Skip to content

SeaweedFS is a distributed storage system for object storage (S3), file systems, Iceberg table buckets, and Lance vector buckets, designed to handle billions of files with O(1) disk access and effortless horizontal scaling.

s3: evaluate multipart completion conditions under the object lock (#12209)

Chris Lu2737315e9d64
diff --git a/other/java/client/src/main/proto/filer.proto b/other/java/client/src/main/proto/filer.proto
index feb37fbd7f..2a21f1e947 100644
--- a/other/java/client/src/main/proto/filer.proto
+++ b/other/java/client/src/main/proto/filer.proto
@@ -359,6 +359,8 @@ message ObjectMutation {
bool touch_mtime = 12; // PATCH_EXTENDED: set the entry's Mtime to now (e.g. a metadata-replace copy)
bool remove_empty_parent = 13; // DELETE: also remove the parent directory when the delete leaves it empty (best-effort)
bool keep_remote_object = 14; // DELETE: leave the object on the mounted remote storage
+ WriteCondition condition = 15; // optional precondition checked under the transaction lock just before this mutation applies
+ string condition_key = 16; // entry the condition evaluates against; empty = the mutation's target
}
// Recompute re-derives a pointer entry (directory/name on the mutation) from the
@@ -404,6 +406,7 @@ message ObjectTransactionRequest {
message ObjectTransactionResponse {
string error = 1;
FilerError error_code = 2;
+ optional int32 failed_mutation = 3; // index of the mutation that failed (its condition or its apply); unset when the request-level condition failed
}
// PosixLockRange is one advisory byte-range lock. Owner identity is (sid, owner):
diff --git a/weed/pb/filer.proto b/weed/pb/filer.proto
index feb37fbd7f..dd1d6edb05 100644
--- a/weed/pb/filer.proto
+++ b/weed/pb/filer.proto
@@ -359,6 +359,8 @@ message ObjectMutation {
bool touch_mtime = 12; // PATCH_EXTENDED: set the entry's Mtime to now (e.g. a metadata-replace copy)
bool remove_empty_parent = 13; // DELETE: also remove the parent directory when the delete leaves it empty (best-effort)
bool keep_remote_object = 14; // DELETE: leave the object on the mounted remote storage
+ WriteCondition condition = 15; // optional precondition checked under the transaction lock just before this mutation applies
+ string condition_key = 16; // entry the condition evaluates against; empty = the mutation's target
}
// Recompute re-derives a pointer entry (directory/name on the mutation) from the
@@ -399,11 +401,13 @@ message ObjectTransactionRequest {
bool is_moved = 8; // set on a forwarded transaction so the receiver applies it locally instead of forwarding again
string request_id = 9; // idempotency token: the owner records the outcome and replays it on a retried request instead of re-applying
string condition_alt_key = 10; // if the condition's entry is absent, evaluate it against this entry instead (e.g. null slot, else latest version)
+ bool condition_prefer_alt = 11; // evaluate condition_alt_key (the latest version) first; fall back to the primary entry only when the alt path cannot resolve
}
message ObjectTransactionResponse {
string error = 1;
FilerError error_code = 2;
+ optional int32 failed_mutation = 3; // index of the mutation that failed (its condition or its apply); unset when the request-level condition failed
}
// PosixLockRange is one advisory byte-range lock. Owner identity is (sid, owner):
diff --git a/weed/pb/filer_pb/filer.pb.go b/weed/pb/filer_pb/filer.pb.go
index cb5f44b2b5..e4efbf7528 100644
--- a/weed/pb/filer_pb/filer.pb.go
+++ b/weed/pb/filer_pb/filer.pb.go
@@ -1561,6 +1561,8 @@ type ObjectMutation struct {
TouchMtime bool `protobuf:"varint,12,opt,name=touch_mtime,json=touchMtime,proto3" json:"touch_mtime,omitempty"` // PATCH_EXTENDED: set the entry's Mtime to now (e.g. a metadata-replace copy)
RemoveEmptyParent bool `protobuf:"varint,13,opt,name=remove_empty_parent,json=removeEmptyParent,proto3" json:"remove_empty_parent,omitempty"` // DELETE: also remove the parent directory when the delete leaves it empty (best-effort)
KeepRemoteObject bool `protobuf:"varint,14,opt,name=keep_remote_object,json=keepRemoteObject,proto3" json:"keep_remote_object,omitempty"` // DELETE: leave the object on the mounted remote storage
+ Condition *WriteCondition `protobuf:"bytes,15,opt,name=condition,proto3" json:"condition,omitempty"` // optional precondition checked under the transaction lock just before this mutation applies
+ ConditionKey string `protobuf:"bytes,16,opt,name=condition_key,json=conditionKey,proto3" json:"condition_key,omitempty"` // entry the condition evaluates against; empty = the mutation's target
unknownFields protoimpl.UnknownFields
sizeCache protoimpl.SizeCache
}
@@ -1693,6 +1695,20 @@ func (x *ObjectMutation) GetKeepRemoteObject() bool {
return false
}
+func (x *ObjectMutation) GetCondition() *WriteCondition {
+ if x != nil {
+ return x.Condition
+ }
+ return nil
+}
+
+func (x *ObjectMutation) GetConditionKey() string {
+ if x != nil {
+ return x.ConditionKey
+ }
+ return ""
+}
+
// Recompute re-derives a pointer entry (directory/name on the mutation) from the
// current contents of a scanned directory, atomically under the transaction's
// lock. It is mechanical: the filer picks the child that sorts first or last by
@@ -1830,11 +1846,12 @@ type ObjectTransactionRequest struct {
Mutations []*ObjectMutation `protobuf:"bytes,3,rep,name=mutations,proto3" json:"mutations,omitempty"`
IsFromOtherCluster bool `protobuf:"varint,4,opt,name=is_from_other_cluster,json=isFromOtherCluster,proto3" json:"is_from_other_cluster,omitempty"`
Signatures []int32 `protobuf:"varint,5,rep,packed,name=signatures,proto3" json:"signatures,omitempty"`
- ConditionKey string `protobuf:"bytes,6,opt,name=condition_key,json=conditionKey,proto3" json:"condition_key,omitempty"` // if set, evaluate the condition against this entry instead of lock_key (still locking lock_key)
- RouteKey string `protobuf:"bytes,7,opt,name=route_key,json=routeKey,proto3" json:"route_key,omitempty"` // ring key identifying the owner filer; a non-owner forwards the whole transaction to it
- IsMoved bool `protobuf:"varint,8,opt,name=is_moved,json=isMoved,proto3" json:"is_moved,omitempty"` // set on a forwarded transaction so the receiver applies it locally instead of forwarding again
- RequestId string `protobuf:"bytes,9,opt,name=request_id,json=requestId,proto3" json:"request_id,omitempty"` // idempotency token: the owner records the outcome and replays it on a retried request instead of re-applying
- ConditionAltKey string `protobuf:"bytes,10,opt,name=condition_alt_key,json=conditionAltKey,proto3" json:"condition_alt_key,omitempty"` // if the condition's entry is absent, evaluate it against this entry instead (e.g. null slot, else latest version)
+ ConditionKey string `protobuf:"bytes,6,opt,name=condition_key,json=conditionKey,proto3" json:"condition_key,omitempty"` // if set, evaluate the condition against this entry instead of lock_key (still locking lock_key)
+ RouteKey string `protobuf:"bytes,7,opt,name=route_key,json=routeKey,proto3" json:"route_key,omitempty"` // ring key identifying the owner filer; a non-owner forwards the whole transaction to it
+ IsMoved bool `protobuf:"varint,8,opt,name=is_moved,json=isMoved,proto3" json:"is_moved,omitempty"` // set on a forwarded transaction so the receiver applies it locally instead of forwarding again
+ RequestId string `protobuf:"bytes,9,opt,name=request_id,json=requestId,proto3" json:"request_id,omitempty"` // idempotency token: the owner records the outcome and replays it on a retried request instead of re-applying
+ ConditionAltKey string `protobuf:"bytes,10,opt,name=condition_alt_key,json=conditionAltKey,proto3" json:"condition_alt_key,omitempty"` // if the condition's entry is absent, evaluate it against this entry instead (e.g. null slot, else latest version)
+ ConditionPreferAlt bool `protobuf:"varint,11,opt,name=condition_prefer_alt,json=conditionPreferAlt,proto3" json:"condition_prefer_alt,omitempty"` // evaluate condition_alt_key (the latest version) first; fall back to the primary entry only when the alt path cannot resolve
unknownFields protoimpl.UnknownFields
sizeCache protoimpl.SizeCache
}
@@ -1939,12 +1956,20 @@ func (x *ObjectTransactionRequest) GetConditionAltKey() string {
return ""
}
+func (x *ObjectTransactionRequest) GetConditionPreferAlt() bool {
+ if x != nil {
+ return x.ConditionPreferAlt
+ }
+ return false
+}
+
type ObjectTransactionResponse struct {
- state protoimpl.MessageState `protogen:"open.v1"`
- Error string `protobuf:"bytes,1,opt,name=error,proto3" json:"error,omitempty"`
- ErrorCode FilerError `protobuf:"varint,2,opt,name=error_code,json=errorCode,proto3,enum=filer_pb.FilerError" json:"error_code,omitempty"`
- unknownFields protoimpl.UnknownFields
- sizeCache protoimpl.SizeCache
+ state protoimpl.MessageState `protogen:"open.v1"`
+ Error string `protobuf:"bytes,1,opt,name=error,proto3" json:"error,omitempty"`
+ ErrorCode FilerError `protobuf:"varint,2,opt,name=error_code,json=errorCode,proto3,enum=filer_pb.FilerError" json:"error_code,omitempty"`
+ FailedMutation *int32 `protobuf:"varint,3,opt,name=failed_mutation,json=failedMutation,proto3,oneof" json:"failed_mutation,omitempty"` // index of the mutation that failed (its condition or its apply); unset when the request-level condition failed
+ unknownFields protoimpl.UnknownFields
+ sizeCache protoimpl.SizeCache
}
func (x *ObjectTransactionResponse) Reset() {
@@ -1991,6 +2016,13 @@ func (x *ObjectTransactionResponse) GetErrorCode() FilerError {
return FilerError_OK
}
+func (x *ObjectTransactionResponse) GetFailedMutation() int32 {
+ if x != nil && x.FailedMutation != nil {
+ return *x.FailedMutation
+ }
+ return 0
+}
+
// PosixLockRange is one advisory byte-range lock. Owner identity is (sid, owner):
// sid is the mount session, owner the FUSE lock owner within it, so owners from
// different mounts never alias. end is inclusive (max uint64 = to EOF); is_flock
@@ -7497,7 +7529,7 @@ const file_filer_proto_rawDesc = "" +
"\x18IF_EXTENDED_TIME_ELAPSED\x10\b\x12\x13\n" +
"\x0fIF_CHUNKS_EQUAL\x10\t\x12\x12\n" +
"\x0eIF_ENTRY_EQUAL\x10\n" +
- "\"\xd0\x05\n" +
+ "\"\xad\x06\n" +
"\x0eObjectMutation\x121\n" +
"\x04type\x18\x01 \x01(\x0e2\x1d.filer_pb.ObjectMutation.TypeR\x04type\x12\x1c\n" +
"\tdirectory\x18\x02 \x01(\tR\tdirectory\x12\x12\n" +
@@ -7515,7 +7547,9 @@ const file_filer_proto_rawDesc = "" +
"\vtouch_mtime\x18\f \x01(\bR\n" +
"touchMtime\x12.\n" +
"\x13remove_empty_parent\x18\r \x01(\bR\x11removeEmptyParent\x12,\n" +
- "\x12keep_remote_object\x18\x0e \x01(\bR\x10keepRemoteObject\x1a>\n" +
+ "\x12keep_remote_object\x18\x0e \x01(\bR\x10keepRemoteObject\x126\n" +
+ "\tcondition\x18\x0f \x01(\v2\x18.filer_pb.WriteConditionR\tcondition\x12#\n" +
+ "\rcondition_key\x18\x10 \x01(\tR\fconditionKey\x1a>\n" +
"\x10SetExtendedEntry\x12\x10\n" +
"\x03key\x18\x01 \x01(\tR\x03key\x12\x14\n" +
"\x05value\x18\x02 \x01(\fR\x05value:\x028\x01\"E\n" +
@@ -7543,7 +7577,7 @@ const file_filer_proto_rawDesc = "" +
" \x01(\tR\x11onlyIfCurrentName\x1a?\n" +
"\x11CopyExtendedEntry\x12\x10\n" +
"\x03key\x18\x01 \x01(\tR\x03key\x12\x14\n" +
- "\x05value\x18\x02 \x01(\tR\x05value:\x028\x01\"\xa0\x03\n" +
+ "\x05value\x18\x02 \x01(\tR\x05value:\x028\x01\"\xd2\x03\n" +
"\x18ObjectTransactionRequest\x12\x19\n" +
"\block_key\x18\x01 \x01(\tR\alockKey\x126\n" +
"\tcondition\x18\x02 \x01(\v2\x18.filer_pb.WriteConditionR\tcondition\x126\n" +
@@ -7558,11 +7592,14 @@ const file_filer_proto_rawDesc = "" +
"\n" +
"request_id\x18\t \x01(\tR\trequestId\x12*\n" +
"\x11condition_alt_key\x18\n" +
- " \x01(\tR\x0fconditionAltKey\"f\n" +
+ " \x01(\tR\x0fconditionAltKey\x120\n" +
+ "\x14condition_prefer_alt\x18\v \x01(\bR\x12conditionPreferAlt\"\xa8\x01\n" +
"\x19ObjectTransactionResponse\x12\x14\n" +
"\x05error\x18\x01 \x01(\tR\x05error\x123\n" +
"\n" +
- "error_code\x18\x02 \x01(\x0e2\x14.filer_pb.FilerErrorR\terrorCode\"\xa1\x01\n" +
+ "error_code\x18\x02 \x01(\x0e2\x14.filer_pb.FilerErrorR\terrorCode\x12,\n" +
+ "\x0ffailed_mutation\x18\x03 \x01(\x05H\x00R\x0efailedMutation\x88\x01\x01B\x12\n" +
+ "\x10_failed_mutation\"\xa1\x01\n" +
"\x0ePosixLockRange\x12\x14\n" +
"\x05start\x18\x01 \x01(\x04R\x05start\x12\x10\n" +
"\x03end\x18\x02 \x01(\x04R\x03end\x12\x12\n" +
@@ -8220,136 +8257,137 @@ var file_filer_proto_depIdxs = []int32{
10, // 17: filer_pb.ObjectMutation.entry:type_name -> filer_pb.Entry
101, // 18: filer_pb.ObjectMutation.set_extended:type_name -> filer_pb.ObjectMutation.SetExtendedEntry
20, // 19: filer_pb.ObjectMutation.recompute:type_name -> filer_pb.Recompute
- 102, // 20: filer_pb.Recompute.copy_extended:type_name -> filer_pb.Recompute.CopyExtendedEntry
- 18, // 21: filer_pb.ObjectTransactionRequest.condition:type_name -> filer_pb.WriteCondition
- 19, // 22: filer_pb.ObjectTransactionRequest.mutations:type_name -> filer_pb.ObjectMutation
- 1, // 23: filer_pb.ObjectTransactionResponse.error_code:type_name -> filer_pb.FilerError
- 2, // 24: filer_pb.PosixLockRequest.op:type_name -> filer_pb.PosixLockOp
- 23, // 25: filer_pb.PosixLockRequest.lock:type_name -> filer_pb.PosixLockRange
- 23, // 26: filer_pb.PosixLockRequest.locks:type_name -> filer_pb.PosixLockRange
- 23, // 27: filer_pb.PosixLockResponse.conflict:type_name -> filer_pb.PosixLockRange
- 21, // 28: filer_pb.ObjectTransactionBatchRequest.transactions:type_name -> filer_pb.ObjectTransactionRequest
- 22, // 29: filer_pb.ObjectTransactionBatchResponse.responses:type_name -> filer_pb.ObjectTransactionResponse
- 59, // 30: filer_pb.CreateEntryResponse.metadata_event:type_name -> filer_pb.SubscribeMetadataResponse
- 1, // 31: filer_pb.CreateEntryResponse.error_code:type_name -> filer_pb.FilerError
- 10, // 32: filer_pb.UpdateEntryRequest.entry:type_name -> filer_pb.Entry
- 103, // 33: filer_pb.UpdateEntryRequest.expected_extended:type_name -> filer_pb.UpdateEntryRequest.ExpectedExtendedEntry
- 18, // 34: filer_pb.UpdateEntryRequest.condition:type_name -> filer_pb.WriteCondition
- 59, // 35: filer_pb.UpdateEntryResponse.metadata_event:type_name -> filer_pb.SubscribeMetadataResponse
- 13, // 36: filer_pb.AppendToEntryRequest.chunks:type_name -> filer_pb.FileChunk
- 104, // 37: filer_pb.DeleteEntryRequest.expected_extended:type_name -> filer_pb.DeleteEntryRequest.ExpectedExtendedEntry
- 59, // 38: filer_pb.DeleteEntryResponse.metadata_event:type_name -> filer_pb.SubscribeMetadataResponse
- 12, // 39: filer_pb.StreamRenameEntryResponse.event_notification:type_name -> filer_pb.EventNotification
- 45, // 40: filer_pb.AssignVolumeResponse.location:type_name -> filer_pb.Location
- 45, // 41: filer_pb.AssignVolumeResponse.replicas:type_name -> filer_pb.Location
- 1, // 42: filer_pb.AssignVolumeResponse.error_code:type_name -> filer_pb.FilerError
- 45, // 43: filer_pb.Locations.locations:type_name -> filer_pb.Location
- 105, // 44: filer_pb.LookupVolumeResponse.locations_map:type_name -> filer_pb.LookupVolumeResponse.LocationsMapEntry
- 47, // 45: filer_pb.CollectionListResponse.collections:type_name -> filer_pb.Collection
- 12, // 46: filer_pb.SubscribeMetadataResponse.event_notification:type_name -> filer_pb.EventNotification
- 59, // 47: filer_pb.SubscribeMetadataResponse.events:type_name -> filer_pb.SubscribeMetadataResponse
- 63, // 48: filer_pb.SubscribeMetadataResponse.log_file_refs:type_name -> filer_pb.LogFileChunkRef
- 62, // 49: filer_pb.ListMetadataSubscribersResponse.subscribers:type_name -> filer_pb.MetadataSubscriber
- 13, // 50: filer_pb.LogFileChunkRef.chunks:type_name -> filer_pb.FileChunk
- 10, // 51: filer_pb.TraverseBfsMetadataResponse.entry:type_name -> filer_pb.Entry
- 106, // 52: filer_pb.LocateBrokerResponse.resources:type_name -> filer_pb.LocateBrokerResponse.Resource
- 107, // 53: filer_pb.FilerConf.locations:type_name -> filer_pb.FilerConf.PathConf
- 10, // 54: filer_pb.CacheRemoteObjectToLocalClusterResponse.entry:type_name -> filer_pb.Entry
- 59, // 55: filer_pb.CacheRemoteObjectToLocalClusterResponse.metadata_event:type_name -> filer_pb.SubscribeMetadataResponse
- 84, // 56: filer_pb.TransferLocksRequest.locks:type_name -> filer_pb.Lock
- 17, // 57: filer_pb.StreamMutateEntryRequest.create_request:type_name -> filer_pb.CreateEntryRequest
- 29, // 58: filer_pb.StreamMutateEntryRequest.update_request:type_name -> filer_pb.UpdateEntryRequest
- 35, // 59: filer_pb.StreamMutateEntryRequest.delete_request:type_name -> filer_pb.DeleteEntryRequest
- 39, // 60: filer_pb.StreamMutateEntryRequest.rename_request:type_name -> filer_pb.StreamRenameEntryRequest
- 28, // 61: filer_pb.StreamMutateEntryResponse.create_response:type_name -> filer_pb.CreateEntryResponse
- 30, // 62: filer_pb.StreamMutateEntryResponse.update_response:type_name -> filer_pb.UpdateEntryResponse
- 36, // 63: filer_pb.StreamMutateEntryResponse.delete_response:type_name -> filer_pb.DeleteEntryResponse
- 40, // 64: filer_pb.StreamMutateEntryResponse.rename_response:type_name -> filer_pb.StreamRenameEntryResponse
- 95, // 65: filer_pb.MountListResponse.mounts:type_name -> filer_pb.MountInfo
- 5, // 66: filer_pb.LookupDirectoryEntriesRequest.requests:type_name -> filer_pb.LookupDirectoryEntryRequest
- 10, // 67: filer_pb.LookupDirectoryEntryResult.entry:type_name -> filer_pb.Entry
- 97, // 68: filer_pb.LookupDirectoryEntriesResponse.results:type_name -> filer_pb.LookupDirectoryEntryResult
- 108, // 69: filer_pb.LookupDirectoryEntriesResponse.locations_map:type_name -> filer_pb.LookupDirectoryEntriesResponse.LocationsMapEntry
- 109, // 70: filer_pb.LookupDirectoryEntriesResponse.read_auth:type_name -> filer_pb.LookupDirectoryEntriesResponse.ReadAuthEntry
- 3, // 71: filer_pb.WriteCondition.Clause.kind:type_name -> filer_pb.WriteCondition.Kind
- 10, // 72: filer_pb.WriteCondition.Clause.expected_entry:type_name -> filer_pb.Entry
- 44, // 73: filer_pb.LookupVolumeResponse.LocationsMapEntry.value:type_name -> filer_pb.Locations
- 44, // 74: filer_pb.LookupDirectoryEntriesResponse.LocationsMapEntry.value:type_name -> filer_pb.Locations
- 5, // 75: filer_pb.SeaweedFiler.LookupDirectoryEntry:input_type -> filer_pb.LookupDirectoryEntryRequest
- 96, // 76: filer_pb.SeaweedFiler.LookupDirectoryEntries:input_type -> filer_pb.LookupDirectoryEntriesRequest
- 7, // 77: filer_pb.SeaweedFiler.ListEntries:input_type -> filer_pb.ListEntriesRequest
- 17, // 78: filer_pb.SeaweedFiler.CreateEntry:input_type -> filer_pb.CreateEntryRequest
- 29, // 79: filer_pb.SeaweedFiler.UpdateEntry:input_type -> filer_pb.UpdateEntryRequest
- 31, // 80: filer_pb.SeaweedFiler.TouchAccessTime:input_type -> filer_pb.TouchAccessTimeRequest
- 33, // 81: filer_pb.SeaweedFiler.AppendToEntry:input_type -> filer_pb.AppendToEntryRequest
- 35, // 82: filer_pb.SeaweedFiler.DeleteEntry:input_type -> filer_pb.DeleteEntryRequest
- 21, // 83: filer_pb.SeaweedFiler.ObjectTransaction:input_type -> filer_pb.ObjectTransactionRequest
- 26, // 84: filer_pb.SeaweedFiler.ObjectTransactionBatch:input_type -> filer_pb.ObjectTransactionBatchRequest
- 24, // 85: filer_pb.SeaweedFiler.PosixLock:input_type -> filer_pb.PosixLockRequest
- 37, // 86: filer_pb.SeaweedFiler.AtomicRenameEntry:input_type -> filer_pb.AtomicRenameEntryRequest
- 39, // 87: filer_pb.SeaweedFiler.StreamRenameEntry:input_type -> filer_pb.StreamRenameEntryRequest
- 89, // 88: filer_pb.SeaweedFiler.StreamMutateEntry:input_type -> filer_pb.StreamMutateEntryRequest
- 41, // 89: filer_pb.SeaweedFiler.AssignVolume:input_type -> filer_pb.AssignVolumeRequest
- 43, // 90: filer_pb.SeaweedFiler.LookupVolume:input_type -> filer_pb.LookupVolumeRequest
- 48, // 91: filer_pb.SeaweedFiler.CollectionList:input_type -> filer_pb.CollectionListRequest
- 50, // 92: filer_pb.SeaweedFiler.DeleteCollection:input_type -> filer_pb.DeleteCollectionRequest
- 52, // 93: filer_pb.SeaweedFiler.Statistics:input_type -> filer_pb.StatisticsRequest
- 54, // 94: filer_pb.SeaweedFiler.Ping:input_type -> filer_pb.PingRequest
- 56, // 95: filer_pb.SeaweedFiler.GetFilerConfiguration:input_type -> filer_pb.GetFilerConfigurationRequest
- 64, // 96: filer_pb.SeaweedFiler.TraverseBfsMetadata:input_type -> filer_pb.TraverseBfsMetadataRequest
- 58, // 97: filer_pb.SeaweedFiler.SubscribeMetadata:input_type -> filer_pb.SubscribeMetadataRequest
- 58, // 98: filer_pb.SeaweedFiler.SubscribeLocalMetadata:input_type -> filer_pb.SubscribeMetadataRequest
- 60, // 99: filer_pb.SeaweedFiler.ListMetadataSubscribers:input_type -> filer_pb.ListMetadataSubscribersRequest
- 71, // 100: filer_pb.SeaweedFiler.KvGet:input_type -> filer_pb.KvGetRequest
- 73, // 101: filer_pb.SeaweedFiler.KvPut:input_type -> filer_pb.KvPutRequest
- 76, // 102: filer_pb.SeaweedFiler.CacheRemoteObjectToLocalCluster:input_type -> filer_pb.CacheRemoteObjectToLocalClusterRequest
- 78, // 103: filer_pb.SeaweedFiler.DistributedLock:input_type -> filer_pb.LockRequest
- 80, // 104: filer_pb.SeaweedFiler.DistributedUnlock:input_type -> filer_pb.UnlockRequest
- 82, // 105: filer_pb.SeaweedFiler.FindLockOwner:input_type -> filer_pb.FindLockOwnerRequest
- 85, // 106: filer_pb.SeaweedFiler.TransferLocks:input_type -> filer_pb.TransferLocksRequest
- 87, // 107: filer_pb.SeaweedFiler.ReplicateLock:input_type -> filer_pb.ReplicateLockRequest
- 91, // 108: filer_pb.SeaweedFiler.MountRegister:input_type -> filer_pb.MountRegisterRequest
- 93, // 109: filer_pb.SeaweedFiler.MountList:input_type -> filer_pb.MountListRequest
- 6, // 110: filer_pb.SeaweedFiler.LookupDirectoryEntry:output_type -> filer_pb.LookupDirectoryEntryResponse
- 98, // 111: filer_pb.SeaweedFiler.LookupDirectoryEntries:output_type -> filer_pb.LookupDirectoryEntriesResponse
- 8, // 112: filer_pb.SeaweedFiler.ListEntries:output_type -> filer_pb.ListEntriesResponse
- 28, // 113: filer_pb.SeaweedFiler.CreateEntry:output_type -> filer_pb.CreateEntryResponse
- 30, // 114: filer_pb.SeaweedFiler.UpdateEntry:output_type -> filer_pb.UpdateEntryResponse
- 32, // 115: filer_pb.SeaweedFiler.TouchAccessTime:output_type -> filer_pb.TouchAccessTimeResponse
- 34, // 116: filer_pb.SeaweedFiler.AppendToEntry:output_type -> filer_pb.AppendToEntryResponse
- 36, // 117: filer_pb.SeaweedFiler.DeleteEntry:output_type -> filer_pb.DeleteEntryResponse
- 22, // 118: filer_pb.SeaweedFiler.ObjectTransaction:output_type -> filer_pb.ObjectTransactionResponse
- 27, // 119: filer_pb.SeaweedFiler.ObjectTransactionBatch:output_type -> filer_pb.ObjectTransactionBatchResponse
- 25, // 120: filer_pb.SeaweedFiler.PosixLock:output_type -> filer_pb.PosixLockResponse
- 38, // 121: filer_pb.SeaweedFiler.AtomicRenameEntry:output_type -> filer_pb.AtomicRenameEntryResponse
- 40, // 122: filer_pb.SeaweedFiler.StreamRenameEntry:output_type -> filer_pb.StreamRenameEntryResponse
- 90, // 123: filer_pb.SeaweedFiler.StreamMutateEntry:output_type -> filer_pb.StreamMutateEntryResponse
- 42, // 124: filer_pb.SeaweedFiler.AssignVolume:output_type -> filer_pb.AssignVolumeResponse
- 46, // 125: filer_pb.SeaweedFiler.LookupVolume:output_type -> filer_pb.LookupVolumeResponse
- 49, // 126: filer_pb.SeaweedFiler.CollectionList:output_type -> filer_pb.CollectionListResponse
- 51, // 127: filer_pb.SeaweedFiler.DeleteCollection:output_type -> filer_pb.DeleteCollectionResponse
- 53, // 128: filer_pb.SeaweedFiler.Statistics:output_type -> filer_pb.StatisticsResponse
- 55, // 129: filer_pb.SeaweedFiler.Ping:output_type -> filer_pb.PingResponse
- 57, // 130: filer_pb.SeaweedFiler.GetFilerConfiguration:output_type -> filer_pb.GetFilerConfigurationResponse
- 65, // 131: filer_pb.SeaweedFiler.TraverseBfsMetadata:output_type -> filer_pb.TraverseBfsMetadataResponse
- 59, // 132: filer_pb.SeaweedFiler.SubscribeMetadata:output_type -> filer_pb.SubscribeMetadataResponse
- 59, // 133: filer_pb.SeaweedFiler.SubscribeLocalMetadata:output_type -> filer_pb.SubscribeMetadataResponse
- 61, // 134: filer_pb.SeaweedFiler.ListMetadataSubscribers:output_type -> filer_pb.ListMetadataSubscribersResponse
- 72, // 135: filer_pb.SeaweedFiler.KvGet:output_type -> filer_pb.KvGetResponse
- 74, // 136: filer_pb.SeaweedFiler.KvPut:output_type -> filer_pb.KvPutResponse
- 77, // 137: filer_pb.SeaweedFiler.CacheRemoteObjectToLocalCluster:output_type -> filer_pb.CacheRemoteObjectToLocalClusterResponse
- 79, // 138: filer_pb.SeaweedFiler.DistributedLock:output_type -> filer_pb.LockResponse
- 81, // 139: filer_pb.SeaweedFiler.DistributedUnlock:output_type -> filer_pb.UnlockResponse
- 83, // 140: filer_pb.SeaweedFiler.FindLockOwner:output_type -> filer_pb.FindLockOwnerResponse
- 86, // 141: filer_pb.SeaweedFiler.TransferLocks:output_type -> filer_pb.TransferLocksResponse
- 88, // 142: filer_pb.SeaweedFiler.ReplicateLock:output_type -> filer_pb.ReplicateLockResponse
- 92, // 143: filer_pb.SeaweedFiler.MountRegister:output_type -> filer_pb.MountRegisterResponse
- 94, // 144: filer_pb.SeaweedFiler.MountList:output_type -> filer_pb.MountListResponse
- 110, // [110:145] is the sub-list for method output_type
- 75, // [75:110] is the sub-list for method input_type
- 75, // [75:75] is the sub-list for extension type_name
- 75, // [75:75] is the sub-list for extension extendee
- 0, // [0:75] is the sub-list for field type_name
+ 18, // 20: filer_pb.ObjectMutation.condition:type_name -> filer_pb.WriteCondition
+ 102, // 21: filer_pb.Recompute.copy_extended:type_name -> filer_pb.Recompute.CopyExtendedEntry
+ 18, // 22: filer_pb.ObjectTransactionRequest.condition:type_name -> filer_pb.WriteCondition
+ 19, // 23: filer_pb.ObjectTransactionRequest.mutations:type_name -> filer_pb.ObjectMutation
+ 1, // 24: filer_pb.ObjectTransactionResponse.error_code:type_name -> filer_pb.FilerError
+ 2, // 25: filer_pb.PosixLockRequest.op:type_name -> filer_pb.PosixLockOp
+ 23, // 26: filer_pb.PosixLockRequest.lock:type_name -> filer_pb.PosixLockRange
+ 23, // 27: filer_pb.PosixLockRequest.locks:type_name -> filer_pb.PosixLockRange
+ 23, // 28: filer_pb.PosixLockResponse.conflict:type_name -> filer_pb.PosixLockRange
+ 21, // 29: filer_pb.ObjectTransactionBatchRequest.transactions:type_name -> filer_pb.ObjectTransactionRequest
+ 22, // 30: filer_pb.ObjectTransactionBatchResponse.responses:type_name -> filer_pb.ObjectTransactionResponse
+ 59, // 31: filer_pb.CreateEntryResponse.metadata_event:type_name -> filer_pb.SubscribeMetadataResponse
+ 1, // 32: filer_pb.CreateEntryResponse.error_code:type_name -> filer_pb.FilerError
+ 10, // 33: filer_pb.UpdateEntryRequest.entry:type_name -> filer_pb.Entry
+ 103, // 34: filer_pb.UpdateEntryRequest.expected_extended:type_name -> filer_pb.UpdateEntryRequest.ExpectedExtendedEntry
+ 18, // 35: filer_pb.UpdateEntryRequest.condition:type_name -> filer_pb.WriteCondition
+ 59, // 36: filer_pb.UpdateEntryResponse.metadata_event:type_name -> filer_pb.SubscribeMetadataResponse
+ 13, // 37: filer_pb.AppendToEntryRequest.chunks:type_name -> filer_pb.FileChunk
+ 104, // 38: filer_pb.DeleteEntryRequest.expected_extended:type_name -> filer_pb.DeleteEntryRequest.ExpectedExtendedEntry
+ 59, // 39: filer_pb.DeleteEntryResponse.metadata_event:type_name -> filer_pb.SubscribeMetadataResponse
+ 12, // 40: filer_pb.StreamRenameEntryResponse.event_notification:type_name -> filer_pb.EventNotification
+ 45, // 41: filer_pb.AssignVolumeResponse.location:type_name -> filer_pb.Location
+ 45, // 42: filer_pb.AssignVolumeResponse.replicas:type_name -> filer_pb.Location
+ 1, // 43: filer_pb.AssignVolumeResponse.error_code:type_name -> filer_pb.FilerError
+ 45, // 44: filer_pb.Locations.locations:type_name -> filer_pb.Location
+ 105, // 45: filer_pb.LookupVolumeResponse.locations_map:type_name -> filer_pb.LookupVolumeResponse.LocationsMapEntry
+ 47, // 46: filer_pb.CollectionListResponse.collections:type_name -> filer_pb.Collection
+ 12, // 47: filer_pb.SubscribeMetadataResponse.event_notification:type_name -> filer_pb.EventNotification
+ 59, // 48: filer_pb.SubscribeMetadataResponse.events:type_name -> filer_pb.SubscribeMetadataResponse
+ 63, // 49: filer_pb.SubscribeMetadataResponse.log_file_refs:type_name -> filer_pb.LogFileChunkRef
+ 62, // 50: filer_pb.ListMetadataSubscribersResponse.subscribers:type_name -> filer_pb.MetadataSubscriber
+ 13, // 51: filer_pb.LogFileChunkRef.chunks:type_name -> filer_pb.FileChunk
+ 10, // 52: filer_pb.TraverseBfsMetadataResponse.entry:type_name -> filer_pb.Entry
+ 106, // 53: filer_pb.LocateBrokerResponse.resources:type_name -> filer_pb.LocateBrokerResponse.Resource
+ 107, // 54: filer_pb.FilerConf.locations:type_name -> filer_pb.FilerConf.PathConf
+ 10, // 55: filer_pb.CacheRemoteObjectToLocalClusterResponse.entry:type_name -> filer_pb.Entry
+ 59, // 56: filer_pb.CacheRemoteObjectToLocalClusterResponse.metadata_event:type_name -> filer_pb.SubscribeMetadataResponse
+ 84, // 57: filer_pb.TransferLocksRequest.locks:type_name -> filer_pb.Lock
+ 17, // 58: filer_pb.StreamMutateEntryRequest.create_request:type_name -> filer_pb.CreateEntryRequest
+ 29, // 59: filer_pb.StreamMutateEntryRequest.update_request:type_name -> filer_pb.UpdateEntryRequest
+ 35, // 60: filer_pb.StreamMutateEntryRequest.delete_request:type_name -> filer_pb.DeleteEntryRequest
+ 39, // 61: filer_pb.StreamMutateEntryRequest.rename_request:type_name -> filer_pb.StreamRenameEntryRequest
+ 28, // 62: filer_pb.StreamMutateEntryResponse.create_response:type_name -> filer_pb.CreateEntryResponse
+ 30, // 63: filer_pb.StreamMutateEntryResponse.update_response:type_name -> filer_pb.UpdateEntryResponse
+ 36, // 64: filer_pb.StreamMutateEntryResponse.delete_response:type_name -> filer_pb.DeleteEntryResponse
+ 40, // 65: filer_pb.StreamMutateEntryResponse.rename_response:type_name -> filer_pb.StreamRenameEntryResponse
+ 95, // 66: filer_pb.MountListResponse.mounts:type_name -> filer_pb.MountInfo
+ 5, // 67: filer_pb.LookupDirectoryEntriesRequest.requests:type_name -> filer_pb.LookupDirectoryEntryRequest
+ 10, // 68: filer_pb.LookupDirectoryEntryResult.entry:type_name -> filer_pb.Entry
+ 97, // 69: filer_pb.LookupDirectoryEntriesResponse.results:type_name -> filer_pb.LookupDirectoryEntryResult
+ 108, // 70: filer_pb.LookupDirectoryEntriesResponse.locations_map:type_name -> filer_pb.LookupDirectoryEntriesResponse.LocationsMapEntry
+ 109, // 71: filer_pb.LookupDirectoryEntriesResponse.read_auth:type_name -> filer_pb.LookupDirectoryEntriesResponse.ReadAuthEntry
+ 3, // 72: filer_pb.WriteCondition.Clause.kind:type_name -> filer_pb.WriteCondition.Kind
+ 10, // 73: filer_pb.WriteCondition.Clause.expected_entry:type_name -> filer_pb.Entry
+ 44, // 74: filer_pb.LookupVolumeResponse.LocationsMapEntry.value:type_name -> filer_pb.Locations
+ 44, // 75: filer_pb.LookupDirectoryEntriesResponse.LocationsMapEntry.value:type_name -> filer_pb.Locations
+ 5, // 76: filer_pb.SeaweedFiler.LookupDirectoryEntry:input_type -> filer_pb.LookupDirectoryEntryRequest
+ 96, // 77: filer_pb.SeaweedFiler.LookupDirectoryEntries:input_type -> filer_pb.LookupDirectoryEntriesRequest
+ 7, // 78: filer_pb.SeaweedFiler.ListEntries:input_type -> filer_pb.ListEntriesRequest
+ 17, // 79: filer_pb.SeaweedFiler.CreateEntry:input_type -> filer_pb.CreateEntryRequest
+ 29, // 80: filer_pb.SeaweedFiler.UpdateEntry:input_type -> filer_pb.UpdateEntryRequest
+ 31, // 81: filer_pb.SeaweedFiler.TouchAccessTime:input_type -> filer_pb.TouchAccessTimeRequest
+ 33, // 82: filer_pb.SeaweedFiler.AppendToEntry:input_type -> filer_pb.AppendToEntryRequest
+ 35, // 83: filer_pb.SeaweedFiler.DeleteEntry:input_type -> filer_pb.DeleteEntryRequest
+ 21, // 84: filer_pb.SeaweedFiler.ObjectTransaction:input_type -> filer_pb.ObjectTransactionRequest
+ 26, // 85: filer_pb.SeaweedFiler.ObjectTransactionBatch:input_type -> filer_pb.ObjectTransactionBatchRequest
+ 24, // 86: filer_pb.SeaweedFiler.PosixLock:input_type -> filer_pb.PosixLockRequest
+ 37, // 87: filer_pb.SeaweedFiler.AtomicRenameEntry:input_type -> filer_pb.AtomicRenameEntryRequest
+ 39, // 88: filer_pb.SeaweedFiler.StreamRenameEntry:input_type -> filer_pb.StreamRenameEntryRequest
+ 89, // 89: filer_pb.SeaweedFiler.StreamMutateEntry:input_type -> filer_pb.StreamMutateEntryRequest
+ 41, // 90: filer_pb.SeaweedFiler.AssignVolume:input_type -> filer_pb.AssignVolumeRequest
+ 43, // 91: filer_pb.SeaweedFiler.LookupVolume:input_type -> filer_pb.LookupVolumeRequest
+ 48, // 92: filer_pb.SeaweedFiler.CollectionList:input_type -> filer_pb.CollectionListRequest
+ 50, // 93: filer_pb.SeaweedFiler.DeleteCollection:input_type -> filer_pb.DeleteCollectionRequest
+ 52, // 94: filer_pb.SeaweedFiler.Statistics:input_type -> filer_pb.StatisticsRequest
+ 54, // 95: filer_pb.SeaweedFiler.Ping:input_type -> filer_pb.PingRequest
+ 56, // 96: filer_pb.SeaweedFiler.GetFilerConfiguration:input_type -> filer_pb.GetFilerConfigurationRequest
+ 64, // 97: filer_pb.SeaweedFiler.TraverseBfsMetadata:input_type -> filer_pb.TraverseBfsMetadataRequest
+ 58, // 98: filer_pb.SeaweedFiler.SubscribeMetadata:input_type -> filer_pb.SubscribeMetadataRequest
+ 58, // 99: filer_pb.SeaweedFiler.SubscribeLocalMetadata:input_type -> filer_pb.SubscribeMetadataRequest
+ 60, // 100: filer_pb.SeaweedFiler.ListMetadataSubscribers:input_type -> filer_pb.ListMetadataSubscribersRequest
+ 71, // 101: filer_pb.SeaweedFiler.KvGet:input_type -> filer_pb.KvGetRequest
+ 73, // 102: filer_pb.SeaweedFiler.KvPut:input_type -> filer_pb.KvPutRequest
+ 76, // 103: filer_pb.SeaweedFiler.CacheRemoteObjectToLocalCluster:input_type -> filer_pb.CacheRemoteObjectToLocalClusterRequest
+ 78, // 104: filer_pb.SeaweedFiler.DistributedLock:input_type -> filer_pb.LockRequest
+ 80, // 105: filer_pb.SeaweedFiler.DistributedUnlock:input_type -> filer_pb.UnlockRequest
+ 82, // 106: filer_pb.SeaweedFiler.FindLockOwner:input_type -> filer_pb.FindLockOwnerRequest
+ 85, // 107: filer_pb.SeaweedFiler.TransferLocks:input_type -> filer_pb.TransferLocksRequest
+ 87, // 108: filer_pb.SeaweedFiler.ReplicateLock:input_type -> filer_pb.ReplicateLockRequest
+ 91, // 109: filer_pb.SeaweedFiler.MountRegister:input_type -> filer_pb.MountRegisterRequest
+ 93, // 110: filer_pb.SeaweedFiler.MountList:input_type -> filer_pb.MountListRequest
+ 6, // 111: filer_pb.SeaweedFiler.LookupDirectoryEntry:output_type -> filer_pb.LookupDirectoryEntryResponse
+ 98, // 112: filer_pb.SeaweedFiler.LookupDirectoryEntries:output_type -> filer_pb.LookupDirectoryEntriesResponse
+ 8, // 113: filer_pb.SeaweedFiler.ListEntries:output_type -> filer_pb.ListEntriesResponse
+ 28, // 114: filer_pb.SeaweedFiler.CreateEntry:output_type -> filer_pb.CreateEntryResponse
+ 30, // 115: filer_pb.SeaweedFiler.UpdateEntry:output_type -> filer_pb.UpdateEntryResponse
+ 32, // 116: filer_pb.SeaweedFiler.TouchAccessTime:output_type -> filer_pb.TouchAccessTimeResponse
+ 34, // 117: filer_pb.SeaweedFiler.AppendToEntry:output_type -> filer_pb.AppendToEntryResponse
+ 36, // 118: filer_pb.SeaweedFiler.DeleteEntry:output_type -> filer_pb.DeleteEntryResponse
+ 22, // 119: filer_pb.SeaweedFiler.ObjectTransaction:output_type -> filer_pb.ObjectTransactionResponse
+ 27, // 120: filer_pb.SeaweedFiler.ObjectTransactionBatch:output_type -> filer_pb.ObjectTransactionBatchResponse
+ 25, // 121: filer_pb.SeaweedFiler.PosixLock:output_type -> filer_pb.PosixLockResponse
+ 38, // 122: filer_pb.SeaweedFiler.AtomicRenameEntry:output_type -> filer_pb.AtomicRenameEntryResponse
+ 40, // 123: filer_pb.SeaweedFiler.StreamRenameEntry:output_type -> filer_pb.StreamRenameEntryResponse
+ 90, // 124: filer_pb.SeaweedFiler.StreamMutateEntry:output_type -> filer_pb.StreamMutateEntryResponse
+ 42, // 125: filer_pb.SeaweedFiler.AssignVolume:output_type -> filer_pb.AssignVolumeResponse
+ 46, // 126: filer_pb.SeaweedFiler.LookupVolume:output_type -> filer_pb.LookupVolumeResponse
+ 49, // 127: filer_pb.SeaweedFiler.CollectionList:output_type -> filer_pb.CollectionListResponse
+ 51, // 128: filer_pb.SeaweedFiler.DeleteCollection:output_type -> filer_pb.DeleteCollectionResponse
+ 53, // 129: filer_pb.SeaweedFiler.Statistics:output_type -> filer_pb.StatisticsResponse
+ 55, // 130: filer_pb.SeaweedFiler.Ping:output_type -> filer_pb.PingResponse
+ 57, // 131: filer_pb.SeaweedFiler.GetFilerConfiguration:output_type -> filer_pb.GetFilerConfigurationResponse
+ 65, // 132: filer_pb.SeaweedFiler.TraverseBfsMetadata:output_type -> filer_pb.TraverseBfsMetadataResponse
+ 59, // 133: filer_pb.SeaweedFiler.SubscribeMetadata:output_type -> filer_pb.SubscribeMetadataResponse
+ 59, // 134: filer_pb.SeaweedFiler.SubscribeLocalMetadata:output_type -> filer_pb.SubscribeMetadataResponse
+ 61, // 135: filer_pb.SeaweedFiler.ListMetadataSubscribers:output_type -> filer_pb.ListMetadataSubscribersResponse
+ 72, // 136: filer_pb.SeaweedFiler.KvGet:output_type -> filer_pb.KvGetResponse
+ 74, // 137: filer_pb.SeaweedFiler.KvPut:output_type -> filer_pb.KvPutResponse
+ 77, // 138: filer_pb.SeaweedFiler.CacheRemoteObjectToLocalCluster:output_type -> filer_pb.CacheRemoteObjectToLocalClusterResponse
+ 79, // 139: filer_pb.SeaweedFiler.DistributedLock:output_type -> filer_pb.LockResponse
+ 81, // 140: filer_pb.SeaweedFiler.DistributedUnlock:output_type -> filer_pb.UnlockResponse
+ 83, // 141: filer_pb.SeaweedFiler.FindLockOwner:output_type -> filer_pb.FindLockOwnerResponse
+ 86, // 142: filer_pb.SeaweedFiler.TransferLocks:output_type -> filer_pb.TransferLocksResponse
+ 88, // 143: filer_pb.SeaweedFiler.ReplicateLock:output_type -> filer_pb.ReplicateLockResponse
+ 92, // 144: filer_pb.SeaweedFiler.MountRegister:output_type -> filer_pb.MountRegisterResponse
+ 94, // 145: filer_pb.SeaweedFiler.MountList:output_type -> filer_pb.MountListResponse
+ 111, // [111:146] is the sub-list for method output_type
+ 76, // [76:111] is the sub-list for method input_type
+ 76, // [76:76] is the sub-list for extension type_name
+ 76, // [76:76] is the sub-list for extension extendee
+ 0, // [0:76] is the sub-list for field type_name
}
func init() { file_filer_proto_init() }
@@ -8358,6 +8396,7 @@ func file_filer_proto_init() {
return
}
file_filer_proto_msgTypes[4].OneofWrappers = []any{}
+ file_filer_proto_msgTypes[17].OneofWrappers = []any{}
file_filer_proto_msgTypes[84].OneofWrappers = []any{
(*StreamMutateEntryRequest_CreateRequest)(nil),
(*StreamMutateEntryRequest_UpdateRequest)(nil),
diff --git a/weed/pb/filer_pb/filer_vtproto.pb.go b/weed/pb/filer_pb/filer_vtproto.pb.go
index 8cd7734a72..9a3b40547c 100644
--- a/weed/pb/filer_pb/filer_vtproto.pb.go
+++ b/weed/pb/filer_pb/filer_vtproto.pb.go
@@ -1294,6 +1294,25 @@ func (m *ObjectMutation) MarshalToSizedBufferVT(dAtA []byte) (int, error) {
i -= len(m.unknownFields)
copy(dAtA[i:], m.unknownFields)
}
+ if len(m.ConditionKey) > 0 {
+ i -= len(m.ConditionKey)
+ copy(dAtA[i:], m.ConditionKey)
+ i = protohelpers.EncodeVarint(dAtA, i, uint64(len(m.ConditionKey)))
+ i--
+ dAtA[i] = 0x1
+ i--
+ dAtA[i] = 0x82
+ }
+ if m.Condition != nil {
+ size, err := m.Condition.MarshalToSizedBufferVT(dAtA[:i])
+ if err != nil {
+ return 0, err
+ }
+ i -= size
+ i = protohelpers.EncodeVarint(dAtA, i, uint64(size))
+ i--
+ dAtA[i] = 0x7a
+ }
if m.KeepRemoteObject {
i--
if m.KeepRemoteObject {
@@ -1579,6 +1598,16 @@ func (m *ObjectTransactionRequest) MarshalToSizedBufferVT(dAtA []byte) (int, err
i -= len(m.unknownFields)
copy(dAtA[i:], m.unknownFields)
}
+ if m.ConditionPreferAlt {
+ i--
+ if m.ConditionPreferAlt {
+ dAtA[i] = 1
+ } else {
+ dAtA[i] = 0
+ }
+ i--
+ dAtA[i] = 0x58
+ }
if len(m.ConditionAltKey) > 0 {
i -= len(m.ConditionAltKey)
copy(dAtA[i:], m.ConditionAltKey)
@@ -1710,6 +1739,11 @@ func (m *ObjectTransactionResponse) MarshalToSizedBufferVT(dAtA []byte) (int, er
i -= len(m.unknownFields)
copy(dAtA[i:], m.unknownFields)
}
+ if m.FailedMutation != nil {
+ i = protohelpers.EncodeVarint(dAtA, i, uint64(*m.FailedMutation))
+ i--
+ dAtA[i] = 0x18
+ }
if m.ErrorCode != 0 {
i = protohelpers.EncodeVarint(dAtA, i, uint64(m.ErrorCode))
i--
@@ -7191,6 +7225,14 @@ func (m *ObjectMutation) SizeVT() (n int) {
if m.KeepRemoteObject {
n += 2
}
+ if m.Condition != nil {
+ l = m.Condition.SizeVT()
+ n += 1 + l + protohelpers.SizeOfVarint(uint64(l))
+ }
+ l = len(m.ConditionKey)
+ if l > 0 {
+ n += 2 + l + protohelpers.SizeOfVarint(uint64(l))
+ }
n += len(m.unknownFields)
return n
}
@@ -7297,6 +7339,9 @@ func (m *ObjectTransactionRequest) SizeVT() (n int) {
if l > 0 {
n += 1 + l + protohelpers.SizeOfVarint(uint64(l))
}
+ if m.ConditionPreferAlt {
+ n += 2
+ }
n += len(m.unknownFields)
return n
}
@@ -7314,6 +7359,9 @@ func (m *ObjectTransactionResponse) SizeVT() (n int) {
if m.ErrorCode != 0 {
n += 1 + protohelpers.SizeOfVarint(uint64(m.ErrorCode))
}
+ if m.FailedMutation != nil {
+ n += 1 + protohelpers.SizeOfVarint(uint64(*m.FailedMutation))
+ }
n += len(m.unknownFields)
return n
}
@@ -13244,6 +13292,74 @@ func (m *ObjectMutation) UnmarshalVT(dAtA []byte) error {
}
}
m.KeepRemoteObject = bool(v != 0)
+ case 15:
+ if wireType != 2 {
+ return fmt.Errorf("proto: wrong wireType = %d for field Condition", wireType)
+ }
+ var msglen int
+ for shift := uint(0); ; shift += 7 {
+ if shift >= 64 {
+ return protohelpers.ErrIntOverflow
+ }
+ if iNdEx >= l {
+ return io.ErrUnexpectedEOF
+ }
+ b := dAtA[iNdEx]
+ iNdEx++
+ msglen |= int(b&0x7F) << shift
+ if b < 0x80 {
+ break
+ }
+ }
+ if msglen < 0 {
+ return protohelpers.ErrInvalidLength
+ }
+ postIndex := iNdEx + msglen
+ if postIndex < 0 {
+ return protohelpers.ErrInvalidLength
+ }
+ if postIndex > l {
+ return io.ErrUnexpectedEOF
+ }
+ if m.Condition == nil {
+ m.Condition = &WriteCondition{}
+ }
+ if err := m.Condition.UnmarshalVT(dAtA[iNdEx:postIndex]); err != nil {
+ return err
+ }
+ iNdEx = postIndex
+ case 16:
+ if wireType != 2 {
+ return fmt.Errorf("proto: wrong wireType = %d for field ConditionKey", wireType)
+ }
+ var stringLen uint64
+ for shift := uint(0); ; shift += 7 {
+ if shift >= 64 {
+ return protohelpers.ErrIntOverflow
+ }
+ if iNdEx >= l {
+ return io.ErrUnexpectedEOF
+ }
+ b := dAtA[iNdEx]
+ iNdEx++
+ stringLen |= uint64(b&0x7F) << shift
+ if b < 0x80 {
+ break
+ }
+ }
+ intStringLen := int(stringLen)
+ if intStringLen < 0 {
+ return protohelpers.ErrInvalidLength
+ }
+ postIndex := iNdEx + intStringLen
+ if postIndex < 0 {
+ return protohelpers.ErrInvalidLength
+ }
+ if postIndex > l {
+ return io.ErrUnexpectedEOF
+ }
+ m.ConditionKey = string(dAtA[iNdEx:postIndex])
+ iNdEx = postIndex
default:
iNdEx = preIndex
skippy, err := protohelpers.Skip(dAtA[iNdEx:])
@@ -14097,6 +14213,26 @@ func (m *ObjectTransactionRequest) UnmarshalVT(dAtA []byte) error {
}
m.ConditionAltKey = string(dAtA[iNdEx:postIndex])
iNdEx = postIndex
+ case 11:
+ if wireType != 0 {
+ return fmt.Errorf("proto: wrong wireType = %d for field ConditionPreferAlt", wireType)
+ }
+ var v int
+ for shift := uint(0); ; shift += 7 {
+ if shift >= 64 {
+ return protohelpers.ErrIntOverflow
+ }
+ if iNdEx >= l {
+ return io.ErrUnexpectedEOF
+ }
+ b := dAtA[iNdEx]
+ iNdEx++
+ v |= int(b&0x7F) << shift
+ if b < 0x80 {
+ break
+ }
+ }
+ m.ConditionPreferAlt = bool(v != 0)
default:
iNdEx = preIndex
skippy, err := protohelpers.Skip(dAtA[iNdEx:])
@@ -14199,6 +14335,26 @@ func (m *ObjectTransactionResponse) UnmarshalVT(dAtA []byte) error {
break
}
}
+ case 3:
+ if wireType != 0 {
+ return fmt.Errorf("proto: wrong wireType = %d for field FailedMutation", wireType)
+ }
+ var v int32
+ for shift := uint(0); ; shift += 7 {
+ if shift >= 64 {
+ return protohelpers.ErrIntOverflow
+ }
+ if iNdEx >= l {
+ return io.ErrUnexpectedEOF
+ }
+ b := dAtA[iNdEx]
+ iNdEx++
+ v |= int32(b&0x7F) << shift
+ if b < 0x80 {
+ break
+ }
+ }
+ m.FailedMutation = &v
default:
iNdEx = preIndex
skippy, err := protohelpers.Skip(dAtA[iNdEx:])
diff --git a/weed/s3api/filer_multipart.go b/weed/s3api/filer_multipart.go
index 710b9db959..753246698a 100644
--- a/weed/s3api/filer_multipart.go
+++ b/weed/s3api/filer_multipart.go
@@ -689,6 +689,13 @@ func (s3a *S3ApiServer) completeMultipartUpload(r *http.Request, input *s3.Compl
// retries. With no owner yet (no ring), keep the lock as the bootstrap path.
owner := s3a.objectWriteOwner(*input.Bucket, *input.Key)
routeKey := s3a.objectRouteKey(*input.Bucket, *input.Key)
+ // The write condition rides the commit transaction so the owner filer
+ // evaluates it under the object's path lock — the gateway's earlier
+ // checkConditionalHeaders is only a fast-path. Conditions it cannot reduce
+ // (time-based, combined, ETag lists, weak ETags) keep the pre-change
+ // semantics: check and commit run under the distributed write lock, which
+ // routed writes still bypass.
+ cond, reducible := buildWriteCondition(r)
completionBody := func() s3err.ErrorCode {
var prepCode s3err.ErrorCode
completionState, output, prepCode = s3a.prepareMultipartCompletionState(r, input, parts, uploadDirectory, entryName, dirName, completedPartNumbers, completedPartMap, maxPartNo)
@@ -796,7 +803,7 @@ func (s3a *S3ApiServer) completeMultipartUpload(r *http.Request, input *s3.Compl
glog.Errorf("completeMultipartUpload %s upload %s unused part cleanup: %v", *input.Bucket, *input.UploadId, err)
return s3err.ErrInternalError
}
- if code := s3a.routedMultipartFinalize(owner, *input.Bucket, *input.Key, useInvertedFormat, versionDir, versionFileName, completionState.finalParts, decorateVersionEntry, *input.UploadId); code != s3err.ErrNone {
+ if code := s3a.routedMultipartFinalize(owner, *input.Bucket, *input.Key, useInvertedFormat, versionDir, versionFileName, completionState.finalParts, decorateVersionEntry, *input.UploadId, cond); code != s3err.ErrNone {
if code == s3err.ErrNoSuchUpload {
return code
}
@@ -850,6 +857,9 @@ func (s3a *S3ApiServer) completeMultipartUpload(r *http.Request, input *s3.Compl
glog.Errorf("completeMultipartUpload %s upload %s unused part cleanup: %v", *input.Bucket, *input.UploadId, err)
return s3err.ErrInternalError
}
+ // The condition checks the null slot first, then the latest version,
+ // same resolution the gateway's conditional check performs.
+ versionsDir := dirName + "/" + entryName + s3_constants.VersionsFolder
if err := s3a.writeMultipartObject(owner, routeKey, dirName, entryName, completionState.finalParts, func(entry *filer_pb.Entry) {
if entry.Extended == nil {
entry.Extended = make(map[string][]byte)
@@ -898,10 +908,13 @@ func (s3a *S3ApiServer) completeMultipartUpload(r *http.Request, input *s3.Compl
entry.Attributes.Mime = completionState.mime
}
entry.Attributes.FileSize = uint64(completionState.offset)
- }, removal); err != nil {
+ }, removal, cond, versionsDir); err != nil {
if errors.Is(err, errUploadRemoved) {
return s3err.ErrNoSuchUpload
}
+ if errors.Is(err, errObjectPrecondition) {
+ return s3err.ErrPreconditionFailed
+ }
// The transaction may have committed the object before failing on
// the upload removal; a surviving entry references the manifests.
if exists, err := s3a.exists(dirName, entryName, false); err != nil || exists {
@@ -991,10 +1004,13 @@ func (s3a *S3ApiServer) completeMultipartUpload(r *http.Request, input *s3.Compl
if completionState.entityWithTtl {
entry.Extended[s3_constants.SeaweedFSExpiresS3] = []byte("true")
}
- }, removal); err != nil {
+ }, removal, cond, ""); err != nil {
if errors.Is(err, errUploadRemoved) {
return s3err.ErrNoSuchUpload
}
+ if errors.Is(err, errObjectPrecondition) {
+ return s3err.ErrPreconditionFailed
+ }
// The transaction may have committed the object before failing on
// the upload removal; a surviving entry references the manifests.
if exists, err := s3a.exists(dirName, entryName, false); err != nil || exists {
@@ -1017,7 +1033,7 @@ func (s3a *S3ApiServer) completeMultipartUpload(r *http.Request, input *s3.Compl
return s3err.ErrNone
}
var finalizeCode s3err.ErrorCode
- if owner != "" {
+ if owner != "" && reducible {
if code := s3a.checkConditionalHeaders(r, *input.Bucket, *input.Key); code != s3err.ErrNone {
finalizeCode = code
} else {
diff --git a/weed/s3api/s3api_object_handlers_copy.go b/weed/s3api/s3api_object_handlers_copy.go
index 1b716126d9..2990eff785 100644
--- a/weed/s3api/s3api_object_handlers_copy.go
+++ b/weed/s3api/s3api_object_handlers_copy.go
@@ -742,7 +742,7 @@ func (s3a *S3ApiServer) finalizeCopyDestination(dstBucket, dstObject, dstVersion
if owner != "" {
err = s3a.routedMkFile(owner, s3a.objectRouteKey(dstBucket, normalizedObject), bucketDir+"/"+normalizedObject+s3_constants.VersionsFolder, versionFileName, dstEntry.Chunks, func(entry *filer_pb.Entry) {
copyEntryToTarget(entry, dstEntry)
- }, nil)
+ }, nil, nil, "")
} else {
err = s3a.mkFile(bucketDir, versionObjectPath, dstEntry.Chunks, func(entry *filer_pb.Entry) {
copyEntryToTarget(entry, dstEntry)
@@ -779,7 +779,7 @@ func (s3a *S3ApiServer) finalizeCopyDestination(dstBucket, dstObject, dstVersion
owner := s3a.objectWriteOwner(dstBucket, normalizedObject)
if err = s3a.writeMultipartObject(owner, s3a.objectRouteKey(dstBucket, normalizedObject), dstDir, dstName, dstEntry.Chunks, func(entry *filer_pb.Entry) {
copyEntryToTarget(entry, dstEntry)
- }, nil); err != nil {
+ }, nil, nil, ""); err != nil {
return "", "", err
}
@@ -799,7 +799,7 @@ func (s3a *S3ApiServer) finalizeCopyDestination(dstBucket, dstObject, dstVersion
// default filer is invisible to the next copy until replication.
if err = s3a.writeMultipartObject(s3a.objectWriteOwner(dstBucket, normalizedObject), s3a.objectRouteKey(dstBucket, normalizedObject), dstDir, dstName, dstEntry.Chunks, func(entry *filer_pb.Entry) {
copyEntryToTarget(entry, dstEntry)
- }, nil); err != nil {
+ }, nil, nil, ""); err != nil {
return "", "", err
}
diff --git a/weed/s3api/s3api_object_handlers_put.go b/weed/s3api/s3api_object_handlers_put.go
index abe4e98607..502c73e7a5 100644
--- a/weed/s3api/s3api_object_handlers_put.go
+++ b/weed/s3api/s3api_object_handlers_put.go
@@ -1084,7 +1084,7 @@ func (s3a *S3ApiServer) putToFiler(r *http.Request, filePath string, dataReader
lockKey, finalizeMutations = finalize.lockKey, finalize.mutations
}
routeKey := s3a.objectRouteKey(bucket, object)
- resp, err := s3a.routedPut(owner, routeKey, lockKey, filePath, entry, cond, "", condAltKey, finalizeMutations)
+ resp, err := s3a.routedPut(owner, routeKey, lockKey, filePath, entry, cond, "", condAltKey, false, nil, "", finalizeMutations)
if err == nil && resp.ErrorCode == filer_pb.FilerError_PRECONDITION_FAILED && condAltKey != "" {
// A filer that ignores condition_alt_key evaluated the empty
// null slot alone; resend the transaction against the entry the
@@ -1097,7 +1097,7 @@ func (s3a *S3ApiServer) putToFiler(r *http.Request, filePath string, dataReader
conditionKey = condAltKey + "/" + resolved.Name
}
glog.Warningf("putToFiler: routed PUT to %s precondition-failed for %s; resending against %s", owner, filePath, conditionKey)
- resp, err = s3a.routedPut(owner, routeKey, lockKey, filePath, entry, cond, conditionKey, condAltKey, finalizeMutations)
+ resp, err = s3a.routedPut(owner, routeKey, lockKey, filePath, entry, cond, conditionKey, condAltKey, false, nil, "", finalizeMutations)
}
switch {
case err != nil:
diff --git a/weed/s3api/s3api_object_routed_write.go b/weed/s3api/s3api_object_routed_write.go
index be70ed1fcb..5c5b5bee7d 100644
--- a/weed/s3api/s3api_object_routed_write.go
+++ b/weed/s3api/s3api_object_routed_write.go
@@ -216,28 +216,37 @@ func (s3a *S3ApiServer) withWriteOwnerFilerClient(owner pb.ServerAddress, fn fun
// object path plus a RECOMPUTE_LATEST finalize, so the version's PUT and its
// .versions pointer flip commit atomically (the recompute scans .versions/ after
// the PUT and sees the new version).
-func (s3a *S3ApiServer) routedPut(owner pb.ServerAddress, routeKey, lockKey, filePath string, entry *filer_pb.Entry, cond *filer_pb.WriteCondition, conditionKey, conditionAltKey string, finalize []*filer_pb.ObjectMutation) (*filer_pb.ObjectTransactionResponse, error) {
+// putCondition, when set, guards the PUT mutation itself against putConditionKey
+// (e.g. the upload directory a multipart commit must still find); the request
+// condition keeps guarding the object the transaction is about.
+func (s3a *S3ApiServer) routedPut(owner pb.ServerAddress, routeKey, lockKey, filePath string, entry *filer_pb.Entry, cond *filer_pb.WriteCondition, conditionKey, conditionAltKey string, conditionPreferAlt bool, putCondition *filer_pb.WriteCondition, putConditionKey string, finalize []*filer_pb.ObjectMutation) (*filer_pb.ObjectTransactionResponse, error) {
mutations := make([]*filer_pb.ObjectMutation, 0, 1+len(finalize))
mutations = append(mutations, &filer_pb.ObjectMutation{
- Type: filer_pb.ObjectMutation_PUT,
- Directory: path.Dir(filePath),
- Entry: entry,
+ Type: filer_pb.ObjectMutation_PUT,
+ Directory: path.Dir(filePath),
+ Entry: entry,
+ Condition: putCondition,
+ ConditionKey: putConditionKey,
})
mutations = append(mutations, finalize...)
return s3a.objectTxnOnFiler(owner, &filer_pb.ObjectTransactionRequest{
- LockKey: lockKey,
- RouteKey: routeKey,
- Condition: cond,
- ConditionKey: conditionKey,
- ConditionAltKey: conditionAltKey,
- Mutations: mutations,
+ LockKey: lockKey,
+ RouteKey: routeKey,
+ Condition: cond,
+ ConditionKey: conditionKey,
+ ConditionAltKey: conditionAltKey,
+ ConditionPreferAlt: conditionPreferAlt,
+ Mutations: mutations,
})
}
// routedMkFile builds an entry like filer_pb.MkFile and writes it through a
// routed PUT on the owner filer, for callers that would otherwise mkFile to the
-// default filer (e.g. multipart completion of a non-versioned object).
-func (s3a *S3ApiServer) routedMkFile(owner pb.ServerAddress, routeKey, parentDir, name string, chunks []*filer_pb.FileChunk, fn func(*filer_pb.Entry), removal *uploadRemovalTxn) error {
+// default filer (e.g. multipart completion of a non-versioned object). cond is
+// the request's write condition, evaluated on the object under the transaction
+// lock; condAltKey falls back to the .versions directory when the object slot
+// is empty (suspended or versioned buckets).
+func (s3a *S3ApiServer) routedMkFile(owner pb.ServerAddress, routeKey, parentDir, name string, chunks []*filer_pb.FileChunk, fn func(*filer_pb.Entry), removal *uploadRemovalTxn, cond *filer_pb.WriteCondition, condAltKey string) error {
now := time.Now().Unix()
entry := &filer_pb.Entry{
Name: name,
@@ -254,14 +263,14 @@ func (s3a *S3ApiServer) routedMkFile(owner pb.ServerAddress, routeKey, parentDir
fn(entry)
}
filePath := parentDir + "/" + name
- var cond *filer_pb.WriteCondition
- var conditionKey string
+ var putCondition *filer_pb.WriteCondition
+ var putConditionKey string
var finalize []*filer_pb.ObjectMutation
if removal != nil {
- cond, conditionKey = removal.condition, removal.conditionKey
+ putCondition, putConditionKey = removal.putCondition, removal.putConditionKey
finalize = []*filer_pb.ObjectMutation{removal.mutation}
}
- resp, err := s3a.routedPut(owner, routeKey, filePath, filePath, entry, cond, conditionKey, "", finalize)
+ resp, err := s3a.routedPut(owner, routeKey, filePath, filePath, entry, cond, "", condAltKey, false, putCondition, putConditionKey, finalize)
if err != nil {
return err
}
@@ -269,12 +278,15 @@ func (s3a *S3ApiServer) routedMkFile(owner pb.ServerAddress, routeKey, parentDir
// Other keys are nested under this one, so the object goes onto the
// directory they live in — the same retry mkFile makes off the ring.
entry.MarkPrefixObject()
- resp, err = s3a.routedPut(owner, routeKey, filePath, filePath, entry, cond, conditionKey, "", finalize)
+ resp, err = s3a.routedPut(owner, routeKey, filePath, filePath, entry, cond, "", condAltKey, false, putCondition, putConditionKey, finalize)
if err != nil {
return err
}
}
if resp.ErrorCode == filer_pb.FilerError_PRECONDITION_FAILED {
+ if resp.FailedMutation == nil {
+ return errObjectPrecondition
+ }
return errUploadRemoved
}
if resp.Error != "" {
@@ -283,32 +295,40 @@ func (s3a *S3ApiServer) routedMkFile(owner pb.ServerAddress, routeKey, parentDir
return nil
}
-// errUploadRemoved is a routed commit rejected because its precondition found
-// the upload directory already deleted, mapped to NoSuchUpload by callers.
+// errUploadRemoved is a routed commit rejected because the upload directory's
+// precondition found it already deleted, mapped to NoSuchUpload by callers.
var errUploadRemoved = errors.New("upload directory already removed")
+// errObjectPrecondition is a routed commit rejected because the request's write
+// condition (e.g. If-None-Match: *) found the object slot already taken by a
+// racing commit, mapped to PreconditionFailed by callers.
+var errObjectPrecondition = errors.New("object precondition failed")
+
// writeMultipartObject writes a completed multipart object entry, routed to the
// owner when known (so it serializes with concurrent writes to the same key)
// and falling back to a plain mkFile otherwise. removal carries the upload
// precondition and deletion applied in the same transaction as the routed PUT.
// routeKey must be the same key the caller used to resolve owner, so owner
-// selection and forwarding stay consistent.
-func (s3a *S3ApiServer) writeMultipartObject(owner pb.ServerAddress, routeKey, dir, name string, chunks []*filer_pb.FileChunk, fn func(*filer_pb.Entry), removal *uploadRemovalTxn) error {
+// selection and forwarding stay consistent. cond is the request's write
+// condition, enforced on the owner under the transaction lock; condAltKey is
+// the .versions directory a conditional check falls back to when the bucket
+// tracks versions.
+func (s3a *S3ApiServer) writeMultipartObject(owner pb.ServerAddress, routeKey, dir, name string, chunks []*filer_pb.FileChunk, fn func(*filer_pb.Entry), removal *uploadRemovalTxn, cond *filer_pb.WriteCondition, condAltKey string) error {
if owner != "" {
- return s3a.routedMkFile(owner, routeKey, dir, name, chunks, fn, removal)
+ return s3a.routedMkFile(owner, routeKey, dir, name, chunks, fn, removal, cond, condAltKey)
}
return s3a.mkFile(dir, name, chunks, fn)
}
-// uploadRemovalTxn carries the precondition and mutation that drop a completed
-// upload's directory inside the object's commit transaction. The precondition
-// fails the commit when a delete that does not take the object lock (abort,
-// lifecycle, s3.clean.uploads) removed the directory first, instead of
-// publishing the object over freed chunks.
+// uploadRemovalTxn carries the PUT precondition and mutation that drop a
+// completed upload's directory inside the object's commit transaction. The
+// precondition fails the commit when a delete that does not take the object
+// lock (abort, lifecycle, s3.clean.uploads) removed the directory first,
+// instead of publishing the object over freed chunks.
type uploadRemovalTxn struct {
- condition *filer_pb.WriteCondition
- conditionKey string
- mutation *filer_pb.ObjectMutation
+ putCondition *filer_pb.WriteCondition
+ putConditionKey string
+ mutation *filer_pb.ObjectMutation
}
// routedUploadRemoval returns the transaction parts that remove a completed
@@ -319,14 +339,20 @@ func (s3a *S3ApiServer) routedUploadRemoval(ctx context.Context, owner pb.Server
if owner == "" {
return nil, nil
}
+ // Older filers ignore the mutation's upload-exists condition, so during an
+ // upgrade this gateway-side check keeps the same guard window the
+ // pre-transaction code had.
+ if exists, err := s3a.exists(s3a.genUploadsFolder(bucket), uploadId, true); err == nil && !exists {
+ return nil, errUploadRemoved
+ }
if err := s3a.deleteUnusedPartEntries(ctx, uploadDirectory, bucket, uploadId, completionState); err != nil {
return nil, err
}
conditionKey, condition := uploadExistsCondition(uploadDirectory)
return &uploadRemovalTxn{
- condition: condition,
- conditionKey: conditionKey,
- mutation: s3a.removeUploadDirMutation(bucket, uploadId),
+ putCondition: condition,
+ putConditionKey: conditionKey,
+ mutation: s3a.removeUploadDirMutation(bucket, uploadId),
}, nil
}
diff --git a/weed/s3api/s3api_object_routed_write_test.go b/weed/s3api/s3api_object_routed_write_test.go
index 0f23779d72..19f202c865 100644
--- a/weed/s3api/s3api_object_routed_write_test.go
+++ b/weed/s3api/s3api_object_routed_write_test.go
@@ -514,7 +514,7 @@ func TestRoutedMultipartFinalize(t *testing.T) {
chunks := []*filer_pb.FileChunk{{FileId: "1,01637037d6"}}
code := s3a.routedMultipartFinalize(owner, "b", "obj", false, "/buckets/b/obj.versions", "v_1", chunks, func(entry *filer_pb.Entry) {
entry.Extended = map[string][]byte{s3_constants.SeaweedFSUploadId: []byte("up1")}
- }, "up1")
+ }, "up1", clause(filer_pb.WriteCondition_IF_NOT_EXISTS))
if code != s3err.ErrNone {
t.Fatalf("routedMultipartFinalize = %v", code)
}
@@ -529,10 +529,15 @@ func TestRoutedMultipartFinalize(t *testing.T) {
if req.RouteKey != "s3.object.write:/buckets/b/obj" {
t.Fatalf("RouteKey = %q", req.RouteKey)
}
- if req.ConditionKey != "/buckets/b/.uploads/up1" {
+ // The request condition guards the object (at lock_key, falling back to
+ // .versions); the upload directory rides the PUT as a mutation guard.
+ if req.ConditionKey != "" {
t.Fatalf("ConditionKey = %q", req.ConditionKey)
}
- if req.Condition == nil || len(req.Condition.Clauses) != 1 || req.Condition.Clauses[0].Kind != filer_pb.WriteCondition_IF_EXISTS {
+ if req.ConditionAltKey != "/buckets/b/obj.versions" {
+ t.Fatalf("ConditionAltKey = %q", req.ConditionAltKey)
+ }
+ if req.Condition == nil || len(req.Condition.Clauses) != 1 || req.Condition.Clauses[0].Kind != filer_pb.WriteCondition_IF_NOT_EXISTS {
t.Fatalf("Condition = %+v", req.Condition)
}
if len(req.Mutations) != 3 {
@@ -548,6 +553,10 @@ func TestRoutedMultipartFinalize(t *testing.T) {
string(put.Entry.Extended[s3_constants.SeaweedFSUploadId]) != "up1" {
t.Fatalf("put mutation = %+v", put)
}
+ if put.ConditionKey != "/buckets/b/.uploads/up1" ||
+ put.Condition == nil || len(put.Condition.Clauses) != 1 || put.Condition.Clauses[0].Kind != filer_pb.WriteCondition_IF_EXISTS {
+ t.Fatalf("put condition = %+v key %q", put.Condition, put.ConditionKey)
+ }
removeUpload := req.Mutations[1]
if removeUpload.Type != filer_pb.ObjectMutation_DELETE ||
removeUpload.Directory != "/buckets/b/.uploads" ||
@@ -580,7 +589,7 @@ func TestWriteMultipartObjectRemovesUploadDir(t *testing.T) {
if err != nil {
t.Fatalf("routedUploadRemoval: %v", err)
}
- if err := s3a.writeMultipartObject(owner, "s3.object.write:/buckets/b/o", "/buckets/b", "o", nil, nil, removal); err != nil {
+ if err := s3a.writeMultipartObject(owner, "s3.object.write:/buckets/b/o", "/buckets/b", "o", nil, nil, removal, nil, ""); err != nil {
t.Fatalf("writeMultipartObject: %v", err)
}
@@ -591,12 +600,6 @@ func TestWriteMultipartObjectRemovesUploadDir(t *testing.T) {
if req.LockKey != "/buckets/b/o" {
t.Fatalf("LockKey = %q", req.LockKey)
}
- if req.ConditionKey != "/buckets/b/.uploads/up1" {
- t.Fatalf("ConditionKey = %q", req.ConditionKey)
- }
- if req.Condition == nil || len(req.Condition.Clauses) != 1 || req.Condition.Clauses[0].Kind != filer_pb.WriteCondition_IF_EXISTS {
- t.Fatalf("Condition = %+v", req.Condition)
- }
if len(req.Mutations) != 2 {
t.Fatalf("mutations = %d, want 2", len(req.Mutations))
}
@@ -604,6 +607,10 @@ func TestWriteMultipartObjectRemovesUploadDir(t *testing.T) {
if put.Type != filer_pb.ObjectMutation_PUT || put.Directory != "/buckets/b" || put.Entry == nil || put.Entry.Name != "o" {
t.Fatalf("put mutation = %+v", put)
}
+ if put.ConditionKey != "/buckets/b/.uploads/up1" ||
+ put.Condition == nil || len(put.Condition.Clauses) != 1 || put.Condition.Clauses[0].Kind != filer_pb.WriteCondition_IF_EXISTS {
+ t.Fatalf("put condition = %+v key %q", put.Condition, put.ConditionKey)
+ }
removeUpload := req.Mutations[1]
if removeUpload.Type != filer_pb.ObjectMutation_DELETE ||
removeUpload.Directory != "/buckets/b/.uploads" ||
diff --git a/weed/s3api/s3api_object_versioned_finalize.go b/weed/s3api/s3api_object_versioned_finalize.go
index 4f470a71de..c9ec86251a 100644
--- a/weed/s3api/s3api_object_versioned_finalize.go
+++ b/weed/s3api/s3api_object_versioned_finalize.go
@@ -116,8 +116,10 @@ func uploadExistsCondition(uploadDirectory string) (string, *filer_pb.WriteCondi
// reachable: PUT first keeps the chunks referenced at all times, and removing
// the upload directory before the recompute means a published object never
// coexists with a leftover .uploads directory that s3.clean.uploads would purge
-// with data.
-func (s3a *S3ApiServer) routedMultipartFinalize(owner pb.ServerAddress, bucket, object string, useInvertedFormat bool, versionDir, versionFileName string, chunks []*filer_pb.FileChunk, decorate func(*filer_pb.Entry), uploadID string) s3err.ErrorCode {
+// with data. cond is the request's write condition, checked against the object
+// (its latest version when the slot at the object path is empty) under the same
+// lock; the upload directory rides the PUT as a mutation precondition.
+func (s3a *S3ApiServer) routedMultipartFinalize(owner pb.ServerAddress, bucket, object string, useInvertedFormat bool, versionDir, versionFileName string, chunks []*filer_pb.FileChunk, decorate func(*filer_pb.Entry), uploadID string, cond *filer_pb.WriteCondition) s3err.ErrorCode {
now := time.Now().Unix()
versionEntry := &filer_pb.Entry{
Name: versionFileName,
@@ -138,8 +140,14 @@ func (s3a *S3ApiServer) routedMultipartFinalize(owner pb.ServerAddress, bucket,
if owner != "" {
routeKey = s3a.objectRouteKey(bucket, object)
}
- conditionKey, condition := uploadExistsCondition(s3a.genUploadsFolder(bucket) + "/" + uploadID)
- resp, err := s3a.routedPut(owner, routeKey, s3a.toFilerPath(bucket, object), versionDir+"/"+versionFileName, versionEntry, condition, conditionKey, "", []*filer_pb.ObjectMutation{
+ // A filer older than the mutation-condition support silently ignores
+ // putCondition, so on upgrade the guard degrades to this gateway-side
+ // check — the same window the pre-transaction code had.
+ if exists, err := s3a.exists(s3a.genUploadsFolder(bucket), uploadID, true); err == nil && !exists {
+ return s3err.ErrNoSuchUpload
+ }
+ putConditionKey, putCondition := uploadExistsCondition(s3a.genUploadsFolder(bucket) + "/" + uploadID)
+ resp, err := s3a.routedPut(owner, routeKey, s3a.toFilerPath(bucket, object), versionDir+"/"+versionFileName, versionEntry, cond, "", versionDir, true, putCondition, putConditionKey, []*filer_pb.ObjectMutation{
s3a.removeUploadDirMutation(bucket, uploadID),
s3a.latestPointerRecompute(bucket, object, useInvertedFormat, "", true),
})
@@ -148,6 +156,9 @@ func (s3a *S3ApiServer) routedMultipartFinalize(owner pb.ServerAddress, bucket,
glog.Errorf("routedMultipartFinalize: %s/%s upload %s on %s: %v", bucket, object, uploadID, owner, err)
return s3err.ErrInternalError
case resp.ErrorCode == filer_pb.FilerError_PRECONDITION_FAILED:
+ if resp.FailedMutation == nil {
+ return s3err.ErrPreconditionFailed
+ }
return s3err.ErrNoSuchUpload
case resp.Error != "":
glog.Errorf("routedMultipartFinalize: %s/%s upload %s: %s", bucket, object, uploadID, resp.Error)
diff --git a/weed/server/filer_grpc_server.go b/weed/server/filer_grpc_server.go
index 21f823410b..498bdd624c 100644
--- a/weed/server/filer_grpc_server.go
+++ b/weed/server/filer_grpc_server.go
@@ -353,6 +353,7 @@ func (fs *FilerServer) ObjectTransaction(ctx context.Context, req *filer_pb.Obje
IsMoved: true,
RequestId: req.RequestId,
ConditionAltKey: req.ConditionAltKey,
+ ConditionPreferAlt: req.ConditionPreferAlt,
}
var resp *filer_pb.ObjectTransactionResponse
handled, err := fs.forwardToWriteOwner(ctx, req.RouteKey, func(owner pb.ServerAddress) error {
@@ -407,11 +408,16 @@ func (fs *FilerServer) ObjectTransaction(ctx context.Context, req *filer_pb.Obje
}
current, findErr := fs.filer.FindEntry(ctx, conditionPath)
if findErr == nil && current != nil && current.IsDirectory() && req.ConditionAltKey != "" {
- // A directory is not the object: implicit directories can share a
- // path with an S3 key, so treat it as an absent slot. Only the
- // alt-key callers want object-slot semantics; an explicit
- // condition_key may legitimately name a directory.
- current, findErr = nil, filer_pb.ErrNotFound
+ // A bare directory is not the object: implicit directories can
+ // share a path with an S3 key, so treat it as an absent slot. A
+ // directory entry holding content or the prefix-object marker is
+ // an object and stays. Only the alt-key callers want object-slot
+ // semantics; an explicit condition_key may legitimately name a
+ // directory.
+ _, isPrefixObject := current.Extended[s3_constants.SeaweedFSPrefixObject]
+ if current.Mime == "" && len(current.GetChunks()) == 0 && len(current.Content) == 0 && !current.IsInRemoteOnly() && !isPrefixObject {
+ current, findErr = nil, filer_pb.ErrNotFound
+ }
}
// The caller can name a fallback evaluated when the primary is absent —
// a suspended-versioning write checks "the null slot if present, else
@@ -419,8 +425,24 @@ func (fs *FilerServer) ObjectTransaction(ctx context.Context, req *filer_pb.Obje
// the object's .versions directory, and the latest it resolves to is
// read from the pointer maintained by the same lock, so a concurrent
// version delete cannot move which entry the condition sees.
- if findErr == filer_pb.ErrNotFound && req.ConditionAltKey != "" {
- current, conditionPath, findErr = fs.latestVersionedConditionEntry(ctx, util.FullPath(req.ConditionAltKey))
+ if req.ConditionAltKey != "" {
+ altEntry, altPath, altErr := fs.latestVersionedConditionEntry(ctx, util.FullPath(req.ConditionAltKey))
+ if req.ConditionPreferAlt {
+ // The versioned path resolves "current" through the latest
+ // pointer, so a stale null-version entry or a tombstoned
+ // latest decides correctly. When the alt path cannot resolve
+ // (no .versions at all), the primary entry stands.
+ switch {
+ case altErr == nil:
+ current, conditionPath, findErr = altEntry, altPath, nil
+ case errors.Is(altErr, filer_pb.ErrNotFound):
+ default:
+ record = false
+ return nil, fmt.Errorf("ObjectTransaction condition %s: %w", conditionPath, altErr)
+ }
+ } else if findErr == filer_pb.ErrNotFound {
+ current, conditionPath, findErr = altEntry, altPath, altErr
+ }
}
if findErr != nil && findErr != filer_pb.ErrNotFound {
record = false // a lookup error decides nothing; a retry must re-run, not replay
@@ -438,10 +460,26 @@ func (fs *FilerServer) ObjectTransaction(ctx context.Context, req *filer_pb.Obje
}
for i, m := range req.Mutations {
+ if code, err := fs.checkMutationCondition(ctx, m); err != nil {
+ glog.V(2).InfofCtx(ctx, "ObjectTransaction %s mutation %d condition lookup: %v", lockPath, i, err)
+ resp.Error = fmt.Sprintf("mutation %d: %v", i, err)
+ failed := int32(i)
+ resp.FailedMutation = &failed
+ return resp, nil
+ } else if code != filer_pb.FilerError_OK {
+ glog.V(3).InfofCtx(ctx, "ObjectTransaction %s mutation %d precondition failed", lockPath, i)
+ resp.Error = fmt.Sprintf("mutation %d: precondition failed", i)
+ resp.ErrorCode = code
+ failed := int32(i)
+ resp.FailedMutation = &failed
+ return resp, nil
+ }
if err := fs.applyObjectMutation(ctx, m, req.IsFromOtherCluster, req.Signatures); err != nil {
glog.V(2).InfofCtx(ctx, "ObjectTransaction %s mutation %d (%v): %v", lockPath, i, m.Type, err)
resp.Error = fmt.Sprintf("mutation %d: %v", i, err)
resp.ErrorCode = filerErrorCode(err)
+ failed := int32(i)
+ resp.FailedMutation = &failed
return resp, nil
}
}
@@ -449,6 +487,36 @@ func (fs *FilerServer) ObjectTransaction(ctx context.Context, req *filer_pb.Obje
return resp, nil
}
+// checkMutationCondition evaluates a mutation's optional precondition under the
+// transaction's lock, against condition_key when set or the mutation's target
+// (the PUT's entry path, else directory/name). Unlike the request-level
+// condition there is no alt-key fallback: a mutation guard names the exact entry
+// it guards.
+func (fs *FilerServer) checkMutationCondition(ctx context.Context, m *filer_pb.ObjectMutation) (filer_pb.FilerError, error) {
+ if !conditionIsSet(m.Condition) {
+ return filer_pb.FilerError_OK, nil
+ }
+ target := util.FullPath(m.ConditionKey)
+ if target == "" {
+ name := m.Name
+ if m.Type == filer_pb.ObjectMutation_PUT && m.Entry != nil {
+ name = m.Entry.Name
+ }
+ target = util.JoinPath(m.Directory, name)
+ }
+ current, err := fs.filer.FindEntry(ctx, target)
+ if err != nil && err != filer_pb.ErrNotFound {
+ return filer_pb.FilerError_OK, fmt.Errorf("condition check %s: %w", target, err)
+ }
+ if err == filer_pb.ErrNotFound {
+ current = nil
+ }
+ if !writeConditionSatisfied(m.Condition, current) {
+ return filer_pb.FilerError_PRECONDITION_FAILED, nil
+ }
+ return filer_pb.FilerError_OK, nil
+}
+
const (
recentObjectTxnTTL = 2 * time.Minute
recentObjectTxnSweepCap = 10000
@@ -525,9 +593,11 @@ func (fs *FilerServer) rememberObjectTxn(requestID string, lockPath util.FullPat
// latestVersionedConditionEntry resolves the object a conditional write should
// be checked against inside a .versions directory: the entry the latest-version
// pointer names, or nothing when the pointer names a delete marker — a
-// tombstone is not a current version. A missing pointer or a pointer to a file
-// that is gone means the metadata cannot answer the question, which reports an
-// error rather than guessing at either branch.
+// tombstone is not a current version. A suspended-versioning write clears the
+// pointer and stamps null-version-is-latest instead, which reports not-found so
+// the regular path's null object decides. Any other missing pointer, or a
+// pointer to a file that is gone, means the metadata cannot answer the
+// question, which reports an error rather than guessing at either branch.
func (fs *FilerServer) latestVersionedConditionEntry(ctx context.Context, versionsDir util.FullPath) (*filer.Entry, util.FullPath, error) {
versionsEntry, err := fs.filer.FindEntry(ctx, versionsDir)
if err != nil {
@@ -538,6 +608,9 @@ func (fs *FilerServer) latestVersionedConditionEntry(ctx context.Context, versio
}
latestName := string(versionsEntry.Extended[s3_constants.ExtLatestVersionFileNameKey])
if latestName == "" {
+ if string(versionsEntry.Extended[s3_constants.ExtNullVersionIsLatestKey]) == "true" {
+ return nil, versionsDir, filer_pb.ErrNotFound
+ }
return nil, versionsDir, fmt.Errorf("%s has no latest-version pointer", versionsDir)
}
latestPath := util.NewFullPath(string(versionsDir), latestName)
@@ -549,7 +622,7 @@ func (fs *FilerServer) latestVersionedConditionEntry(ctx context.Context, versio
return nil, latestPath, err
}
if string(entry.Extended[s3_constants.ExtDeleteMarkerKey]) == "true" {
- return nil, latestPath, filer_pb.ErrNotFound
+ return nil, latestPath, nil
}
return entry, latestPath, nil
}
diff --git a/weed/server/filer_grpc_server_object_txn_test.go b/weed/server/filer_grpc_server_object_txn_test.go
index c73badd7a3..b7fd0b5117 100644
--- a/weed/server/filer_grpc_server_object_txn_test.go
+++ b/weed/server/filer_grpc_server_object_txn_test.go
@@ -229,6 +229,90 @@ func TestObjectTransactionPreconditionAborts(t *testing.T) {
}
}
+// A mutation's own precondition is checked under the same lock just before the
+// mutation applies; a failure aborts the transaction at that mutation and is
+// reported through failed_mutation, leaving earlier mutations' work visible.
+func TestObjectTransactionMutationCondition(t *testing.T) {
+ now := time.Unix(1700000000, 0)
+ fs, store := newTxnTestServer(t, map[string]*filer.Entry{
+ "/buckets/bkt/obj": {
+ Attr: filer.Attr{Inode: 1, Mtime: now, Crtime: now, Mode: 0644},
+ },
+ })
+
+ mkReq := func() *filer_pb.ObjectTransactionRequest {
+ return &filer_pb.ObjectTransactionRequest{
+ LockKey: "/buckets/bkt/obj",
+ Mutations: []*filer_pb.ObjectMutation{
+ {Type: filer_pb.ObjectMutation_PUT, Directory: "/buckets/bkt", Entry: &filer_pb.Entry{
+ Name: "other",
+ Attributes: &filer_pb.FuseAttributes{Mtime: now.Unix(), FileMode: 0644, Inode: 2},
+ }, Condition: &filer_pb.WriteCondition{Clauses: []*filer_pb.WriteCondition_Clause{
+ {Kind: filer_pb.WriteCondition_IF_NOT_EXISTS},
+ }}, ConditionKey: "/buckets/bkt/obj"},
+ {Type: filer_pb.ObjectMutation_DELETE, Directory: "/buckets/bkt", Name: "obj"},
+ },
+ }
+ }
+
+ resp, err := fs.ObjectTransaction(context.Background(), mkReq())
+ if err != nil {
+ t.Fatalf("unexpected err: %v", err)
+ }
+ if resp.ErrorCode != filer_pb.FilerError_PRECONDITION_FAILED {
+ t.Fatalf("want PRECONDITION_FAILED, got %v (%q)", resp.ErrorCode, resp.Error)
+ }
+ if resp.FailedMutation == nil || *resp.FailedMutation != 0 {
+ t.Fatalf("want failed_mutation=0, got %v", resp.FailedMutation)
+ }
+ if _, ok := store.entries["/buckets/bkt/other"]; ok {
+ t.Errorf("PUT whose condition failed must not store an entry")
+ }
+ if _, ok := store.entries["/buckets/bkt/obj"]; !ok {
+ t.Errorf("a failed mutation condition must abort before later mutations run")
+ }
+
+ // With the slot absent the same transaction applies both mutations.
+ delete(store.entries, "/buckets/bkt/obj")
+ resp, err = fs.ObjectTransaction(context.Background(), mkReq())
+ if err != nil || resp.Error != "" {
+ t.Fatalf("txn failed: err=%v resp=%q", err, resp.Error)
+ }
+ if _, ok := store.entries["/buckets/bkt/other"]; !ok {
+ t.Errorf("condition satisfied: PUT entry missing")
+ }
+}
+
+// The request-level condition keeps reporting no failed_mutation, so callers
+// can tell it apart from a mutation's own guard.
+func TestObjectTransactionMutationConditionVsRequest(t *testing.T) {
+ now := time.Unix(1700000000, 0)
+ fs, _ := newTxnTestServer(t, map[string]*filer.Entry{
+ "/buckets/bkt/obj": {
+ Attr: filer.Attr{Inode: 1, Mtime: now, Crtime: now, Mode: 0644},
+ },
+ })
+
+ resp, err := fs.ObjectTransaction(context.Background(), &filer_pb.ObjectTransactionRequest{
+ LockKey: "/buckets/bkt/obj",
+ Condition: &filer_pb.WriteCondition{Clauses: []*filer_pb.WriteCondition_Clause{
+ {Kind: filer_pb.WriteCondition_IF_NOT_EXISTS},
+ }},
+ Mutations: []*filer_pb.ObjectMutation{
+ {Type: filer_pb.ObjectMutation_DELETE, Directory: "/buckets/bkt", Name: "obj"},
+ },
+ })
+ if err != nil {
+ t.Fatalf("unexpected err: %v", err)
+ }
+ if resp.ErrorCode != filer_pb.FilerError_PRECONDITION_FAILED {
+ t.Fatalf("want PRECONDITION_FAILED, got %v (%q)", resp.ErrorCode, resp.Error)
+ }
+ if resp.FailedMutation != nil {
+ t.Fatalf("request-level failure must not set failed_mutation, got %d", *resp.FailedMutation)
+ }
+}
+
// Deleting the latest version and recomputing re-points the pointer at the new
// highest-named remaining version; the scan runs under the transaction lock.
func TestObjectTransactionRecomputeLatest(t *testing.T) {
@@ -1095,3 +1179,42 @@ func TestObjectTransactionConditionIgnoresDirectory(t *testing.T) {
t.Fatalf("directory at key with live latest: err=%v code=%v, want precondition failed", err, resp.GetErrorCode())
}
}
+
+// A suspended-versioning write clears the latest-version pointer and leaves
+// the null object at the regular path as current. The prefer-alt condition
+// must keep resolving against that object, not reject on the empty pointer.
+func TestObjectTransactionConditionPreferAltHonorsNullIsLatest(t *testing.T) {
+ now := time.Unix(1700000000, 0)
+ seed := map[string]*filer.Entry{
+ "/buckets/bkt/obj": {
+ Attr: filer.Attr{Inode: 1, Mtime: now, Crtime: now, Mode: 0644},
+ Extended: map[string][]byte{s3_constants.ExtETagKey: []byte("nulletag")},
+ },
+ "/buckets/bkt/obj/.versions": {
+ Attr: filer.Attr{Inode: 2, Mtime: now, Crtime: now, Mode: 0755},
+ Extended: map[string][]byte{s3_constants.ExtNullVersionIsLatestKey: []byte("true")},
+ },
+ }
+ req := &filer_pb.ObjectTransactionRequest{
+ LockKey: "/buckets/bkt/obj",
+ ConditionAltKey: "/buckets/bkt/obj/.versions",
+ ConditionPreferAlt: true,
+ Condition: &filer_pb.WriteCondition{
+ Clauses: []*filer_pb.WriteCondition_Clause{{Kind: filer_pb.WriteCondition_IF_ETAG_MATCH, Etags: []string{`"nulletag"`}}},
+ },
+ Mutations: []*filer_pb.ObjectMutation{{
+ Type: filer_pb.ObjectMutation_PUT,
+ Directory: "/buckets/bkt",
+ Entry: &filer_pb.Entry{Name: "obj", Attributes: &filer_pb.FuseAttributes{Mtime: now.Unix(), FileMode: 0644, Inode: 9}},
+ }},
+ }
+
+ fs, store := newTxnTestServer(t, seed)
+ resp, err := fs.ObjectTransaction(context.Background(), req)
+ if err != nil || resp.Error != "" {
+ t.Fatalf("null-is-latest: err=%v resp=%q, want applied", err, resp.Error)
+ }
+ if store.entries["/buckets/bkt/obj"] == nil {
+ t.Fatal("mutation did not apply over a null-is-latest .versions")
+ }
+}