Files
seaweedfs/weed/iam/integration/oidc_provider_store.go
T
Chris Lu 4ded97a321 feat(iam): OIDC provider store + read-only IAM API (Phase 2a) (#9319)
* feat(iam): STS web-identity AWS-fidelity polish

- OIDC discovery via .well-known/openid-configuration; falls back to
  /.well-known/jwks.json when discovery is absent. Reject discovery docs
  whose issuer claim does not match the configured issuer to defend
  against issuer-substitution.
- ComputeParentUser derives a stable per-identity hash from (sub, iss).
  Surface as aws:userid in the request context and as a parent_user
  claim in the session JWT so per-user state survives token rotation.
- Per-role MaxSessionDuration (3600..43200) clamps requested
  DurationSeconds before the STS service applies its own caps.
- Tighten RoleSessionName to the AWS contract: 2..64 chars from
  [\w+=,.@-].
- Populate PackedPolicySize in AssumeRole / AssumeRoleWithWebIdentity /
  AssumeRoleWithLDAPIdentity responses as a percentage of the 2048-byte
  inline session policy budget.

* fix(iam): leave omitted DurationSeconds nil so STS default applies

capDurationByRole was substituting the role's MaxSessionDuration
when the caller omitted DurationSeconds entirely. AWS returns the
configured default (typically 1 hour) in that case, not the role's
upper bound — a 12h MaxSessionDuration shouldn't silently make every
no-duration assume-role mint a 12h session.

Return nil when requested is nil; let the downstream
calculateSessionDuration in the STS service apply its TokenDuration
default. The role-max upper bound still clamps when the request
arrives with a concrete value above the cap.

Addresses gemini high-priority review on PR #9318.

* fix(iam): synchronize OIDCProvider JWKS cache fields

jwksCache, jwksFetchedAt, resolvedJWKSUri, and discoveryFailed are
mutated lazily on the first token-validate call and refreshed
afterwards on TTL expiry. Multiple S3 requests can land here in
parallel, so the writes were racing against subsequent reads on
every other goroutine. resolvedJWKSUri/discoveryFailed inherited
the same un-protected pattern when discovery shipped.

Add sync.RWMutex; getPublicKey takes the read lock for the
common cache-hit path and promotes to the write lock for misses
+ refreshes. fetchJWKSLocked / resolveJWKSUriLocked assume the
write lock is held by the caller; fetchJWKS keeps the
test-friendly entry point that acquires the lock itself.

Addresses gemini high-priority review on PR #9318.

* fix(iam): trim trailing slash + retry discovery after transient failure

Two OIDC discovery edge cases reviewers flagged:

1. Issuer comparison was sensitive to trailing slashes. resolveJWKSUri
   trims them when building the discovery URL, but the doc.Issuer ↔
   p.config.Issuer check did not, so an IDP whose issuer claim drops or
   adds the slash relative to the configured value would be falsely
   rejected. Trim a single trailing slash on each side before comparing.

2. discoveryFailed flipped to true on any error and stayed there for the
   process lifetime. A transient 5xx at startup permanently locked the
   provider into the /.well-known/jwks.json fallback. Reset the flag at
   the top of fetchJWKSLocked when no URI has been cached yet, so each
   JWKS refresh (typically once per TTL = 1h) reattempts discovery.
   Successful discovery remains cached via resolvedJWKSUri so we don't
   pay the discovery RTT on every refresh.

Addresses gemini security-medium + medium reviews on PR #9318.

* fix(iam): require non-empty issuer in OIDC discovery doc

The previous "doc.Issuer != "" && ..." guard let a discovery document
that omitted the issuer field bypass the issuer-mismatch check
entirely, letting the doc steer fetchJWKS at any URL it provided.
OIDC Discovery 1.0 §3 mandates the issuer field; treat missing as a
hard failure same as mismatched. Trailing-slash equivalence still
applies.

Adds TestDiscoveryRejectsMissingIssuer alongside the existing
TestDiscoveryRejectsIssuerMismatch via a new omitDiscoveryIssuer
toggle on fakeIDP.

* feat(iam): OIDC provider store + read-only IAM API

Add OIDCProviderRecord — the persisted, IAM-managed view of an OIDC
identity provider — and an OIDCProviderStore interface with memory and
filer implementations mirroring the existing role-store pattern.

The store is hydrated at boot from the static STS.Providers list so the
new IAM API surfaces the same set the STS service already validates
against. Two read-only actions land now:

- ListOpenIDConnectProviders -> ARN-only list, AWS-shape XML.
- GetOpenIDConnectProvider   -> URL, ClientIDList, ThumbprintList,
                                Tags, CreateDate.

Mutations (Create/Delete/Add-Remove ClientID/Update Thumbprint), multiple
client_ids per provider, and TLS thumbprint pinning come in Phase 2b.

* fix(iam): preserve CreatedAt across boots + paginate ListProviders

Two medium-priority issues gemini flagged on the read-only IAM API:

1. The static-config bootstrap was setting CreatedAt = time.Now() on
   every server start, so the IAM GetOpenIDConnectProvider response's
   CreateDate shifted on each restart even when backed by a persistent
   store. Look up the existing record via GetProviderByARN first and
   preserve its CreatedAt; only the UpdatedAt advances.

2. FilerOIDCProviderStore.ListProviders had a hardcoded Limit: 1000
   that silently truncated above that. Stream-paginate via
   StartFromFileName, returning io.EOF naturally and surfacing all
   other errors instead of swallowing them.

Addresses two gemini medium reviews on PR #9319.
2026-05-04 22:15:03 -07:00

422 lines
14 KiB
Go

package integration
import (
"context"
"crypto/sha1"
"encoding/hex"
"encoding/json"
"errors"
"fmt"
"io"
"net/url"
"sort"
"strings"
"sync"
"time"
"github.com/seaweedfs/seaweedfs/weed/glog"
"github.com/seaweedfs/seaweedfs/weed/pb"
"github.com/seaweedfs/seaweedfs/weed/pb/filer_pb"
"google.golang.org/grpc"
)
// OIDCProviderRecord is the persisted, IAM-managed view of an OIDC identity
// provider. It is the source of truth consulted at AssumeRoleWithWebIdentity
// time. Static configuration entries are loaded into this same store at
// startup so the resolution path is uniform.
type OIDCProviderRecord struct {
// AccountID scopes the record. Empty means "global" — usable from any
// account. Non-empty records are only resolvable when the assuming
// caller's RoleArn lives in the same account. Phase 2a leaves this
// field unenforced; Phase 3c lights up the cross-account check.
AccountID string `json:"accountId,omitempty"`
// ARN is the canonical IAM identifier for the provider:
// arn:aws:iam::<account>:oidc-provider/<host>/<path>.
ARN string `json:"arn"`
// URL is the issuer URL (no trailing slash), e.g.
// https://token.actions.githubusercontent.com.
URL string `json:"url"`
// ClientIDs are the audiences AWS calls "client IDs". A token's `aud` or
// `azp` must match one of these for the token to be accepted.
ClientIDs []string `json:"clientIds,omitempty"`
// Thumbprints are SHA-1 hex digests of the IDP's TLS certificate, matching
// the AWS-compatible thumbprint algorithm. Used at JWKS-fetch time when
// non-empty; an empty list means "trust the system root store".
Thumbprints []string `json:"thumbprints,omitempty"`
// AllowedPrincipalTagKeys is the per-provider ABAC allowlist for the
// `https://aws.amazon.com/tags/principal_tags` claim namespace. Empty
// means no tags are surfaced from this provider.
AllowedPrincipalTagKeys []string `json:"allowedPrincipalTagKeys,omitempty"`
// PolicyClaim, when non-empty, enables claim-based policy mode for this
// provider: tokens whose RoleArn matches the configured sentinel will
// derive their effective policies from this JWT claim.
PolicyClaim string `json:"policyClaim,omitempty"`
// Tags is the AWS-style tag set attached to the IAM resource itself
// (audit/inventory metadata, not propagated into sessions).
Tags map[string]string `json:"tags,omitempty"`
CreatedAt time.Time `json:"createdAt"`
UpdatedAt time.Time `json:"updatedAt"`
}
// OIDCProviderStore stores OIDCProviderRecord entries. Implementations are
// expected to be safe for concurrent use.
type OIDCProviderStore interface {
StoreProvider(ctx context.Context, filerAddress string, record *OIDCProviderRecord) error
GetProviderByARN(ctx context.Context, filerAddress string, arn string) (*OIDCProviderRecord, error)
GetProviderByIssuer(ctx context.Context, filerAddress string, issuer string) (*OIDCProviderRecord, error)
ListProviders(ctx context.Context, filerAddress string) ([]*OIDCProviderRecord, error)
DeleteProvider(ctx context.Context, filerAddress string, arn string) error
}
// MemoryOIDCProviderStore is a process-local store, suitable for tests and
// single-node deployments. It also acts as the in-memory cache hydrated from
// static config at boot.
type MemoryOIDCProviderStore struct {
mu sync.RWMutex
providers map[string]*OIDCProviderRecord // keyed by ARN
}
// NewMemoryOIDCProviderStore creates an empty in-memory store.
func NewMemoryOIDCProviderStore() *MemoryOIDCProviderStore {
return &MemoryOIDCProviderStore{providers: make(map[string]*OIDCProviderRecord)}
}
// StoreProvider replaces any existing record with the same ARN.
func (m *MemoryOIDCProviderStore) StoreProvider(ctx context.Context, _ string, record *OIDCProviderRecord) error {
if record == nil {
return fmt.Errorf("record cannot be nil")
}
if record.ARN == "" {
return fmt.Errorf("record.ARN is required")
}
m.mu.Lock()
defer m.mu.Unlock()
m.providers[record.ARN] = copyOIDCProviderRecord(record)
return nil
}
// GetProviderByARN returns the record with the given ARN, or an error if absent.
func (m *MemoryOIDCProviderStore) GetProviderByARN(ctx context.Context, _ string, arn string) (*OIDCProviderRecord, error) {
m.mu.RLock()
defer m.mu.RUnlock()
rec, ok := m.providers[arn]
if !ok {
return nil, fmt.Errorf("OIDC provider not found: %s", arn)
}
return copyOIDCProviderRecord(rec), nil
}
// GetProviderByIssuer scans for the first record whose URL matches `issuer`.
// Comparison strips trailing slashes and is case-insensitive on host.
func (m *MemoryOIDCProviderStore) GetProviderByIssuer(ctx context.Context, _ string, issuer string) (*OIDCProviderRecord, error) {
want := normalizeIssuer(issuer)
m.mu.RLock()
defer m.mu.RUnlock()
for _, rec := range m.providers {
if normalizeIssuer(rec.URL) == want {
return copyOIDCProviderRecord(rec), nil
}
}
return nil, fmt.Errorf("no OIDC provider registered for issuer: %s", issuer)
}
// ListProviders returns every record sorted by ARN for stable output.
func (m *MemoryOIDCProviderStore) ListProviders(ctx context.Context, _ string) ([]*OIDCProviderRecord, error) {
m.mu.RLock()
defer m.mu.RUnlock()
out := make([]*OIDCProviderRecord, 0, len(m.providers))
for _, rec := range m.providers {
out = append(out, copyOIDCProviderRecord(rec))
}
sort.Slice(out, func(i, j int) bool { return out[i].ARN < out[j].ARN })
return out, nil
}
// DeleteProvider removes the record. Idempotent — deleting a missing ARN is success.
func (m *MemoryOIDCProviderStore) DeleteProvider(ctx context.Context, _ string, arn string) error {
m.mu.Lock()
defer m.mu.Unlock()
delete(m.providers, arn)
return nil
}
// FilerOIDCProviderStore persists records as JSON files in a filer directory,
// mirroring FilerRoleStore.
type FilerOIDCProviderStore struct {
grpcDialOption grpc.DialOption
basePath string
filerAddressProvider func() string
}
// NewFilerOIDCProviderStore returns a filer-backed store. Default path
// `/etc/iam/oidc-providers` aligns with the existing roles directory.
func NewFilerOIDCProviderStore(config map[string]interface{}, filerAddressProvider func() string) *FilerOIDCProviderStore {
store := &FilerOIDCProviderStore{
basePath: "/etc/iam/oidc-providers",
filerAddressProvider: filerAddressProvider,
}
if config != nil {
if bp, ok := config["basePath"].(string); ok && bp != "" {
store.basePath = strings.TrimSuffix(bp, "/")
}
}
glog.V(2).Infof("Initialized FilerOIDCProviderStore with basePath %s", store.basePath)
return store
}
func (f *FilerOIDCProviderStore) resolveFilerAddress(filerAddress string) string {
if filerAddress != "" {
return filerAddress
}
if f.filerAddressProvider != nil {
return f.filerAddressProvider()
}
return ""
}
func (f *FilerOIDCProviderStore) fileName(arn string) string {
// Hash the ARN to a fixed-width filename so we can store any character set
// (including the URL fragments AWS lets through) without filer-path issues.
h := sha1.Sum([]byte(arn))
return hex.EncodeToString(h[:]) + ".json"
}
// StoreProvider persists `record` as JSON.
func (f *FilerOIDCProviderStore) StoreProvider(ctx context.Context, filerAddress string, record *OIDCProviderRecord) error {
filerAddress = f.resolveFilerAddress(filerAddress)
if filerAddress == "" {
return fmt.Errorf("filer address is required")
}
if record == nil || record.ARN == "" {
return fmt.Errorf("record with ARN is required")
}
data, err := json.MarshalIndent(record, "", " ")
if err != nil {
return fmt.Errorf("marshal OIDC provider: %v", err)
}
return f.withFilerClient(filerAddress, func(client filer_pb.SeaweedFilerClient) error {
_, err := client.CreateEntry(ctx, &filer_pb.CreateEntryRequest{
Directory: f.basePath,
Entry: &filer_pb.Entry{
Name: f.fileName(record.ARN),
IsDirectory: false,
Attributes: &filer_pb.FuseAttributes{
Mtime: time.Now().Unix(),
Crtime: time.Now().Unix(),
FileMode: uint32(0o600),
},
Content: data,
},
})
if err != nil {
return fmt.Errorf("store OIDC provider %s: %v", record.ARN, err)
}
return nil
})
}
// GetProviderByARN looks up the record file by its ARN-derived filename.
func (f *FilerOIDCProviderStore) GetProviderByARN(ctx context.Context, filerAddress string, arn string) (*OIDCProviderRecord, error) {
filerAddress = f.resolveFilerAddress(filerAddress)
if filerAddress == "" {
return nil, fmt.Errorf("filer address is required")
}
var data []byte
err := f.withFilerClient(filerAddress, func(client filer_pb.SeaweedFilerClient) error {
resp, err := client.LookupDirectoryEntry(ctx, &filer_pb.LookupDirectoryEntryRequest{
Directory: f.basePath,
Name: f.fileName(arn),
})
if err != nil {
return fmt.Errorf("OIDC provider not found: %v", err)
}
if resp.Entry == nil {
return fmt.Errorf("OIDC provider not found: %s", arn)
}
data = resp.Entry.Content
return nil
})
if err != nil {
return nil, err
}
var rec OIDCProviderRecord
if err := json.Unmarshal(data, &rec); err != nil {
return nil, fmt.Errorf("unmarshal OIDC provider: %v", err)
}
return &rec, nil
}
// GetProviderByIssuer linearly scans the directory. Phase 2b adds an issuer
// index when write throughput grows; for read-only this is good enough.
func (f *FilerOIDCProviderStore) GetProviderByIssuer(ctx context.Context, filerAddress string, issuer string) (*OIDCProviderRecord, error) {
records, err := f.ListProviders(ctx, filerAddress)
if err != nil {
return nil, err
}
want := normalizeIssuer(issuer)
for _, rec := range records {
if normalizeIssuer(rec.URL) == want {
return rec, nil
}
}
return nil, fmt.Errorf("no OIDC provider registered for issuer: %s", issuer)
}
// ListProviders enumerates every record in the directory.
func (f *FilerOIDCProviderStore) ListProviders(ctx context.Context, filerAddress string) ([]*OIDCProviderRecord, error) {
filerAddress = f.resolveFilerAddress(filerAddress)
if filerAddress == "" {
return nil, fmt.Errorf("filer address is required")
}
var out []*OIDCProviderRecord
err := f.withFilerClient(filerAddress, func(client filer_pb.SeaweedFilerClient) error {
// Stream-paginate via StartFromFileName so deployments with more
// than 1000 providers don't get a silently truncated list.
const pageSize = 1000
startFrom := ""
for {
stream, err := client.ListEntries(ctx, &filer_pb.ListEntriesRequest{
Directory: f.basePath,
Limit: pageSize,
StartFromFileName: startFrom,
InclusiveStartFrom: false,
})
if err != nil {
return fmt.Errorf("list OIDC providers: %v", err)
}
lastName := ""
pageCount := 0
for {
resp, recvErr := stream.Recv()
if recvErr != nil {
if errors.Is(recvErr, io.EOF) {
break
}
return fmt.Errorf("recv OIDC provider entry: %w", recvErr)
}
if resp.Entry == nil || resp.Entry.IsDirectory {
continue
}
lastName = resp.Entry.Name
pageCount++
if !strings.HasSuffix(resp.Entry.Name, ".json") {
continue
}
var rec OIDCProviderRecord
if err := json.Unmarshal(resp.Entry.Content, &rec); err != nil {
glog.Warningf("skipping malformed OIDC provider record %s: %v", resp.Entry.Name, err)
continue
}
out = append(out, &rec)
}
if pageCount < pageSize {
break
}
startFrom = lastName
}
return nil
})
if err != nil {
return nil, err
}
sort.Slice(out, func(i, j int) bool { return out[i].ARN < out[j].ARN })
return out, nil
}
// DeleteProvider removes the record. Missing ARN is treated as success.
func (f *FilerOIDCProviderStore) DeleteProvider(ctx context.Context, filerAddress string, arn string) error {
filerAddress = f.resolveFilerAddress(filerAddress)
if filerAddress == "" {
return fmt.Errorf("filer address is required")
}
return f.withFilerClient(filerAddress, func(client filer_pb.SeaweedFilerClient) error {
resp, err := client.DeleteEntry(ctx, &filer_pb.DeleteEntryRequest{
Directory: f.basePath,
Name: f.fileName(arn),
IsDeleteData: true,
})
if err != nil {
if strings.Contains(err.Error(), "not found") {
return nil
}
return fmt.Errorf("delete OIDC provider %s: %v", arn, err)
}
if resp.Error != "" && !strings.Contains(resp.Error, "not found") {
return fmt.Errorf("delete OIDC provider %s: %s", arn, resp.Error)
}
return nil
})
}
func (f *FilerOIDCProviderStore) withFilerClient(filerAddress string, fn func(filer_pb.SeaweedFilerClient) error) error {
return pb.WithGrpcFilerClient(false, 0, pb.ServerAddress(filerAddress), f.grpcDialOption, fn)
}
// DeriveOIDCProviderARN turns an issuer URL into the canonical IAM ARN AWS uses.
// Empty `accountID` produces a global-style ARN.
//
// Examples:
//
// https://accounts.google.com -> arn:aws:iam::<acct>:oidc-provider/accounts.google.com
// https://oidc.eks.us-west-2.amazonaws.com/id/EXAMPLED ->
// arn:aws:iam::<acct>:oidc-provider/oidc.eks.us-west-2.amazonaws.com/id/EXAMPLED
func DeriveOIDCProviderARN(accountID, issuerURL string) (string, error) {
if issuerURL == "" {
return "", fmt.Errorf("issuer URL is required")
}
u, err := url.Parse(issuerURL)
if err != nil || u.Host == "" {
return "", fmt.Errorf("invalid issuer URL: %s", issuerURL)
}
host := strings.ToLower(u.Host)
resource := host + strings.TrimSuffix(u.Path, "/")
return fmt.Sprintf("arn:aws:iam::%s:oidc-provider/%s", accountID, resource), nil
}
// normalizeIssuer compares-friendly form: lowercased host, no trailing slash.
func normalizeIssuer(issuer string) string {
u, err := url.Parse(issuer)
if err != nil || u.Host == "" {
return strings.TrimSuffix(strings.ToLower(issuer), "/")
}
u.Host = strings.ToLower(u.Host)
u.Path = strings.TrimSuffix(u.Path, "/")
return u.String()
}
func copyOIDCProviderRecord(rec *OIDCProviderRecord) *OIDCProviderRecord {
if rec == nil {
return nil
}
cp := *rec
if rec.ClientIDs != nil {
cp.ClientIDs = append([]string(nil), rec.ClientIDs...)
}
if rec.Thumbprints != nil {
cp.Thumbprints = append([]string(nil), rec.Thumbprints...)
}
if rec.AllowedPrincipalTagKeys != nil {
cp.AllowedPrincipalTagKeys = append([]string(nil), rec.AllowedPrincipalTagKeys...)
}
if rec.Tags != nil {
cp.Tags = make(map[string]string, len(rec.Tags))
for k, v := range rec.Tags {
cp.Tags[k] = v
}
}
return &cp
}