Files
git.stella-ops.org/docs/modules/concelier/federation-operations.md

499 lines
14 KiB
Markdown

# Federation Setup and Operations
Per SPRINT_8200_0014_0003.
> **Related:** [Bundle Export Format](federation-bundle-export.md) for detailed bundle schema.
## Overview
Federation enables secure, cursor-based synchronization of canonical vulnerability advisories between StellaOps sites. It supports:
- **Delta exports**: Only changed records since the last cursor are included
- **Air-gap transfers**: Bundles can be written to files for offline transfer
- **Multi-site topology**: Multiple sites can synchronize independently
- **Cryptographic verification**: DSSE signatures ensure bundle authenticity
## Bundle Format
Federation bundles are ZST-compressed TAR archives containing:
| File | Description |
|------|-------------|
| `MANIFEST.json` | Bundle metadata, cursor, counts, hash |
| `canonicals.ndjson` | Canonical advisories (one per line) |
| `edges.ndjson` | Source edges linking advisories to sources |
| `deletions.ndjson` | Withdrawn/deleted advisory IDs |
| `SIGNATURE.json` | Optional DSSE signature envelope |
## Cursor Format
Cursors use ISO-8601 timestamp with sequence number:
```
{ISO-8601 timestamp}#{sequence number}
Examples:
2025-01-15T10:00:00.000Z#0001
2025-01-15T10:00:00.000Z#0002
```
- Cursors are site-specific (each site maintains independent cursors)
- Sequence numbers distinguish concurrent exports
- Cursors are monotonically increasing within a site
## Architecture
```
┌─────────────────────────────────────────────────────────────────────────────┐
│ Federation Topology │
├─────────────────────────────────────────────────────────────────────────────┤
│ │
│ ┌──────────────────┐ ┌──────────────────┐ │
│ │ Site A (HQ) │ │ Site B (Branch) │ │
│ │ │ │ │ │
│ │ ┌────────────┐ │ Export │ ┌────────────┐ │ │
│ │ │ Concelier │──┼──────────►│ │ Concelier │ │ │
│ │ │ │ │ Bundle │ │ │ │ │
│ │ └────────────┘ │ │ └────────────┘ │ │
│ │ │ │ │ │ │ │
│ │ ▼ │ │ ▼ │ │
│ │ ┌────────────┐ │ │ ┌────────────┐ │ │
│ │ │ PostgreSQL │ │ │ │ PostgreSQL │ │ │
│ │ └────────────┘ │ │ └────────────┘ │ │
│ └──────────────────┘ └──────────────────┘ │
│ │
│ ┌──────────────────┐ │
│ │ Site C (Air-Gap)│ │
│ │ │ │
│ │ ┌────────────┐ │ USB/Secure │
│ │ │ Concelier │◄─┼───Transfer │
│ │ │ │ │ │
│ │ └────────────┘ │ │
│ └──────────────────┘ │
│ │
└─────────────────────────────────────────────────────────────────────────────┘
```
## Setup
### 1. Enable Federation
Configure federation in `concelier.yaml`:
```yaml
Federation:
Enabled: true
SiteId: "site-us-west-1" # Unique identifier for this site
DefaultCompressionLevel: 3
DefaultMaxItems: 10000
RequireSignature: true
FederationImport:
AllowedSites:
- "site-us-east-1"
- "site-eu-central-1"
MaxBundleSizeBytes: 104857600 # 100 MB
SkipSignatureOnTrustedSites: false
```
### 2. Configure Site Policies
Create site policies for each trusted federation partner:
```bash
# Add trusted site
stella feedser sites add site-us-east-1 \
--display-name "US East Production" \
--enabled
# Configure policy
stella feedser sites policy site-us-east-1 \
--max-bundle-size 100MB \
--allowed-sources nvd,ghsa,debian
```
### 3. Generate Signing Keys
For signed bundles, configure Authority keys:
```bash
# Generate federation signing key
stella authority keys generate \
--name federation-signer \
--algorithm ES256 \
--purpose federation
# Export public key for distribution
stella authority keys export federation-signer --public
```
## Import Operations
### API Import
```
POST /api/v1/federation/import
Content-Type: application/zstd
```
**Query Parameters:**
| Parameter | Type | Default | Description |
|-----------|------|---------|-------------|
| `dry_run` | bool | false | Validate without importing |
| `skip_signature` | bool | false | Skip signature verification (requires trust) |
| `on_conflict` | enum | prefer_remote | `prefer_remote`, `prefer_local`, `fail` |
| `force` | bool | false | Import even if cursor is not after current |
**Response:**
```json
{
"success": true,
"bundle_hash": "sha256:a1b2c3...",
"imported_cursor": "2025-01-15T10:30:00.000Z#0042",
"counts": {
"canonical_created": 100,
"canonical_updated": 25,
"canonical_skipped": 10,
"edges_added": 200,
"deletions_processed": 5
},
"conflicts": [],
"duration_ms": 1234
}
```
### CLI Import
```bash
# Import from file
stella feedser bundle import ./bundle.zst
# Import with dry run
stella feedser bundle import ./bundle.zst --dry-run
# Import from stdin (for pipes)
cat bundle.zst | stella feedser bundle import -
# Import without signature verification (testing only)
stella feedser bundle import ./bundle.zst --skip-signature
# Force import (override cursor check)
stella feedser bundle import ./bundle.zst --force
```
### Conflict Resolution
When conflicts occur between local and remote values:
| Strategy | Behavior |
|----------|----------|
| `prefer_remote` | Remote value wins (default) |
| `prefer_local` | Local value preserved |
| `fail` | Abort import on first conflict |
Conflicts are logged with full details:
```json
{
"merge_hash": "sha256:abc...",
"field": "severity",
"local_value": "high",
"remote_value": "critical",
"resolution": "prefer_remote"
}
```
## Site Management
### List Sites
```bash
stella feedser sites list
```
Output:
```
SITE ID STATUS LAST SYNC CURSOR
───────────────────────── ──────── ─────────────────── ──────────────────────────
site-us-east-1 enabled 2025-01-15 10:30 2025-01-15T10:30:00Z#0042
site-eu-central-1 enabled 2025-01-15 09:15 2025-01-15T09:15:00Z#0038
site-asia-pacific-1 disabled never -
```
### View Site History
```bash
stella feedser sites history site-us-east-1 --limit 10
```
### Update Site Policy
```bash
stella feedser sites policy site-us-east-1 \
--enabled false # Disable imports from this site
```
## Air-Gap Operations
For sites without network connectivity:
### Export for Transfer
```bash
# On connected site
stella feedser bundle export \
-c "2025-01-14T00:00:00Z#0000" \
-o ./delta-2025-01-15.zst
# Transfer via USB/secure media
```
### Import on Air-Gap Site
```bash
# On air-gapped site
stella feedser bundle import ./delta-2025-01-15.zst
# Verify import
stella feedser sites list
```
### Full Sync Workflow
1. **Initial Sync:**
```bash
# Export full dataset
stella feedser bundle export -o ./full-sync.zst
```
2. **Transfer to air-gap site**
3. **Import on air-gap:**
```bash
stella feedser bundle import ./full-sync.zst
```
4. **Subsequent Delta Syncs:**
```bash
# Get current cursor from air-gap site
stella feedser sites list # Note the cursor
# On connected site, export delta
stella feedser bundle export -c "{cursor}" -o ./delta.zst
# Transfer and import on air-gap
```
## Verification
### Validate Bundle Without Import
```bash
stella feedser bundle validate ./bundle.zst
```
Output:
```
Bundle: bundle.zst
Version: feedser-bundle/1.0
Site: site-us-east-1
Cursor: 2025-01-15T10:30:00.000Z#0042
Counts:
Canonicals: 1,234
Edges: 3,456
Deletions: 12
Total: 4,702
Verification:
Hash: ✓ Valid
Signature: ✓ Valid (key: sha256:abc...)
Format: ✓ Valid
Ready for import.
```
### Preview Import Impact
```bash
stella feedser bundle import ./bundle.zst --dry-run --json
```
## Monitoring
### Sync Status Endpoint
```
GET /api/v1/federation/sync/status
```
Response:
```json
{
"sites": [
{
"site_id": "site-us-east-1",
"enabled": true,
"last_sync_at": "2025-01-15T10:30:00Z",
"last_cursor": "2025-01-15T10:30:00.000Z#0042",
"bundles_imported": 156,
"total_items_imported": 45678
}
],
"local_cursor": "2025-01-15T10:35:00.000Z#0044"
}
```
### Event Stream
Import events are published to the `canonical-imported` stream:
```json
{
"canonical_id": "uuid",
"cve": "CVE-2024-1234",
"affects_key": "pkg:npm/express@4.0.0",
"merge_hash": "sha256:...",
"action": "Created",
"bundle_hash": "sha256:...",
"site_id": "site-us-east-1",
"imported_at": "2025-01-15T10:30:15Z"
}
```
### Cache Invalidation
After import, cache indexes are automatically updated:
- PURL index updated for affected packages
- CVE index updated for vulnerability lookups
- Existing cache entries invalidated for refresh
## Troubleshooting
### Common Issues
| Issue | Cause | Solution |
|-------|-------|----------|
| "Cursor not after current" | Bundle is stale | Use `--force` or export newer bundle |
| "Signature verification failed" | Key mismatch | Verify signing key is trusted |
| "Site not allowed" | Policy restriction | Add site to `AllowedSites` config |
| "Bundle too large" | Size limit exceeded | Increase `MaxBundleSizeBytes` or export smaller delta |
### Debug Logging
Enable verbose logging for federation operations:
```yaml
Logging:
LogLevel:
StellaOps.Concelier.Federation: Debug
```
### Verify Sync State
```bash
# Check local vs remote cursor
stella feedser sites status site-us-east-1
# List recent imports
stella feedser sites history site-us-east-1 --limit 5
# Verify specific canonical was imported
stella feedser canonical get sha256:mergehash...
```
## Best Practices
1. **Regular Sync Schedule:** Configure automated delta exports/imports on a schedule (e.g., hourly)
2. **Monitor Cursor Drift:** Alert if cursor falls too far behind
3. **Verify Signatures:** Only disable signature verification in development
4. **Size Bundles Appropriately:** For large deltas, split into multiple bundles
5. **Test Import Before Production:** Use `--dry-run` to validate bundles
6. **Maintain Key Trust:** Regularly rotate and verify federation signing keys
7. **Document Site Policies:** Keep a registry of trusted sites and their policies
## Multi-Site Topologies
### Hub-and-Spoke Topology
```
┌─────────────┐
│ Hub Site │
│ (Primary) │
└──────┬──────┘
┌──────────┼──────────┐
▼ ▼ ▼
┌──────────┐ ┌──────────┐ ┌──────────┐
│ Site A │ │ Site B │ │ Site C │
│ (Spoke) │ │ (Spoke) │ │ (Spoke) │
└──────────┘ └──────────┘ └──────────┘
```
### Mesh Topology
Each site can import from multiple sources for redundancy:
```yaml
federation:
import:
allowed_sites:
- "hub-primary"
- "hub-secondary" # Redundancy
```
## Verification Details
### Hash Verification
Bundle hash is computed over compressed content:
```
SHA256(compressed bundle content)
```
### DSSE Signature Format
DSSE envelope contains:
```json
{
"payloadType": "application/stellaops.federation.bundle+json",
"payload": "base64(bundle_hash + site_id + cursor)",
"signatures": [
{
"keyId": "signing-key-001",
"algorithm": "ES256",
"signature": "base64(signature)"
}
]
}
```
## Monitoring Metrics
### Key Prometheus Metrics
- `federation_export_duration_seconds` - Export time
- `federation_import_duration_seconds` - Import time
- `federation_bundle_size_bytes` - Bundle sizes
- `federation_items_processed_total` - Items processed by type
- `federation_conflicts_total` - Merge conflicts encountered
## Security Considerations
1. **Never skip signature verification in production**
2. **Validate allowed_sites whitelist**
3. **Use TLS for API endpoints**
4. **Rotate signing keys periodically**
5. **Audit import events**
6. **Monitor for duplicate bundle imports**