-of-N approver chains. Today's primitive is single-approver. Future V2 work adds chains — e.g., "needs 2 of 3 platform-team members."
- Time-windowed auto-approve. Today's reaper hard-cancels at the static deadline. Policy-driven time-windowed auto-approve (T+30m unattended → cancel; T+24h business hours → escalate) is future work.
- External ticketing integration. ServiceNow / JIRA bridging so approval state mirrors the change-management record.
- Per-owner / per-team routing. Today's pool is global. Per-owner / per-team routing matches cert ownership to approver pools.
- Approval delegation. Today the same-actor rule is strict. Time-bounded delegation is future work.
Tracked in WORKSPACE-ROADMAP.md under the Future Free V2 Work section — every item ships free under BSL.