Files
seaweedfs/test/s3/iam/STS_DISTRIBUTED.md
T
Chris Lu dc5621d2ae s3: use oidc: prefix for trust-policy conditions in IAM example (#9653)
* s3: use oidc: prefix for trust-policy conditions in IAM example

Trust-policy conditions for AssumeRoleWithWebIdentity see OIDC claims
under the oidc: prefix, so the docker example's bare "roles" key never
matched and denied every web-identity assume against those roles. Switch
the three roles to oidc:roles.

Also document the available trust-policy condition keys (oidc:iss/sub/aud,
oidc:<claim>, aws:FederatedProvider, aws:userid, sts:DurationSeconds) and
note that roleMapping selects the role for direct OIDC bearer auth while
STS uses the explicit RoleArn plus trust policy.

* s3: clarify aws:userid differs between trust policy and request auth

aws:userid is the raw sub claim during trust-policy evaluation, but a
stable sub+iss hash (ComputeParentUser) during S3 request authorization
after the role is assumed. Note both so the two contexts aren't conflated.
2026-05-23 20:02:48 -07:00

560 lines
17 KiB
Markdown

# Distributed STS Service for SeaweedFS S3 Gateway
This document explains how to configure and deploy the STS (Security Token Service) for distributed SeaweedFS S3 Gateway deployments with consistent identity provider configurations.
## Problem Solved
Previously, identity providers had to be **manually registered** on each S3 gateway instance, leading to:
-**Inconsistent authentication**: Different instances might have different providers
-**Manual synchronization**: No guarantee all instances have same provider configs
-**Authentication failures**: Users getting different responses from different instances
-**Operational complexity**: Difficult to manage provider configurations at scale
## Solution: Configuration-Driven Providers
The STS service now supports **automatic provider loading** from configuration files, ensuring:
-**Consistent providers**: All instances load identical providers from config
-**Automatic synchronization**: Configuration-driven, no manual registration needed
-**Reliable authentication**: Same behavior from all instances
-**Easy management**: Update config file, restart services
## Configuration Schema
### Basic STS Configuration
```json
{
"sts": {
"tokenDuration": "1h",
"maxSessionLength": "12h",
"issuer": "seaweedfs-sts",
"signingKey": "base64-encoded-signing-key-32-chars-min"
}
}
```
**Note**: The STS service uses a **stateless JWT design** where all session information is embedded directly in the JWT token. No external session storage is required.
### Configuration-Driven Providers
```json
{
"sts": {
"tokenDuration": "1h",
"maxSessionLength": "12h",
"issuer": "seaweedfs-sts",
"signingKey": "base64-encoded-signing-key",
"providers": [
{
"name": "keycloak-oidc",
"type": "oidc",
"enabled": true,
"config": {
"issuer": "https://keycloak.company.com/realms/seaweedfs",
"clientId": "seaweedfs-s3",
"clientSecret": "super-secret-key",
"jwksUri": "https://keycloak.company.com/realms/seaweedfs/protocol/openid-connect/certs",
"scopes": ["openid", "profile", "email", "roles"],
"claimsMapping": {
"usernameClaim": "preferred_username",
"groupsClaim": "roles"
}
}
},
{
"name": "backup-oidc",
"type": "oidc",
"enabled": false,
"config": {
"issuer": "https://backup-oidc.company.com",
"clientId": "seaweedfs-backup"
}
},
{
"name": "dev-mock-provider",
"type": "mock",
"enabled": true,
"config": {
"issuer": "http://localhost:9999",
"clientId": "mock-client"
}
}
]
}
}
```
## Supported Provider Types
### 1. OIDC Provider (`"type": "oidc"`)
For production authentication with OpenID Connect providers like Keycloak, Auth0, Google, etc.
**Required Configuration:**
- `issuer`: OIDC issuer URL
- `clientId`: OAuth2 client ID
**Optional Configuration:**
- `clientSecret`: OAuth2 client secret (for confidential clients)
- `jwksUri`: JSON Web Key Set URI (auto-discovered if not provided)
- `userInfoUri`: UserInfo endpoint URI (auto-discovered if not provided)
- `scopes`: OAuth2 scopes to request (default: `["openid"]`)
- `claimsMapping`: Map OIDC claims to identity attributes
**Example:**
```json
{
"name": "corporate-keycloak",
"type": "oidc",
"enabled": true,
"config": {
"issuer": "https://sso.company.com/realms/production",
"clientId": "seaweedfs-prod",
"clientSecret": "confidential-secret",
"scopes": ["openid", "profile", "email", "groups"],
"claimsMapping": {
"usernameClaim": "preferred_username",
"groupsClaim": "groups",
"emailClaim": "email"
}
}
}
```
### 2. Mock Provider (`"type": "mock"`)
For development, testing, and staging environments.
**Configuration:**
- `issuer`: Mock issuer URL (default: `http://localhost:9999`)
- `clientId`: Mock client ID
**Example:**
```json
{
"name": "dev-mock",
"type": "mock",
"enabled": true,
"config": {
"issuer": "http://dev-mock:9999",
"clientId": "dev-client"
}
}
```
**Built-in Test Tokens:**
- `valid_test_token`: Returns test user with developer groups
- `valid-oidc-token`: Compatible with integration tests
- `expired_token`: Returns token expired error
- `invalid_token`: Returns invalid token error
### 3. Future Provider Types
The factory pattern supports easy addition of new provider types:
- `"type": "ldap"`: LDAP/Active Directory authentication
- `"type": "saml"`: SAML 2.0 authentication
- `"type": "oauth2"`: Generic OAuth2 providers
- `"type": "custom"`: Custom authentication backends
## Deployment Patterns
### Single Instance (Development)
```bash
# Standard deployment with config-driven providers
weed s3 -filer=localhost:8888 -port=8333 -iam.config=/path/to/sts_config.json
```
### Multiple Instances (Production)
```bash
# Instance 1
weed s3 -filer=prod-filer:8888 -port=8333 -iam.config=/shared/sts_distributed.json
# Instance 2
weed s3 -filer=prod-filer:8888 -port=8334 -iam.config=/shared/sts_distributed.json
# Instance N
weed s3 -filer=prod-filer:8888 -port=833N -iam.config=/shared/sts_distributed.json
```
**Critical Requirements for Distributed Deployment:**
1. **Identical Configuration Files**: All instances must use the exact same configuration file
2. **Same Signing Keys**: All instances must have identical `signingKey` values
3. **Same Issuer**: All instances must use the same `issuer` value
**Note**: STS now uses stateless JWT tokens, eliminating the need for shared session storage.
### High Availability Setup
```yaml
# docker-compose.yml for production deployment
services:
filer:
image: seaweedfs/seaweedfs:latest
command: "filer -master=master:9333"
volumes:
- filer-data:/data
s3-gateway-1:
image: seaweedfs/seaweedfs:latest
command: "s3 -filer=filer:8888 -port=8333 -iam.config=/config/sts_distributed.json"
ports:
- "8333:8333"
volumes:
- ./sts_distributed.json:/config/sts_distributed.json:ro
depends_on: [filer]
s3-gateway-2:
image: seaweedfs/seaweedfs:latest
command: "s3 -filer=filer:8888 -port=8333 -iam.config=/config/sts_distributed.json"
ports:
- "8334:8333"
volumes:
- ./sts_distributed.json:/config/sts_distributed.json:ro
depends_on: [filer]
s3-gateway-3:
image: seaweedfs/seaweedfs:latest
command: "s3 -filer=filer:8888 -port=8333 -iam.config=/config/sts_distributed.json"
ports:
- "8335:8333"
volumes:
- ./sts_distributed.json:/config/sts_distributed.json:ro
depends_on: [filer]
load-balancer:
image: nginx:alpine
ports:
- "80:80"
volumes:
- ./nginx.conf:/etc/nginx/nginx.conf:ro
depends_on: [s3-gateway-1, s3-gateway-2, s3-gateway-3]
```
## Authentication Flow
### 1. OIDC Authentication Flow
```
1. User authenticates with OIDC provider (Keycloak, Auth0, etc.)
2. User receives OIDC JWT token from provider
3. User calls SeaweedFS STS AssumeRoleWithWebIdentity
POST /sts/assume-role-with-web-identity
{
"RoleArn": "arn:aws:iam::role/S3AdminRole",
"WebIdentityToken": "eyJ0eXAiOiJKV1QiLCJhbGc...",
"RoleSessionName": "user-session"
}
4. STS validates OIDC token with configured provider
- Verifies JWT signature using provider's JWKS
- Validates issuer, audience, expiration
- Extracts user identity and groups
5. STS checks role trust policy
- Verifies user/groups can assume the requested role
- Validates conditions in trust policy
6. STS generates temporary credentials
- Creates temporary access key, secret key, session token
- Session token is signed JWT with all session information embedded (stateless)
7. User receives temporary credentials
{
"Credentials": {
"AccessKeyId": "AKIA...",
"SecretAccessKey": "base64-secret",
"SessionToken": "eyJ0eXAiOiJKV1QiLCJhbGc...",
"Expiration": "2024-01-01T12:00:00Z"
}
}
8. User makes S3 requests with temporary credentials
- AWS SDK signs requests with temporary credentials
- SeaweedFS S3 gateway validates session token
- Gateway checks permissions via policy engine
```
### 2. Cross-Instance Token Validation
```
User Request → Load Balancer → Any S3 Gateway Instance
Extract JWT Session Token
Validate JWT Token
(Self-contained - no external storage needed)
Check Permissions
(Shared policy engine)
Allow/Deny Request
```
## Trust Policy Conditions
Step 5 above evaluates the role's trust policy against context keys derived from the
OIDC token's claims. The available keys are:
| Condition key | Source |
|---------------|--------|
| `oidc:iss` | `iss` claim (issuer URL) |
| `oidc:sub` | `sub` claim |
| `oidc:aud` | `aud` claim |
| `oidc:<claim>` | any other token claim, e.g. `oidc:roles`, `oidc:groups`, `oidc:email` |
| `aws:FederatedProvider` | the provider `name` (e.g. `keycloak-oidc`) when its configured issuer matches the token, otherwise the raw issuer URL |
| `aws:userid` | `sub` claim (same value as `oidc:sub` during trust-policy evaluation) |
| `sts:DurationSeconds` | requested session duration, when supplied |
During trust-policy evaluation `aws:userid` is the raw `sub` claim. Once the
role has been assumed, the keys seen by request authorization differ: there
`aws:userid` is a stable per-identity hash of `sub` and `iss` (see
`ComputeParentUser`), so do not assume the two contexts carry the same value.
Custom claims are always exposed under the `oidc:` prefix, so a trust policy must use
`oidc:roles` (not a bare `roles`) to match a `roles` claim:
```json
"Condition": {
"StringEquals": {
"oidc:roles": "s3-admin"
}
}
```
A multi-valued claim (such as a `roles` array) matches when any of its values equals
the condition value. The same `oidc:` keys can be interpolated into policy resources,
e.g. `arn:aws:s3:::bucket/${oidc:sub}/*`.
### roleMapping vs. trust policy
A provider's `roleMapping` and a role's trust policy apply to two different entry
points and are not interchangeable:
- **Direct OIDC** — an S3 request carrying `Authorization: Bearer <OIDC-JWT>`. The
gateway applies `roleMapping` to choose the caller's role from the token claims; the
first matching rule (or `defaultRole`) wins.
- **STS `AssumeRoleWithWebIdentity`** — the caller names the role explicitly via
`RoleArn`, and that role's trust policy decides whether the assumption is allowed.
`roleMapping` does not select the role on this path; instead the token claims are
surfaced as the `oidc:` condition keys above for the trust policy to evaluate.
## Configuration Management
### Development Environment
```json
{
"sts": {
"tokenDuration": "1h",
"maxSessionLength": "12h",
"issuer": "seaweedfs-dev-sts",
"signingKey": "ZGV2LXNpZ25pbmcta2V5LTMyLWNoYXJhY3RlcnMtbG9uZw==",
"providers": [
{
"name": "dev-mock",
"type": "mock",
"enabled": true,
"config": {
"issuer": "http://localhost:9999",
"clientId": "dev-mock-client"
}
}
]
}
}
```
### Production Environment
```json
{
"sts": {
"tokenDuration": "1h",
"maxSessionLength": "12h",
"issuer": "seaweedfs-prod-sts",
"signingKey": "cHJvZC1zaWduaW5nLWtleS0zMi1jaGFyYWN0ZXJzLWxvbmctcmFuZG9t",
"providers": [
{
"name": "corporate-sso",
"type": "oidc",
"enabled": true,
"config": {
"issuer": "https://sso.company.com/realms/production",
"clientId": "seaweedfs-prod",
"clientSecret": "${SSO_CLIENT_SECRET}",
"scopes": ["openid", "profile", "email", "groups"],
"claimsMapping": {
"usernameClaim": "preferred_username",
"groupsClaim": "groups"
}
}
},
{
"name": "backup-auth",
"type": "oidc",
"enabled": false,
"config": {
"issuer": "https://backup-sso.company.com",
"clientId": "seaweedfs-backup"
}
}
]
}
}
```
## Operational Best Practices
### 1. Configuration Management
- **Version Control**: Store configurations in Git with proper versioning
- **Environment Separation**: Use separate configs for dev/staging/production
- **Secret Management**: Use environment variable substitution for secrets
- **Configuration Validation**: Test configurations before deployment
### 2. Security Considerations
- **Signing Key Security**: Use strong, randomly generated signing keys (32+ bytes)
- **Key Rotation**: Implement signing key rotation procedures
- **Secret Storage**: Store client secrets in secure secret management systems
- **TLS Encryption**: Always use HTTPS for OIDC providers in production
### 3. Monitoring and Troubleshooting
- **Provider Health**: Monitor OIDC provider availability and response times
- **Session Metrics**: Track active sessions, token validation errors
- **Configuration Drift**: Alert on configuration inconsistencies between instances
- **Authentication Logs**: Log authentication attempts for security auditing
### 4. Capacity Planning
- **Provider Performance**: Monitor OIDC provider response times and rate limits
- **Token Validation**: Monitor JWT validation performance and caching
- **Memory Usage**: Monitor JWT token validation caching and provider metadata
## Migration Guide
### From Manual Provider Registration
**Before (Manual Registration):**
```go
// Each instance needs this code
keycloakProvider := oidc.NewOIDCProvider("keycloak-oidc")
keycloakProvider.Initialize(keycloakConfig)
stsService.RegisterProvider(keycloakProvider)
```
**After (Configuration-Driven):**
```json
{
"sts": {
"providers": [
{
"name": "keycloak-oidc",
"type": "oidc",
"enabled": true,
"config": {
"issuer": "https://keycloak.company.com/realms/seaweedfs",
"clientId": "seaweedfs-s3"
}
}
]
}
}
```
### Migration Steps
1. **Create Configuration File**: Convert manual provider registrations to JSON config
2. **Test Single Instance**: Deploy config to one instance and verify functionality
3. **Validate Consistency**: Ensure all instances load identical providers
4. **Rolling Deployment**: Update instances one by one with new configuration
5. **Remove Manual Code**: Clean up manual provider registration code
## Troubleshooting
### Common Issues
#### 1. Provider Inconsistency
**Symptoms**: Authentication works on some instances but not others
**Diagnosis**:
```bash
# Check provider counts on each instance
curl http://instance1:8333/sts/providers | jq '.providers | length'
curl http://instance2:8334/sts/providers | jq '.providers | length'
```
**Solution**: Ensure all instances use identical configuration files
#### 2. Token Validation Failures
**Symptoms**: "Invalid signature" or "Invalid issuer" errors
**Diagnosis**: Check signing key and issuer consistency
**Solution**: Verify `signingKey` and `issuer` are identical across all instances
#### 3. Provider Loading Failures
**Symptoms**: Providers not loaded at startup
**Diagnosis**: Check logs for provider initialization errors
**Solution**: Validate provider configuration against schema
#### 4. OIDC Provider Connectivity
**Symptoms**: "Failed to fetch JWKS" errors
**Diagnosis**: Test OIDC provider connectivity from all instances
**Solution**: Check network connectivity, DNS resolution, certificates
### Debug Commands
```bash
# Test configuration loading
weed s3 -iam.config=/path/to/config.json -test.config
# Validate JWT tokens
curl -X POST http://localhost:8333/sts/validate-token \
-H "Content-Type: application/json" \
-d '{"sessionToken": "eyJ0eXAiOiJKV1QiLCJhbGc..."}'
# List loaded providers
curl http://localhost:8333/sts/providers
# Check session store
curl http://localhost:8333/sts/sessions/count
```
## Performance Considerations
### Token Validation Performance
- **JWT Validation**: ~1-5ms per token validation
- **JWKS Caching**: Cache JWKS responses to reduce OIDC provider load
- **Session Lookup**: Filer session lookup adds ~10-20ms latency
- **Concurrent Requests**: Each instance can handle 1000+ concurrent validations
### Scaling Recommendations
- **Horizontal Scaling**: Add more S3 gateway instances behind load balancer
- **Session Store Optimization**: Use SSD storage for filer session store
- **Provider Caching**: Implement JWKS caching to reduce provider load
- **Connection Pooling**: Use connection pooling for filer communication
## Summary
The configuration-driven provider system solves critical distributed deployment issues:
-**Automatic Provider Loading**: No manual registration code required
-**Configuration Consistency**: All instances load identical providers from config
-**Easy Management**: Update config file, restart services
-**Production Ready**: Supports OIDC, proper session management, distributed storage
-**Backwards Compatible**: Existing manual registration still works
This enables SeaweedFS S3 Gateway to **scale horizontally** with **consistent authentication** across all instances, making it truly **production-ready for enterprise deployments**.