Merge pull request 'Workstream I: Implement VS-04 Audit Trail + GDPR' (#24) from feat/I-vs04-audit-trail into main
deploy / notify (push) Has been cancelled
deploy / deploy (push) Has been cancelled

Reviewed-on: #24
This commit was merged in pull request #24.
This commit is contained in:
2026-08-07 17:18:15 +09:00
25 changed files with 3312 additions and 0 deletions
@@ -0,0 +1,130 @@
-- Migration 0033: Market Data Import Logs (KRX, OpenDart, KIS)
-- Purpose: Append-only audit trail for external API data imports with PIT tracking
-- ============================================================================
-- MARKET_DATA SCHEMA: Import Audit & Evidence
-- ============================================================================
CREATE SCHEMA IF NOT EXISTS market_data;
-- KRX OpenAPI import log (indices, stocks, sectors)
CREATE TABLE IF NOT EXISTS market_data.krx_imports (
id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
import_at TIMESTAMP WITH TIME ZONE NOT NULL,
row_count INT NOT NULL,
checksum VARCHAR(256), -- SHA256 of imported data for deduplication
status VARCHAR(50) NOT NULL, -- 'SUCCESS', 'FAILURE', 'PARTIAL'
error_message TEXT,
details JSONB, -- Event-specific metadata (endpoint, records_skipped, api_latency_ms)
published_at TIMESTAMP WITH TIME ZONE NOT NULL DEFAULT CURRENT_TIMESTAMP,
correlation_id UUID NOT NULL,
revision INT NOT NULL DEFAULT 1,
CONSTRAINT krx_imports_status_check CHECK (status IN ('SUCCESS', 'FAILURE', 'PARTIAL'))
);
CREATE INDEX IF NOT EXISTS idx_krx_imports_import_at ON market_data.krx_imports(import_at DESC);
CREATE INDEX IF NOT EXISTS idx_krx_imports_status ON market_data.krx_imports(status);
CREATE INDEX IF NOT EXISTS idx_krx_imports_correlation_id ON market_data.krx_imports(correlation_id);
CREATE INDEX IF NOT EXISTS idx_krx_imports_published_at ON market_data.krx_imports(published_at);
-- OpenDart API import log (company disclosures, quarterly financials)
CREATE TABLE IF NOT EXISTS market_data.opendart_imports (
id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
import_at TIMESTAMP WITH TIME ZONE NOT NULL,
row_count INT NOT NULL,
checksum VARCHAR(256), -- SHA256 of imported data for deduplication
status VARCHAR(50) NOT NULL, -- 'SUCCESS', 'FAILURE', 'PARTIAL'
error_message TEXT,
details JSONB, -- Event-specific metadata (api_endpoint, query_params, quota_used)
published_at TIMESTAMP WITH TIME ZONE NOT NULL DEFAULT CURRENT_TIMESTAMP,
correlation_id UUID NOT NULL,
revision INT NOT NULL DEFAULT 1,
CONSTRAINT opendart_imports_status_check CHECK (status IN ('SUCCESS', 'FAILURE', 'PARTIAL'))
);
CREATE INDEX IF NOT EXISTS idx_opendart_imports_import_at ON market_data.opendart_imports(import_at DESC);
CREATE INDEX IF NOT EXISTS idx_opendart_imports_status ON market_data.opendart_imports(status);
CREATE INDEX IF NOT EXISTS idx_opendart_imports_correlation_id ON market_data.opendart_imports(correlation_id);
CREATE INDEX IF NOT EXISTS idx_opendart_imports_published_at ON market_data.opendart_imports(published_at);
-- KIS API import log (trading orders, portfolio reconciliation)
CREATE TABLE IF NOT EXISTS market_data.kis_imports (
id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
import_at TIMESTAMP WITH TIME ZONE NOT NULL,
row_count INT NOT NULL,
checksum VARCHAR(256), -- SHA256 of imported data for deduplication
status VARCHAR(50) NOT NULL, -- 'SUCCESS', 'FAILURE', 'PARTIAL'
error_message TEXT,
details JSONB, -- Event-specific metadata (order_count, execution_latency_ms, token_refresh_required)
published_at TIMESTAMP WITH TIME ZONE NOT NULL DEFAULT CURRENT_TIMESTAMP,
correlation_id UUID NOT NULL,
revision INT NOT NULL DEFAULT 1,
CONSTRAINT kis_imports_status_check CHECK (status IN ('SUCCESS', 'FAILURE', 'PARTIAL'))
);
CREATE INDEX IF NOT EXISTS idx_kis_imports_import_at ON market_data.kis_imports(import_at DESC);
CREATE INDEX IF NOT EXISTS idx_kis_imports_status ON market_data.kis_imports(status);
CREATE INDEX IF NOT EXISTS idx_kis_imports_correlation_id ON market_data.kis_imports(correlation_id);
CREATE INDEX IF NOT EXISTS idx_kis_imports_published_at ON market_data.kis_imports(published_at);
-- ============================================================================
-- IMPORT ERROR CLASSIFICATION (for DQ quarantine & retry logic)
-- ============================================================================
-- Error classification for transient vs permanent failures
CREATE TABLE IF NOT EXISTS market_data.import_error_classification (
id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
import_id UUID NOT NULL, -- References one of krx/opendart/kis_imports
api_name VARCHAR(50) NOT NULL, -- 'krx', 'opendart', 'kis'
error_type VARCHAR(100) NOT NULL, -- e.g., 'TIMEOUT', 'RATE_LIMIT', 'INVALID_SCHEMA', 'AUTHENTICATION_FAILED'
classification VARCHAR(50) NOT NULL, -- 'TRANSIENT', 'PERMANENT', 'DATA_QUALITY'
retry_eligible BOOLEAN NOT NULL DEFAULT FALSE,
escalation_required BOOLEAN NOT NULL DEFAULT FALSE,
published_at TIMESTAMP WITH TIME ZONE NOT NULL DEFAULT CURRENT_TIMESTAMP,
correlation_id UUID NOT NULL
);
CREATE INDEX IF NOT EXISTS idx_import_error_classification_api ON market_data.import_error_classification(api_name);
CREATE INDEX IF NOT EXISTS idx_import_error_classification_error_type ON market_data.import_error_classification(error_type);
CREATE INDEX IF NOT EXISTS idx_import_error_classification_retry_eligible ON market_data.import_error_classification(retry_eligible);
-- ============================================================================
-- IMPORT SLA TRACKING (for compliance & monitoring)
-- ============================================================================
-- Daily SLA target: import should complete within 4 hours of market close (16:30 KST)
-- Target window: 16:30-20:30 KST
CREATE TABLE IF NOT EXISTS market_data.import_sla_tracking (
id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
api_name VARCHAR(50) NOT NULL, -- 'krx', 'opendart', 'kis'
import_date DATE NOT NULL,
scheduled_at TIMESTAMP WITH TIME ZONE NOT NULL,
started_at TIMESTAMP WITH TIME ZONE,
completed_at TIMESTAMP WITH TIME ZONE,
duration_seconds INT,
sla_met BOOLEAN, -- True if completed within 4 hours of market close
published_at TIMESTAMP WITH TIME ZONE NOT NULL DEFAULT CURRENT_TIMESTAMP,
correlation_id UUID NOT NULL,
UNIQUE(api_name, import_date)
);
CREATE INDEX IF NOT EXISTS idx_import_sla_tracking_api ON market_data.import_sla_tracking(api_name);
CREATE INDEX IF NOT EXISTS idx_import_sla_tracking_import_date ON market_data.import_sla_tracking(import_date);
CREATE INDEX IF NOT EXISTS idx_import_sla_tracking_sla_met ON market_data.import_sla_tracking(sla_met);
-- Last Known Good (LKG) cache for fallback
CREATE TABLE IF NOT EXISTS market_data.lkg_cache (
id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
api_name VARCHAR(50) NOT NULL, -- 'krx', 'opendart', 'kis'
cache_date DATE NOT NULL,
data_snapshot JSONB NOT NULL,
created_at TIMESTAMP WITH TIME ZONE NOT NULL DEFAULT CURRENT_TIMESTAMP,
published_at TIMESTAMP WITH TIME ZONE NOT NULL DEFAULT CURRENT_TIMESTAMP,
UNIQUE(api_name, cache_date)
);
CREATE INDEX IF NOT EXISTS idx_lkg_cache_api ON market_data.lkg_cache(api_name);
CREATE INDEX IF NOT EXISTS idx_lkg_cache_date ON market_data.lkg_cache(cache_date);
-- Permissions: schema owned by executing role
-- In production, add explicit GRANT via separate admin script after schema creation
+84
View File
@@ -0,0 +1,84 @@
-- Workstream I: VS-04 Audit Trail (Immutable events + GDPR compliance)
-- Creates compliance audit trail for model operations, regulatory reporting, and GDPR redaction
-- Audit events (immutable, INSERT-only)
CREATE TABLE IF NOT EXISTS compliance.audit_events (
id UUID PRIMARY KEY,
event_type VARCHAR(100) NOT NULL, -- MODEL_CREATED, APPROVAL_PROPOSED, APPROVAL_APPROVED, MODEL_ACTIVATED, SELL_DECISION_MADE, SELL_EXECUTED, BACKTEST_COMPLETED, DATA_CORRECTION, etc.
entity_type VARCHAR(50) NOT NULL, -- MODEL, APPROVAL, SELL_DECISION, TRADE_EXECUTION
entity_id UUID NOT NULL,
actor_email VARCHAR(255) NOT NULL,
actor_role VARCHAR(50), -- MAKER, CHECKER, SRE, SYSTEM
event_at TIMESTAMPTZ NOT NULL,
result VARCHAR(50) NOT NULL, -- SUCCESS, FAILURE, PARTIAL
error_message TEXT,
details JSONB, -- Event-specific metadata
evidence_links TEXT[], -- S3 artifact URLs (PBO scores, OOS returns, backtest reports)
ip_address INET, -- Source IP for forensics
user_agent TEXT, -- Client identifier
published_at TIMESTAMPTZ NOT NULL DEFAULT NOW(),
correlation_id UUID NOT NULL, -- Links related events
revision INT NOT NULL DEFAULT 1
);
-- Indexes for compliance querying
CREATE INDEX idx_audit_events_entity_id ON compliance.audit_events(entity_id);
CREATE INDEX idx_audit_events_event_type ON compliance.audit_events(event_type);
CREATE INDEX idx_audit_events_actor_email ON compliance.audit_events(actor_email);
CREATE INDEX idx_audit_events_event_at ON compliance.audit_events(event_at);
CREATE INDEX idx_audit_events_correlation_id ON compliance.audit_events(correlation_id);
-- GDPR retention tracking (personal data retention policy)
CREATE TABLE IF NOT EXISTS compliance.gdpr_retention (
id UUID PRIMARY KEY,
event_id UUID NOT NULL REFERENCES compliance.audit_events(id),
customer_id UUID, -- Links to personal data
data_categories VARCHAR(50)[], -- PII, EMAIL, TRADING_HISTORY, PORTFOLIO_DATA, etc.
retention_ends_at DATE, -- When to purge
purge_status VARCHAR(50) NOT NULL DEFAULT 'PENDING', -- PENDING, PURGED, EXCEPTION
purged_at TIMESTAMPTZ,
exception_reason TEXT,
published_at TIMESTAMPTZ NOT NULL DEFAULT NOW(),
revision INT NOT NULL DEFAULT 1
);
-- Indexes for GDPR processing
CREATE INDEX idx_gdpr_retention_customer_id ON compliance.gdpr_retention(customer_id);
CREATE INDEX idx_gdpr_retention_purge_status ON compliance.gdpr_retention(purge_status);
-- Event types enumeration (reference, not enforced at DB level)
CREATE TABLE IF NOT EXISTS compliance.audit_event_types (
event_type VARCHAR(100) PRIMARY KEY,
description TEXT,
entity_type VARCHAR(50), -- MODEL, APPROVAL, SELL_DECISION, TRADE_EXECUTION
created_at TIMESTAMPTZ NOT NULL DEFAULT NOW()
);
-- Seed event types
INSERT INTO compliance.audit_event_types (event_type, description, entity_type) VALUES
('MODEL_CREATED', 'New model version created', 'MODEL'),
('MODEL_ARCHIVED', 'Model retired from use', 'MODEL'),
('APPROVAL_PROPOSED', 'Maker submitted activation proposal', 'APPROVAL'),
('APPROVAL_APPROVED', 'Checker approved proposal', 'APPROVAL'),
('APPROVAL_REJECTED', 'Checker rejected proposal', 'APPROVAL'),
('MODEL_ACTIVATED', 'SRE activated model in production', 'MODEL'),
('MODEL_DEACTIVATED', 'SRE deactivated model', 'MODEL'),
('SELL_DECISION_MADE', 'Signal engine generated sell signal', 'SELL_DECISION'),
('SELL_EXECUTED', 'Trade executed based on signal', 'TRADE_EXECUTION'),
('BACKTEST_COMPLETED', 'Shadow run/backtest finished', 'MODEL'),
('DATA_CORRECTION', 'Source data corrected retroactively', 'MODEL'),
('COMPLIANCE_AUDIT', 'Auditor reviewed trail', 'MODEL')
ON CONFLICT (event_type) DO NOTHING;
-- Schema ownership
ALTER TABLE compliance.audit_events OWNER TO kartsell;
ALTER TABLE compliance.gdpr_retention OWNER TO kartsell;
ALTER TABLE compliance.audit_event_types OWNER TO kartsell;
-- Immutability constraints (enforced via code, not DB triggers)
-- INSERT-only: no UPDATE, no DELETE permitted on audit_events
-- Timestamps: immutable after insertion (enforced in application layer)
-- Correlation_id: immutable for traceability
-- 7-year retention policy (FSS requirement)
-- retention_ends_at defaults to now() + 7 years (enforced in application)