Files
seaweedfs/weed/s3api/iceberg/handlers_view.go
T
Chris LuandClaude Opus 5.5 4f24b756ba s3tables, iceberg: give tables created without a location their own directory (#12475)
* s3tables: give a table created without a location its own directory

An Iceberg table created without a location kept its files under its name
path, so a rename left them where the next table of that name would land.
Such a table now gets s3://<bucket>/<namespace>/<name>-<uuid>, stored as the
record's warehouseLocation and returned by GetTable; its first commit writes
there without moving the root. Records without one keep resolving to their
name path, so existing tables need no rewrite. Lance datasets stay at their
name path, where Lance clients and the vector engine look for them.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* iceberg: create tables and views in a directory of their own

A table or view created without a location took its name path, or a
unique one only when that path was already taken. It now always gets
<name>-<uuid>, so a later rename never leaves its files under a name
another table can take.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* iceberg: load and commit a table at its warehouse location

A table created through S3 Tables without a location has no metadata until
its first commit; loading or committing it now uses the directory recorded
for it, and only a record without one falls back to its name path.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* iceberg: fail a stage-create whose marker is not saved

The marker is how the commit finds the location the stage-create handed
out. Without it the commit fell back to the name path while the client's
files sat under the returned location.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* test: expect tables created without a location in their own directory

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* test: rename a table that keeps its files elsewhere before reusing its name

Enterprise keeps the storage marker on a renamed table's files, so the name
of a table whose files sit under its name path stays taken there.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* s3tables: remove the table directory when a create backs out

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* iceberg: answer a create that lost the race with the registered table

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* iceberg: answer a view create that lost the race with the registered view

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* iceberg: fail a staged commit whose name another create registered

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* iceberg: remove the directory of a create whose name another table took

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* iceberg: keep the metadata of a staged commit whose name is taken

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* test: write the lakekeeper table's data under its warehouse location

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* iceberg: commit a staged table at the location of its own stage

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* iceberg: never continue another stage when a commit's own marker is gone

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* iceberg: test filer honors exclusive creates and token-guarded deletes

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-07 16:37:31 -08:00

467 lines
16 KiB
Go

package iceberg
import (
"encoding/json"
"errors"
"fmt"
"net/http"
"os"
"path"
"strings"
"github.com/apache/iceberg-go"
"github.com/apache/iceberg-go/view"
"github.com/google/uuid"
"github.com/gorilla/mux"
"github.com/seaweedfs/seaweedfs/weed/glog"
"github.com/seaweedfs/seaweedfs/weed/pb/filer_pb"
"github.com/seaweedfs/seaweedfs/weed/s3api/s3_constants"
"github.com/seaweedfs/seaweedfs/weed/s3api/s3tables"
)
// handleListViews lists views in a namespace.
func (s *Server) handleListViews(w http.ResponseWriter, r *http.Request) {
vars := mux.Vars(r)
namespace := parseNamespace(vars["namespace"])
if len(namespace) == 0 {
writeError(w, http.StatusBadRequest, "BadRequestException", "Namespace is required")
return
}
bucketName := getBucketFromPrefix(r)
bucketARN := buildTableBucketARN(bucketName)
identityName := s3_constants.GetIdentityNameFromContext(r)
pageToken, pageSize, err := parsePagination(r)
if err != nil {
writeError(w, http.StatusBadRequest, "BadRequestException", err.Error())
return
}
listReq := &s3tables.ListViewsRequest{
TableBucketARN: bucketARN,
Namespace: namespace,
ContinuationToken: pageToken,
MaxViews: pageSize,
}
var listResp s3tables.ListViewsResponse
err = s.filerClient.WithFilerClient(false, func(client filer_pb.SeaweedFilerClient) error {
mgrClient := s3tables.NewManagerClient(client)
return s.tablesManager.Execute(r.Context(), mgrClient, "ListViews", listReq, &listResp, identityName)
})
if err != nil {
if strings.Contains(err.Error(), "not found") {
writeError(w, http.StatusNotFound, "NoSuchNamespaceException", fmt.Sprintf("Namespace does not exist: %v", namespace))
return
}
glog.V(1).Infof("Iceberg: ListViews error: %v", err)
writeManagerError(w, err)
return
}
identifiers := make([]TableIdentifier, 0, len(listResp.Views))
for _, v := range listResp.Views {
identifiers = append(identifiers, TableIdentifier{Namespace: namespace, Name: v.Name})
}
writeJSON(w, http.StatusOK, ListViewsResponse{
NextPageToken: listResp.ContinuationToken,
Identifiers: identifiers,
})
}
// handleCreateView creates a new view and writes its v1 metadata.json.
func (s *Server) handleCreateView(w http.ResponseWriter, r *http.Request) {
vars := mux.Vars(r)
namespace := parseNamespace(vars["namespace"])
if len(namespace) == 0 {
writeError(w, http.StatusBadRequest, "BadRequestException", "Namespace is required")
return
}
var req CreateViewRequest
if err := json.NewDecoder(r.Body).Decode(&req); err != nil {
writeError(w, http.StatusBadRequest, "BadRequestException", "Invalid request body")
return
}
if req.Name == "" {
writeError(w, http.StatusBadRequest, "BadRequestException", "view name is required")
return
}
if req.Schema == nil {
writeError(w, http.StatusBadRequest, "BadRequestException", "view schema is required")
return
}
if req.ViewVersion == nil {
writeError(w, http.StatusBadRequest, "BadRequestException", "view-version is required")
return
}
bucketName := getBucketFromPrefix(r)
bucketARN := buildTableBucketARN(bucketName)
identityName := s3_constants.GetIdentityNameFromContext(r)
viewPath := path.Join(flattenNamespacePath(namespace), req.Name+"-"+uuid.NewString())
location := strings.TrimSuffix(req.Location, "/")
if location == "" {
if req.Properties != nil {
if warehouse := strings.TrimSuffix(req.Properties["warehouse"], "/"); warehouse != "" {
location = fmt.Sprintf("%s/%s", warehouse, viewPath)
}
}
if location == "" {
if warehouse := strings.TrimSuffix(os.Getenv("ICEBERG_WAREHOUSE"), "/"); warehouse != "" {
location = fmt.Sprintf("%s/%s", warehouse, viewPath)
}
}
if location == "" {
location = fmt.Sprintf("s3://%s/%s", bucketName, viewPath)
}
} else {
parsedBucket, parsedPath, err := parseS3Location(location)
if err != nil {
writeError(w, http.StatusBadRequest, "BadRequestException", "Invalid view location: "+err.Error())
return
}
if parsedPath == "" {
location = fmt.Sprintf("s3://%s/%s", parsedBucket, viewPath)
}
}
metadataBucket, metadataPath, err := parseS3Location(location)
if err != nil {
writeError(w, http.StatusInternalServerError, "InternalServerError", "Invalid view location: "+err.Error())
return
}
if metadataBucket != bucketName {
writeError(w, http.StatusBadRequest, "BadRequestException", "view location must be within bucket "+bucketName)
return
}
if !isValidTablePath(metadataPath) {
writeError(w, http.StatusBadRequest, "BadRequestException", "invalid view location path")
return
}
metadata, err := view.NewMetadata(req.ViewVersion, req.Schema, location, viewProperties(req.Properties))
if err != nil {
writeError(w, http.StatusBadRequest, "BadRequestException", "Failed to build view metadata: "+err.Error())
return
}
metadataFileName := "v1.metadata.json"
metadataLocation := fmt.Sprintf("%s/metadata/%s", location, metadataFileName)
metadataBytes, err := json.Marshal(metadata)
if err != nil {
writeError(w, http.StatusInternalServerError, "InternalServerError", "Failed to serialize view metadata: "+err.Error())
return
}
// Authoritative existence check before touching storage: a registered view
// short-circuits with its stored definition (idempotent CreateView) so we
// never overwrite the persisted metadata of an existing view.
if existsResp, existsErr := s.getView(r, namespace, req.Name); existsErr == nil {
result, buildErr := s.buildViewResponse(r, existsResp, bucketName, namespace, req.Name)
if buildErr != nil {
writeError(w, http.StatusInternalServerError, "InternalServerError", buildErr.Error())
return
}
writeLoadResult(w, http.StatusOK, result)
return
} else if !isViewNotFound(existsErr) {
glog.V(1).Infof("Iceberg: CreateView existence check failed for %s.%s: %v", flattenNamespacePath(namespace), req.Name, existsErr)
writeManagerError(w, existsErr)
return
}
createReq := &s3tables.CreateViewRequest{
TableBucketARN: bucketARN,
Namespace: namespace,
Name: req.Name,
Metadata: &s3tables.TableMetadata{
Iceberg: &s3tables.IcebergMetadata{TableUUID: metadata.ViewUUID().String()},
FullMetadata: metadataBytes,
},
MetadataLocation: metadataLocation,
MetadataVersion: 1,
}
var createResp s3tables.CreateViewResponse
err = s.filerClient.WithFilerClient(false, func(client filer_pb.SeaweedFilerClient) error {
mgrClient := s3tables.NewManagerClient(client)
return s.tablesManager.Execute(r.Context(), mgrClient, "CreateView", createReq, &createResp, identityName)
})
if err != nil {
if isViewAlreadyExists(err) {
writeError(w, http.StatusConflict, "AlreadyExistsException", err.Error())
return
}
var viewErr *s3tables.S3TablesError
if errors.As(err, &viewErr) && viewErr.Type == s3tables.ErrCodeNoSuchNamespace {
writeError(w, http.StatusNotFound, "NoSuchNamespaceException", fmt.Sprintf("Namespace does not exist: %v", namespace))
return
}
glog.V(1).Infof("Iceberg: CreateView error: %v", err)
writeManagerError(w, err)
return
}
if createResp.MetadataLocation != metadataLocation {
// Another create registered the name first; answer with its view.
existsResp, existsErr := s.getView(r, namespace, req.Name)
if existsErr != nil {
writeError(w, http.StatusConflict, "AlreadyExistsException", fmt.Sprintf("View already exists: %s", req.Name))
return
}
result, buildErr := s.buildViewResponse(r, existsResp, bucketName, namespace, req.Name)
if buildErr != nil {
writeError(w, http.StatusInternalServerError, "InternalServerError", buildErr.Error())
return
}
writeLoadResult(w, http.StatusOK, result)
return
}
// Persist the metadata file only after the catalog registers the view, so a
// missing namespace or name collision fails before any bytes hit storage.
if err := s.saveMetadataFile(r.Context(), metadataBucket, metadataPath, metadataFileName, metadataBytes, false); err != nil {
// Roll back the registered view so it doesn't linger pointing at metadata
// that was never written.
if dropErr := s.dropView(r, namespace, req.Name); dropErr != nil {
glog.V(1).Infof("Iceberg: failed to roll back view %s.%s after metadata write error: %v", flattenNamespacePath(namespace), req.Name, dropErr)
}
writeError(w, http.StatusInternalServerError, "InternalServerError", "Failed to save view metadata file: "+err.Error())
return
}
config, _ := s.buildFileIOConfig(r, location, func() bool {
return s.mayUpdateView(r, bucketName, namespace, req.Name)
})
writeLoadResult(w, http.StatusOK, ViewResponse{
MetadataLocation: metadataLocation,
Metadata: metadata,
Config: config,
})
}
// handleLoadView loads view metadata.
func (s *Server) handleLoadView(w http.ResponseWriter, r *http.Request) {
vars := mux.Vars(r)
namespace := parseNamespace(vars["namespace"])
viewName := vars["view"]
if len(namespace) == 0 || viewName == "" {
writeError(w, http.StatusBadRequest, "BadRequestException", "Namespace and view name are required")
return
}
getResp, err := s.getView(r, namespace, viewName)
if err != nil {
if isViewNotFound(err) {
writeError(w, http.StatusNotFound, "NoSuchViewException", fmt.Sprintf("View does not exist: %s", viewName))
return
}
glog.V(1).Infof("Iceberg: LoadView error: %v", err)
writeManagerError(w, err)
return
}
result, err := s.buildViewResponse(r, getResp, getBucketFromPrefix(r), namespace, viewName)
if err != nil {
writeError(w, http.StatusInternalServerError, "InternalServerError", err.Error())
return
}
writeLoadResult(w, http.StatusOK, result)
}
// handleViewExists checks if a view exists.
func (s *Server) handleViewExists(w http.ResponseWriter, r *http.Request) {
vars := mux.Vars(r)
namespace := parseNamespace(vars["namespace"])
viewName := vars["view"]
if len(namespace) == 0 || viewName == "" {
w.WriteHeader(http.StatusBadRequest)
return
}
if _, err := s.getView(r, namespace, viewName); err != nil {
w.WriteHeader(http.StatusNotFound)
return
}
w.WriteHeader(http.StatusNoContent)
}
// handleDropView deletes a view.
func (s *Server) handleDropView(w http.ResponseWriter, r *http.Request) {
vars := mux.Vars(r)
namespace := parseNamespace(vars["namespace"])
viewName := vars["view"]
if len(namespace) == 0 || viewName == "" {
writeError(w, http.StatusBadRequest, "BadRequestException", "Namespace and view name are required")
return
}
var storedMetadataLocation string
if getResp, err := s.getView(r, namespace, viewName); err == nil {
storedMetadataLocation = getResp.MetadataLocation
}
err := s.dropView(r, namespace, viewName)
if err != nil {
if isViewNotFound(err) {
writeError(w, http.StatusNotFound, "NoSuchViewException", fmt.Sprintf("View does not exist: %s", viewName))
return
}
glog.V(1).Infof("Iceberg: DropView error: %v", err)
writeManagerError(w, err)
return
}
if storedMetadataLocation != "" {
if viewLoc := tableLocationFromMetadataLocation(storedMetadataLocation); viewLoc != "" {
if dataBucket, dataPath, parseErr := parseS3Location(viewLoc); parseErr == nil {
if cleanupErr := s.cleanupStaleTableLocation(r.Context(), dataBucket, dataPath); cleanupErr != nil {
glog.V(1).Infof("Iceberg: failed to purge dropped view location s3://%s/%s: %v", dataBucket, dataPath, cleanupErr)
}
}
}
}
w.WriteHeader(http.StatusNoContent)
}
// dropView removes a view registration from the catalog.
func (s *Server) dropView(r *http.Request, namespace []string, viewName string) error {
deleteReq := &s3tables.DeleteViewRequest{
TableBucketARN: buildTableBucketARN(getBucketFromPrefix(r)),
Namespace: namespace,
Name: viewName,
}
identityName := s3_constants.GetIdentityNameFromContext(r)
return s.filerClient.WithFilerClient(false, func(client filer_pb.SeaweedFilerClient) error {
mgrClient := s3tables.NewManagerClient(client)
return s.tablesManager.Execute(r.Context(), mgrClient, "DeleteView", deleteReq, nil, identityName)
})
}
// getView fetches a view's stored metadata pointer from the catalog.
func (s *Server) getView(r *http.Request, namespace []string, viewName string) (s3tables.GetViewResponse, error) {
getReq := &s3tables.GetViewRequest{
TableBucketARN: buildTableBucketARN(getBucketFromPrefix(r)),
Namespace: namespace,
Name: viewName,
}
var getResp s3tables.GetViewResponse
identityName := s3_constants.GetIdentityNameFromContext(r)
err := s.filerClient.WithFilerClient(false, func(client filer_pb.SeaweedFilerClient) error {
mgrClient := s3tables.NewManagerClient(client)
return s.tablesManager.Execute(r.Context(), mgrClient, "GetView", getReq, &getResp, identityName)
})
return getResp, err
}
// buildViewResponse parses the stored view metadata into a ViewResponse.
func (s *Server) buildViewResponse(r *http.Request, getResp s3tables.GetViewResponse, bucketName string, namespace []string, viewName string) (ViewResponse, error) {
if getResp.Metadata == nil || len(getResp.Metadata.FullMetadata) == 0 {
return ViewResponse{}, fmt.Errorf("view %s has no metadata", viewName)
}
metadata, err := view.ParseMetadataBytes(getResp.Metadata.FullMetadata)
if err != nil {
return ViewResponse{}, fmt.Errorf("failed to parse view metadata: %w", err)
}
config, _ := s.buildFileIOConfig(r, tableLocationFromMetadataLocation(getResp.MetadataLocation), func() bool {
return s.mayUpdateView(r, bucketName, namespace, viewName)
})
return ViewResponse{
MetadataLocation: getResp.MetadataLocation,
Metadata: metadata,
Config: config,
}, nil
}
// viewProperties returns a non-nil copy of props for view metadata construction.
func viewProperties(props iceberg.Properties) iceberg.Properties {
out := make(iceberg.Properties, len(props))
for k, v := range props {
out[k] = v
}
return out
}
func isViewNotFound(err error) bool {
if err == nil {
return false
}
var viewErr *s3tables.S3TablesError
if errors.As(err, &viewErr) {
if viewErr.Type == s3tables.ErrCodeNoSuchView || viewErr.Type == s3tables.ErrCodeNoSuchNamespace {
return true
}
}
return strings.Contains(strings.ToLower(err.Error()), "not found")
}
func isViewAlreadyExists(err error) bool {
if err == nil {
return false
}
var viewErr *s3tables.S3TablesError
if errors.As(err, &viewErr) && viewErr.Type == s3tables.ErrCodeViewAlreadyExists {
return true
}
return strings.Contains(strings.ToLower(err.Error()), "already exists")
}
// handleRenameView moves a view's catalog pointer to a new namespace/name,
// the view counterpart of POST /v1/{prefix}/tables/rename.
func (s *Server) handleRenameView(w http.ResponseWriter, r *http.Request) {
var req RenameTableRequest
if err := json.NewDecoder(r.Body).Decode(&req); err != nil {
writeError(w, http.StatusBadRequest, "BadRequestException", "Invalid request body")
return
}
source := parseNamespace(encodeNamespace(req.Source.Namespace))
dest := parseNamespace(encodeNamespace(req.Destination.Namespace))
if len(source) == 0 || req.Source.Name == "" || len(dest) == 0 || req.Destination.Name == "" {
writeError(w, http.StatusBadRequest, "BadRequestException", "source and destination namespace and name are required")
return
}
bucketName := getBucketFromPrefix(r)
identityName := s3_constants.GetIdentityNameFromContext(r)
renameReq := &s3tables.RenameTableRequest{
TableBucketARN: buildTableBucketARN(bucketName),
SourceNamespace: source,
SourceName: req.Source.Name,
DestNamespace: dest,
DestName: req.Destination.Name,
}
err := s.filerClient.WithFilerClient(false, func(client filer_pb.SeaweedFilerClient) error {
mgrClient := s3tables.NewManagerClient(client)
return s.tablesManager.Execute(r.Context(), mgrClient, "RenameView", renameReq, nil, identityName)
})
if err != nil {
var viewErr *s3tables.S3TablesError
if errors.As(err, &viewErr) {
switch viewErr.Type {
case s3tables.ErrCodeNoSuchView:
writeError(w, http.StatusNotFound, "NoSuchViewException", fmt.Sprintf("View does not exist: %s", req.Source.Name))
return
case s3tables.ErrCodeNoSuchNamespace:
writeError(w, http.StatusNotFound, "NoSuchNamespaceException", fmt.Sprintf("Namespace does not exist: %v", dest))
return
case s3tables.ErrCodeViewAlreadyExists:
writeError(w, http.StatusConflict, "AlreadyExistsException", fmt.Sprintf("View already exists: %s", req.Destination.Name))
return
case s3tables.ErrCodeInvalidRequest:
writeError(w, http.StatusBadRequest, "BadRequestException", viewErr.Message)
return
}
}
glog.V(1).Infof("Iceberg: RenameView error: %v", err)
writeManagerError(w, err)
return
}
w.WriteHeader(http.StatusNoContent)
}