Files
seaweedfs/weed/credential/propagating_store.go
T
Chris Lu 902a12fd6f wdclient: bound the wait for a master leader by the caller's context (#11002)
* wdclient: bound the wait for a master leader by the caller's context

WithClient waited on GetMaster with context.Background(), so a caller that
arrived while no master leader was known parked in a 200ms poll loop until one
appeared, whatever deadline it had already set on the RPC. Each retry above it
then left another goroutine in the same wait.

Take the context in WithClient and WithClientCustomGetMaster and hand it to
GetMaster, and stop the retry loop once it is done. The dial keeps
context.Background(): fn brings its own RPC context, so a cancellation seen
here cannot be attributed to the shared connection.

Call sites pass whatever they hold: the request context in the filer's
CollectionList, DeleteCollection and Statistics handlers and in the credential
store's propagation, the operation context in the shell's s3.bucket.delete and
the kafka gateway's broker and filer discovery, and context.Background() where
there is none - the shell commands, the admin dashboard wrapper, and the
exclusive locker's initial lease. The locker's release keeps its own
uncancelled context so a slow unlock cannot turn into a ghost lock.

Claude-Session: https://claude.ai/code/session_01BjDWtZsCoZY6x4pdDmGWxU

* wdclient: test that WithClient gives up with the caller's context

Claude-Session: https://claude.ai/code/session_01BjDWtZsCoZY6x4pdDmGWxU

* wdclient: cut the master retry backoff short when the caller gives up

util.Retry sleeps unconditionally between attempts, so a transient error
arriving just before the caller's deadline still cost it a full backoff step.
Use the context-aware util.RetryWithBackoff, the same helper the volume lookup
in this file already uses.

Two call sites went with it: the shell's lock-holder lookup builds its three
second bound before WithClient so it also covers finding the leader, as its
comment already promised, and the filer's post-delete collection cleanup goes
back to an uncancelled context - the entry is already gone, so a caller that
hung up must not leave the collection behind.

Claude-Session: https://claude.ai/code/session_01BjDWtZsCoZY6x4pdDmGWxU

* wdclient: test that a cancel during backoff ends the retry

Claude-Session: https://claude.ai/code/session_01BjDWtZsCoZY6x4pdDmGWxU
2026-08-27 22:27:45 -07:00

574 lines
20 KiB
Go

package credential
import (
"context"
"encoding/json"
"sync"
"time"
"github.com/seaweedfs/seaweedfs/weed/cluster"
"github.com/seaweedfs/seaweedfs/weed/glog"
"github.com/seaweedfs/seaweedfs/weed/pb"
"github.com/seaweedfs/seaweedfs/weed/pb/iam_pb"
"github.com/seaweedfs/seaweedfs/weed/pb/master_pb"
"github.com/seaweedfs/seaweedfs/weed/pb/s3_pb"
"github.com/seaweedfs/seaweedfs/weed/s3api/policy_engine"
"github.com/seaweedfs/seaweedfs/weed/wdclient"
"google.golang.org/grpc"
)
var _ CredentialStore = &PropagatingCredentialStore{}
var _ PolicyManager = &PropagatingCredentialStore{}
type propagatingManagedPolicyLoader interface {
LoadManagedPolicies(ctx context.Context) ([]*iam_pb.Policy, error)
}
type propagatingInlinePolicyLoader interface {
LoadInlinePolicies(ctx context.Context) (map[string]map[string]policy_engine.PolicyDocument, error)
}
type PropagatingCredentialStore struct {
CredentialStore
masterClient *wdclient.MasterClient
grpcDialOption grpc.DialOption
}
func NewPropagatingCredentialStore(upstream CredentialStore, masterClient *wdclient.MasterClient, grpcDialOption grpc.DialOption) *PropagatingCredentialStore {
return &PropagatingCredentialStore{
CredentialStore: upstream,
masterClient: masterClient,
grpcDialOption: grpcDialOption,
}
}
func (s *PropagatingCredentialStore) SetFilerAddressFunc(getFiler func() pb.ServerAddress, grpcDialOption grpc.DialOption) {
if setter, ok := s.CredentialStore.(FilerAddressSetter); ok {
setter.SetFilerAddressFunc(getFiler, grpcDialOption)
}
}
func (s *PropagatingCredentialStore) propagateChange(ctx context.Context, fn func(context.Context, s3_pb.SeaweedS3IamCacheClient) error) {
if s.masterClient == nil {
return
}
// List S3 servers
var s3Servers []string
err := s.masterClient.WithClient(ctx, false, func(client master_pb.SeaweedClient) error {
glog.V(4).Infof("IAM: listing S3 servers (FilerGroup: '%s')", s.masterClient.FilerGroup)
resp, err := client.ListClusterNodes(ctx, &master_pb.ListClusterNodesRequest{
ClientType: cluster.S3Type,
FilerGroup: s.masterClient.FilerGroup,
})
if err != nil {
glog.Warningf("failed to list S3 servers: %v", err)
return err
}
for _, node := range resp.ClusterNodes {
s3Servers = append(s3Servers, node.Address)
}
return nil
})
if err != nil {
glog.Warningf("failed to list s3 servers via master client: %v", err)
return
}
glog.V(1).Infof("IAM: propagating change to %d S3 servers: %v", len(s3Servers), s3Servers)
// Create context with timeout for the propagation process
propagateCtx, cancel := context.WithTimeout(ctx, 10*time.Second)
defer cancel()
var wg sync.WaitGroup
for _, server := range s3Servers {
wg.Add(1)
go func(server string) {
defer wg.Done()
err := pb.WithGrpcClient(context.Background(), false, 0, func(conn *grpc.ClientConn) error {
glog.V(4).Infof("IAM: successfully connected to S3 server %s for propagation", server)
client := s3_pb.NewSeaweedS3IamCacheClient(conn)
return fn(propagateCtx, client)
}, server, false, s.grpcDialOption)
if err != nil {
glog.Warningf("failed to propagate change to s3 server %s: %v", server, err)
}
}(server)
}
wg.Wait()
}
func (s *PropagatingCredentialStore) AttachUserPolicy(ctx context.Context, username string, policyName string) error {
glog.V(4).Infof("IAM: PropagatingCredentialStore.AttachUserPolicy %s -> %s", username, policyName)
if err := s.CredentialStore.AttachUserPolicy(ctx, username, policyName); err != nil {
return err
}
// Fetch updated identity to propagate
identity, err := s.CredentialStore.GetUser(ctx, username)
if err != nil {
glog.Warningf("failed to get user %s after attaching policy: %v", username, err)
return nil
}
s.propagateChange(ctx, func(tx context.Context, client s3_pb.SeaweedS3IamCacheClient) error {
_, err := client.PutIdentity(tx, &iam_pb.PutIdentityRequest{Identity: identity})
return err
})
return nil
}
func (s *PropagatingCredentialStore) DetachUserPolicy(ctx context.Context, username string, policyName string) error {
glog.V(4).Infof("IAM: PropagatingCredentialStore.DetachUserPolicy %s -> %s", username, policyName)
if err := s.CredentialStore.DetachUserPolicy(ctx, username, policyName); err != nil {
return err
}
// Fetch updated identity to propagate
identity, err := s.CredentialStore.GetUser(ctx, username)
if err != nil {
glog.Warningf("failed to get user %s after detaching policy: %v", username, err)
return nil
}
s.propagateChange(ctx, func(tx context.Context, client s3_pb.SeaweedS3IamCacheClient) error {
_, err := client.PutIdentity(tx, &iam_pb.PutIdentityRequest{Identity: identity})
return err
})
return nil
}
func (s *PropagatingCredentialStore) ListAttachedUserPolicies(ctx context.Context, username string) ([]string, error) {
return s.CredentialStore.ListAttachedUserPolicies(ctx, username)
}
// SaveConfiguration overrides the embedded CredentialStore.SaveConfiguration
// so bulk identity / group updates also push to running S3 caches. The IAM
// API flow ends each handler with SaveConfiguration after recomputing
// identity.Actions (the legacy authorization field), so reusing PutIdentity
// here is what keeps inline-policy changes visible to S3 servers without
// requiring a restart. We diff against the prior store state so deletions
// also fan out (RemoveIdentity / RemoveGroup); without the diff a postgres
// user who got pruned by SaveConfiguration would linger in the S3 cache.
func (s *PropagatingCredentialStore) SaveConfiguration(ctx context.Context, config *iam_pb.S3ApiConfiguration) error {
priorUsers, priorErr := s.CredentialStore.ListUsers(ctx)
if priorErr != nil {
glog.V(1).Infof("failed to list users before SaveConfiguration; skipping deletion propagation: %v", priorErr)
priorUsers = nil
}
priorGroups, gPriorErr := s.CredentialStore.ListGroups(ctx)
if gPriorErr != nil {
glog.V(1).Infof("failed to list groups before SaveConfiguration; skipping deletion propagation: %v", gPriorErr)
priorGroups = nil
}
if err := s.CredentialStore.SaveConfiguration(ctx, config); err != nil {
return err
}
keptUsers := make(map[string]struct{}, len(config.Identities))
for _, ident := range config.Identities {
if ident != nil {
keptUsers[ident.Name] = struct{}{}
}
}
keptGroups := make(map[string]struct{}, len(config.Groups))
for _, g := range config.Groups {
if g != nil {
keptGroups[g.Name] = struct{}{}
}
}
s.propagateChange(ctx, func(tx context.Context, client s3_pb.SeaweedS3IamCacheClient) error {
for _, ident := range config.Identities {
if ident == nil {
continue
}
if _, err := client.PutIdentity(tx, &iam_pb.PutIdentityRequest{Identity: ident}); err != nil {
return err
}
}
for _, g := range config.Groups {
if g == nil {
continue
}
if _, err := client.PutGroup(tx, &iam_pb.PutGroupRequest{Group: g}); err != nil {
return err
}
}
for _, name := range priorUsers {
if _, kept := keptUsers[name]; kept {
continue
}
if _, err := client.RemoveIdentity(tx, &iam_pb.RemoveIdentityRequest{Username: name}); err != nil {
return err
}
}
for _, name := range priorGroups {
if _, kept := keptGroups[name]; kept {
continue
}
if _, err := client.RemoveGroup(tx, &iam_pb.RemoveGroupRequest{GroupName: name}); err != nil {
return err
}
}
return nil
})
return nil
}
func (s *PropagatingCredentialStore) CreateUser(ctx context.Context, identity *iam_pb.Identity) error {
glog.V(4).Infof("IAM: PropagatingCredentialStore.CreateUser %s", identity.Name)
if err := s.CredentialStore.CreateUser(ctx, identity); err != nil {
return err
}
s.propagateChange(ctx, func(tx context.Context, client s3_pb.SeaweedS3IamCacheClient) error {
_, err := client.PutIdentity(tx, &iam_pb.PutIdentityRequest{Identity: identity})
return err
})
return nil
}
func (s *PropagatingCredentialStore) UpdateUser(ctx context.Context, username string, identity *iam_pb.Identity) error {
glog.V(4).Infof("IAM: PropagatingCredentialStore.UpdateUser %s", username)
if err := s.CredentialStore.UpdateUser(ctx, username, identity); err != nil {
return err
}
s.propagateChange(ctx, func(tx context.Context, client s3_pb.SeaweedS3IamCacheClient) error {
if _, err := client.PutIdentity(tx, &iam_pb.PutIdentityRequest{Identity: identity}); err != nil {
return err
}
if username != identity.Name {
if _, err := client.RemoveIdentity(tx, &iam_pb.RemoveIdentityRequest{Username: username}); err != nil {
return err
}
}
return nil
})
return nil
}
func (s *PropagatingCredentialStore) DeleteUser(ctx context.Context, username string) error {
glog.V(4).Infof("IAM: PropagatingCredentialStore.DeleteUser %s", username)
if err := s.CredentialStore.DeleteUser(ctx, username); err != nil {
return err
}
s.propagateChange(ctx, func(tx context.Context, client s3_pb.SeaweedS3IamCacheClient) error {
_, err := client.RemoveIdentity(tx, &iam_pb.RemoveIdentityRequest{Username: username})
return err
})
return nil
}
func (s *PropagatingCredentialStore) CreateAccessKey(ctx context.Context, username string, credential *iam_pb.Credential) error {
if err := s.CredentialStore.CreateAccessKey(ctx, username, credential); err != nil {
return err
}
// Fetch updated identity to propagate
identity, err := s.CredentialStore.GetUser(ctx, username)
if err != nil {
glog.Warningf("failed to get user %s after creating access key: %v", username, err)
return nil
}
s.propagateChange(ctx, func(tx context.Context, client s3_pb.SeaweedS3IamCacheClient) error {
_, err := client.PutIdentity(tx, &iam_pb.PutIdentityRequest{Identity: identity})
return err
})
return nil
}
func (s *PropagatingCredentialStore) DeleteAccessKey(ctx context.Context, username string, accessKey string) error {
if err := s.CredentialStore.DeleteAccessKey(ctx, username, accessKey); err != nil {
return err
}
// Fetch updated identity to propagate
identity, err := s.CredentialStore.GetUser(ctx, username)
if err != nil {
glog.Warningf("failed to get user %s after deleting access key: %v", username, err)
return nil
}
s.propagateChange(ctx, func(tx context.Context, client s3_pb.SeaweedS3IamCacheClient) error {
_, err := client.PutIdentity(tx, &iam_pb.PutIdentityRequest{Identity: identity})
return err
})
return nil
}
func (s *PropagatingCredentialStore) PutPolicy(ctx context.Context, name string, document policy_engine.PolicyDocument) error {
glog.V(4).Infof("IAM: PropagatingCredentialStore.PutPolicy %s", name)
if err := s.CredentialStore.PutPolicy(ctx, name, document); err != nil {
return err
}
s.propagateChange(ctx, func(tx context.Context, client s3_pb.SeaweedS3IamCacheClient) error {
content, err := json.Marshal(document)
if err != nil {
return err
}
_, err = client.PutPolicy(tx, &iam_pb.PutPolicyRequest{Name: name, Content: string(content)})
return err
})
return nil
}
func (s *PropagatingCredentialStore) DeletePolicy(ctx context.Context, name string) error {
glog.V(4).Infof("IAM: PropagatingCredentialStore.DeletePolicy %s", name)
if err := s.CredentialStore.DeletePolicy(ctx, name); err != nil {
return err
}
s.propagateChange(ctx, func(tx context.Context, client s3_pb.SeaweedS3IamCacheClient) error {
_, err := client.DeletePolicy(tx, &iam_pb.DeletePolicyRequest{Name: name})
return err
})
return nil
}
func (s *PropagatingCredentialStore) ListPolicyNames(ctx context.Context) ([]string, error) {
return s.CredentialStore.ListPolicyNames(ctx)
}
func (s *PropagatingCredentialStore) LoadManagedPolicies(ctx context.Context) ([]*iam_pb.Policy, error) {
if loader, ok := s.CredentialStore.(propagatingManagedPolicyLoader); ok {
return loader.LoadManagedPolicies(ctx)
}
policies, err := s.CredentialStore.GetPolicies(ctx)
if err != nil {
return nil, err
}
managedPolicies := make([]*iam_pb.Policy, 0, len(policies))
for name, policyDocument := range policies {
content, err := json.Marshal(policyDocument)
if err != nil {
return nil, err
}
managedPolicies = append(managedPolicies, &iam_pb.Policy{
Name: name,
Content: string(content),
})
}
return managedPolicies, nil
}
func (s *PropagatingCredentialStore) LoadInlinePolicies(ctx context.Context) (map[string]map[string]policy_engine.PolicyDocument, error) {
if loader, ok := s.CredentialStore.(propagatingInlinePolicyLoader); ok {
return loader.LoadInlinePolicies(ctx)
}
return nil, nil
}
func (s *PropagatingCredentialStore) PutUserInlinePolicy(ctx context.Context, userName, policyName string, document policy_engine.PolicyDocument) error {
if store, ok := s.CredentialStore.(InlinePolicyStore); ok {
return store.PutUserInlinePolicy(ctx, userName, policyName, document)
}
return nil
}
func (s *PropagatingCredentialStore) GetUserInlinePolicy(ctx context.Context, userName, policyName string) (*policy_engine.PolicyDocument, error) {
if store, ok := s.CredentialStore.(InlinePolicyStore); ok {
return store.GetUserInlinePolicy(ctx, userName, policyName)
}
return nil, nil
}
func (s *PropagatingCredentialStore) DeleteUserInlinePolicy(ctx context.Context, userName, policyName string) error {
if store, ok := s.CredentialStore.(InlinePolicyStore); ok {
return store.DeleteUserInlinePolicy(ctx, userName, policyName)
}
return nil
}
func (s *PropagatingCredentialStore) ListUserInlinePolicies(ctx context.Context, userName string) ([]string, error) {
if store, ok := s.CredentialStore.(InlinePolicyStore); ok {
return store.ListUserInlinePolicies(ctx, userName)
}
return nil, nil
}
func (s *PropagatingCredentialStore) PutGroupInlinePolicy(ctx context.Context, groupName, policyName string, document policy_engine.PolicyDocument) error {
if store, ok := s.CredentialStore.(GroupInlinePolicyStore); ok {
return store.PutGroupInlinePolicy(ctx, groupName, policyName, document)
}
return nil
}
func (s *PropagatingCredentialStore) GetGroupInlinePolicy(ctx context.Context, groupName, policyName string) (*policy_engine.PolicyDocument, error) {
if store, ok := s.CredentialStore.(GroupInlinePolicyStore); ok {
return store.GetGroupInlinePolicy(ctx, groupName, policyName)
}
return nil, nil
}
func (s *PropagatingCredentialStore) DeleteGroupInlinePolicy(ctx context.Context, groupName, policyName string) error {
if store, ok := s.CredentialStore.(GroupInlinePolicyStore); ok {
return store.DeleteGroupInlinePolicy(ctx, groupName, policyName)
}
return nil
}
func (s *PropagatingCredentialStore) ListGroupInlinePolicies(ctx context.Context, groupName string) ([]string, error) {
if store, ok := s.CredentialStore.(GroupInlinePolicyStore); ok {
return store.ListGroupInlinePolicies(ctx, groupName)
}
return nil, nil
}
func (s *PropagatingCredentialStore) LoadGroupInlinePolicies(ctx context.Context) (map[string]map[string]policy_engine.PolicyDocument, error) {
if loader, ok := s.CredentialStore.(GroupInlinePoliciesLoader); ok {
return loader.LoadGroupInlinePolicies(ctx)
}
return nil, nil
}
func (s *PropagatingCredentialStore) CreatePolicy(ctx context.Context, name string, document policy_engine.PolicyDocument) error {
if pm, ok := s.CredentialStore.(PolicyManager); ok {
if err := pm.CreatePolicy(ctx, name, document); err != nil {
return err
}
} else {
if err := s.CredentialStore.PutPolicy(ctx, name, document); err != nil {
return err
}
}
s.propagateChange(ctx, func(tx context.Context, client s3_pb.SeaweedS3IamCacheClient) error {
content, err := json.Marshal(document)
if err != nil {
return err
}
_, err = client.PutPolicy(tx, &iam_pb.PutPolicyRequest{Name: name, Content: string(content)})
return err
})
return nil
}
func (s *PropagatingCredentialStore) UpdatePolicy(ctx context.Context, name string, document policy_engine.PolicyDocument) error {
if pm, ok := s.CredentialStore.(PolicyManager); ok {
if err := pm.UpdatePolicy(ctx, name, document); err != nil {
return err
}
} else {
if err := s.CredentialStore.PutPolicy(ctx, name, document); err != nil {
return err
}
}
s.propagateChange(ctx, func(tx context.Context, client s3_pb.SeaweedS3IamCacheClient) error {
content, err := json.Marshal(document)
if err != nil {
return err
}
_, err = client.PutPolicy(tx, &iam_pb.PutPolicyRequest{Name: name, Content: string(content)})
return err
})
return nil
}
func (s *PropagatingCredentialStore) CreateServiceAccount(ctx context.Context, sa *iam_pb.ServiceAccount) error {
glog.V(4).Infof("IAM: PropagatingCredentialStore.CreateServiceAccount %s (parent: %s)", sa.Id, sa.ParentUser)
if err := s.CredentialStore.CreateServiceAccount(ctx, sa); err != nil {
return err
}
// Fetch parent identity to propagate
identity, err := s.CredentialStore.GetUser(ctx, sa.ParentUser)
if err != nil {
glog.Warningf("failed to get parent user %s after creating service account: %v", sa.ParentUser, err)
return nil
}
s.propagateChange(ctx, func(tx context.Context, client s3_pb.SeaweedS3IamCacheClient) error {
_, err := client.PutIdentity(tx, &iam_pb.PutIdentityRequest{Identity: identity})
return err
})
return nil
}
func (s *PropagatingCredentialStore) UpdateServiceAccount(ctx context.Context, id string, sa *iam_pb.ServiceAccount) error {
if err := s.CredentialStore.UpdateServiceAccount(ctx, id, sa); err != nil {
return err
}
// Fetch parent identity to propagate
identity, err := s.CredentialStore.GetUser(ctx, sa.ParentUser)
if err != nil {
glog.Warningf("failed to get parent user %s after updating service account: %v", sa.ParentUser, err)
return nil
}
s.propagateChange(ctx, func(tx context.Context, client s3_pb.SeaweedS3IamCacheClient) error {
_, err := client.PutIdentity(tx, &iam_pb.PutIdentityRequest{Identity: identity})
return err
})
return nil
}
func (s *PropagatingCredentialStore) DeleteServiceAccount(ctx context.Context, id string) error {
// Retrieve SA first to get ParentUser
sa, err := s.CredentialStore.GetServiceAccount(ctx, id)
if err != nil {
// If accessing non-existent SA, just proceed to delete (idempotency)
// But we can't propagate to parent...
if err := s.CredentialStore.DeleteServiceAccount(ctx, id); err != nil {
return err
}
return nil
}
if err := s.CredentialStore.DeleteServiceAccount(ctx, id); err != nil {
return err
}
// Fetch parent identity to propagate
identity, err := s.CredentialStore.GetUser(ctx, sa.ParentUser)
if err != nil {
glog.Warningf("failed to get parent user %s after deleting service account: %v", sa.ParentUser, err)
return nil
}
s.propagateChange(ctx, func(tx context.Context, client s3_pb.SeaweedS3IamCacheClient) error {
_, err := client.PutIdentity(tx, &iam_pb.PutIdentityRequest{Identity: identity})
return err
})
return nil
}
func (s *PropagatingCredentialStore) CreateGroup(ctx context.Context, group *iam_pb.Group) error {
if group != nil {
glog.V(4).Infof("IAM: PropagatingCredentialStore.CreateGroup %s", group.Name)
}
if err := s.CredentialStore.CreateGroup(ctx, group); err != nil {
return err
}
s.propagateChange(ctx, func(tx context.Context, client s3_pb.SeaweedS3IamCacheClient) error {
_, err := client.PutGroup(tx, &iam_pb.PutGroupRequest{Group: group})
return err
})
return nil
}
func (s *PropagatingCredentialStore) GetGroup(ctx context.Context, groupName string) (*iam_pb.Group, error) {
return s.CredentialStore.GetGroup(ctx, groupName)
}
func (s *PropagatingCredentialStore) DeleteGroup(ctx context.Context, groupName string) error {
glog.V(4).Infof("IAM: PropagatingCredentialStore.DeleteGroup %s", groupName)
if err := s.CredentialStore.DeleteGroup(ctx, groupName); err != nil {
return err
}
s.propagateChange(ctx, func(tx context.Context, client s3_pb.SeaweedS3IamCacheClient) error {
_, err := client.RemoveGroup(tx, &iam_pb.RemoveGroupRequest{GroupName: groupName})
return err
})
return nil
}
func (s *PropagatingCredentialStore) ListGroups(ctx context.Context) ([]string, error) {
return s.CredentialStore.ListGroups(ctx)
}
func (s *PropagatingCredentialStore) UpdateGroup(ctx context.Context, group *iam_pb.Group) error {
if group != nil {
glog.V(4).Infof("IAM: PropagatingCredentialStore.UpdateGroup %s", group.Name)
}
if err := s.CredentialStore.UpdateGroup(ctx, group); err != nil {
return err
}
s.propagateChange(ctx, func(tx context.Context, client s3_pb.SeaweedS3IamCacheClient) error {
_, err := client.PutGroup(tx, &iam_pb.PutGroupRequest{Group: group})
return err
})
return nil
}