org.apache.hadoop.fs.s3a.s3guard

Interface MetadataStore

All Superinterfaces:

AutoCloseable, Closeable

All Known Implementing Classes:

DynamoDBMetadataStore, LocalMetadataStore, NullMetadataStore

@InterfaceAudience.Private
 @InterfaceStability.Evolving
public interface MetadataStore
extends Closeable

MetadataStore defines the set of operations that any metadata store implementation must provide. Note that all Path objects provided to methods must be absolute, not relative paths. Implementations must implement any retries needed internally, such that transient errors are generally recovered from without throwing exceptions from this API.

Nested Class Summary

Nested Classes

Modifier and Type	Interface and Description
static class	MetadataStore.PruneMode Modes of operation for prune.

Method Summary

Modifier and Type	Method and Description
void	addAncestors(org.apache.hadoop.fs.Path qualifiedPath, BulkOperationState operationState) This adds all new ancestors of a path as directories.
void	delete(org.apache.hadoop.fs.Path path, BulkOperationState operationState) Deletes exactly one path, leaving a tombstone to prevent lingering, inconsistent copies of it from being listed.
void	deletePaths(Collection<org.apache.hadoop.fs.Path> paths, BulkOperationState operationState) Delete the paths.
void	deleteSubtree(org.apache.hadoop.fs.Path path, BulkOperationState operationState) Deletes the entire sub-tree rooted at the given path, leaving tombstones to prevent lingering, inconsistent copies of it from being listed.
void	destroy() Destroy all resources associated with the metadata store.
void	forgetMetadata(org.apache.hadoop.fs.Path path) Removes the record of exactly one path.
PathMetadata	get(org.apache.hadoop.fs.Path path) Gets metadata for a path.
PathMetadata	get(org.apache.hadoop.fs.Path path, boolean wantEmptyDirectoryFlag) Gets metadata for a path.
Map<String,String>	getDiagnostics() Get any diagnostics information from a store, as a list of (key, value) tuples for display.
default MetastoreInstrumentation	getInstrumentation() Get any S3GuardInstrumentation for this store...must not be null.
void	initialize(org.apache.hadoop.conf.Configuration conf, ITtlTimeProvider ttlTimeProvider) Performs one-time initialization of the metadata store via configuration.
void	initialize(org.apache.hadoop.fs.FileSystem fs, ITtlTimeProvider ttlTimeProvider) Performs one-time initialization of the metadata store.
default BulkOperationState	initiateBulkWrite(BulkOperationState.OperationType operation, org.apache.hadoop.fs.Path dest) Initiate a bulk update and create an operation state for it.
RenameTracker	initiateRenameOperation(StoreContext storeContext, org.apache.hadoop.fs.Path source, S3AFileStatus sourceStatus, org.apache.hadoop.fs.Path dest) Start a rename operation.
DirListingMetadata	listChildren(org.apache.hadoop.fs.Path path) Lists metadata for all direct children of a path.
default int	markAsAuthoritative(org.apache.hadoop.fs.Path dest, BulkOperationState operationState) Mark all directories created/touched in an operation as authoritative.
void	move(Collection<org.apache.hadoop.fs.Path> pathsToDelete, Collection<PathMetadata> pathsToCreate, BulkOperationState operationState) Record the effects of a FileSystem.rename(Path, Path) in the MetadataStore.
void	prune(MetadataStore.PruneMode pruneMode, long cutoff) Prune method with two modes of operation: MetadataStore.PruneMode.ALL_BY_MODTIME Clear any metadata older than a specified mod_time from the store.
long	prune(MetadataStore.PruneMode pruneMode, long cutoff, String keyPrefix) Same as prune(PruneMode, long), but with an additional keyPrefix parameter to filter the pruned keys with a prefix.
void	put(Collection<? extends PathMetadata> metas, BulkOperationState operationState) Saves metadata for any number of paths.
void	put(DirListingMetadata meta, List<org.apache.hadoop.fs.Path> unchangedEntries, BulkOperationState operationState) Save directory listing metadata.
void	put(PathMetadata meta) Saves metadata for exactly one path.
void	put(PathMetadata meta, BulkOperationState operationState) Saves metadata for exactly one path, potentially using any bulk operation state to eliminate duplicate work.
void	setTtlTimeProvider(ITtlTimeProvider ttlTimeProvider) The TtlTimeProvider has to be set during the initialization for the metadatastore, but this method can be used for testing, and change the instance during runtime.
void	updateParameters(Map<String,String> parameters) Tune/update parameters for an existing table.

Methods inherited from interface java.io.Closeable

close

Method Detail

initialize

void initialize(org.apache.hadoop.fs.FileSystem fs,
                ITtlTimeProvider ttlTimeProvider)
         throws IOException

Performs one-time initialization of the metadata store.

Parameters:

fs - FileSystem associated with the MetadataStore

ttlTimeProvider - the time provider to use for metadata expiry

Throws:

IOException - if there is an error

initialize

void initialize(org.apache.hadoop.conf.Configuration conf,
                ITtlTimeProvider ttlTimeProvider)
         throws IOException

Performs one-time initialization of the metadata store via configuration.

Parameters:

conf - Configuration.

ttlTimeProvider - the time provider to use for metadata expiry

Throws:

IOException - if there is an error

See Also:

initialize(FileSystem, ITtlTimeProvider)

delete

void delete(org.apache.hadoop.fs.Path path,
            @Nullable
            BulkOperationState operationState)
     throws IOException

Deletes exactly one path, leaving a tombstone to prevent lingering, inconsistent copies of it from being listed. Deleting an entry with a tombstone needs a S3Guard.TtlTimeProvider because the lastUpdated field of the record has to be updated to

now

.

Parameters:

path - the path to delete

operationState - (nullable) operational state for a bulk update

Throws:

IOException - if there is an error

forgetMetadata

void forgetMetadata(org.apache.hadoop.fs.Path path)
             throws IOException

Removes the record of exactly one path. Does not leave a tombstone (see delete(Path, BulkOperationState). It is currently intended for testing only, and a need to use it as part of normal FileSystem usage is not anticipated.

Parameters:

path - the path to delete

Throws:

IOException - if there is an error

deleteSubtree

@Retries.RetryTranslated
void deleteSubtree(org.apache.hadoop.fs.Path path,
                                            @Nullable
                                            BulkOperationState operationState)
                                     throws IOException

Deletes the entire sub-tree rooted at the given path, leaving tombstones to prevent lingering, inconsistent copies of it from being listed. In addition to affecting future calls to get(Path), implementations must also update any stored DirListingMetadata objects which track the parent of this file. Deleting a subtree with a tombstone needs a S3Guard.TtlTimeProvider because the lastUpdated field of all records have to be updated to

now

.

Parameters:

path - the root of the sub-tree to delete

operationState - (nullable) operational state for a bulk update

Throws:

IOException - if there is an error

deletePaths

@Retries.RetryTranslated
void deletePaths(Collection<org.apache.hadoop.fs.Path> paths,
                                          @Nullable
                                          BulkOperationState operationState)
                                   throws IOException

Delete the paths. There's no attempt to order the paths: they are deleted in the order passed in.

Parameters:

paths - paths to delete.

operationState - Nullable operation state

Throws:

IOException - failure

get

PathMetadata get(org.apache.hadoop.fs.Path path)
          throws IOException

Gets metadata for a path.

Parameters:

path - the path to get

Returns:

metadata for path, null if not found

Throws:

IOException - if there is an error

get

PathMetadata get(org.apache.hadoop.fs.Path path,
                 boolean wantEmptyDirectoryFlag)
          throws IOException

Gets metadata for a path. Alternate method that includes a hint whether or not the MetadataStore should do work to compute the value for PathMetadata.isEmptyDirectory(). Since determining emptiness may be an expensive operation, this can save wasted work.

Parameters:

path - the path to get

wantEmptyDirectoryFlag - Set to true to give a hint to the MetadataStore that it should try to compute the empty directory flag.

Returns:

metadata for path, null if not found

Throws:

IOException - if there is an error

listChildren

@Retries.RetryTranslated
DirListingMetadata listChildren(org.apache.hadoop.fs.Path path)
                                                  throws IOException

Lists metadata for all direct children of a path.

Parameters:

path - the path to list

Returns:

metadata for all direct children of path which are being tracked by the MetadataStore, or null if the path was not found in the MetadataStore.

Throws:

IOException - if there is an error

addAncestors

@Retries.RetryTranslated
void addAncestors(org.apache.hadoop.fs.Path qualifiedPath,
                                           @Nullable
                                           BulkOperationState operationState)
                                    throws IOException

This adds all new ancestors of a path as directories.

Important: to propagate TTL information, any new ancestors added must have their last updated timestamps set through S3Guard.patchLastUpdated(Collection, ITtlTimeProvider).

Parameters:

qualifiedPath - path to update

operationState - (nullable) operational state for a bulk update

Throws:

IOException - failure

move

void move(@Nullable
          Collection<org.apache.hadoop.fs.Path> pathsToDelete,
          @Nullable
          Collection<PathMetadata> pathsToCreate,
          @Nullable
          BulkOperationState operationState)
   throws IOException

Record the effects of a FileSystem.rename(Path, Path) in the MetadataStore. Clients provide explicit enumeration of the affected paths (recursively), before and after the rename. This operation is not atomic, unless specific implementations claim otherwise. On the need to provide an enumeration of directory trees instead of just source and destination paths: Since a MetadataStore does not have to track all metadata for the underlying storage system, and a new MetadataStore may be created on an existing underlying filesystem, this move() may be the first time the MetadataStore sees the affected paths. Therefore, simply providing src and destination paths may not be enough to record the deletions (under src path) and creations (at destination) that are happening during the rename().

Parameters:

pathsToDelete - Collection of all paths that were removed from the source directory tree of the move.

pathsToCreate - Collection of all PathMetadata for the new paths that were created at the destination of the rename().

operationState - Any ongoing state supplied to the rename tracker which is to be passed in with each move operation.

Throws:

IOException - if there is an error

put

@Retries.RetryTranslated
void put(PathMetadata meta)
                           throws IOException

Saves metadata for exactly one path. Implementations may pre-create all the path's ancestors automatically. Implementations must update any DirListingMetadata objects which track the immediate parent of this file.

Parameters:

meta - the metadata to save

Throws:

IOException - if there is an error

put

@Retries.RetryTranslated
void put(PathMetadata meta,
                                  @Nullable
                                  BulkOperationState operationState)
                           throws IOException

Saves metadata for exactly one path, potentially using any bulk operation state to eliminate duplicate work. Implementations may pre-create all the path's ancestors automatically. Implementations must update any DirListingMetadata objects which track the immediate parent of this file.

Parameters:

meta - the metadata to save

operationState - operational state for a bulk update

Throws:

IOException - if there is an error

put

void put(Collection<? extends PathMetadata> metas,
         @Nullable
         BulkOperationState operationState)
  throws IOException

Saves metadata for any number of paths. Semantics are otherwise the same as single-path puts.

Parameters:

metas - the metadata to save

operationState - (nullable) operational state for a bulk update

Throws:

IOException - if there is an error

put

void put(DirListingMetadata meta,
         List<org.apache.hadoop.fs.Path> unchangedEntries,
         @Nullable
         BulkOperationState operationState)
  throws IOException

Save directory listing metadata. Callers may save a partial directory listing for a given path, or may store a complete and authoritative copy of the directory listing. MetadataStore implementations may subsequently keep track of all modifications to the directory contents at this path, and return authoritative results from subsequent calls to listChildren(Path). See DirListingMetadata. Any authoritative results returned are only authoritative for the scope of the MetadataStore: A per-process MetadataStore, for example, would only show results visible to that process, potentially missing metadata updates (create, delete) made to the same path by another process. To optimize updates and avoid overwriting existing entries which may contain extra data, entries in the list of unchangedEntries may be excluded. That is: the listing metadata has the full list of what it believes are children, but implementations can opt to ignore some.

Parameters:

meta - Directory listing metadata.

unchangedEntries - list of entries in the dir listing which have not changed since the directory was list scanned on s3guard.

operationState - operational state for a bulk update

Throws:

IOException - if there is an error

destroy

void destroy()
      throws IOException

Destroy all resources associated with the metadata store. The destroyed resources can be DynamoDB tables, MySQL databases/tables, or HDFS directories. Any operations after calling this method may possibly fail. This operation is idempotent.

Throws:

IOException - if there is an error

prune

void prune(MetadataStore.PruneMode pruneMode,
           long cutoff)
    throws IOException,
           UnsupportedOperationException

Prune method with two modes of operation:

• MetadataStore.PruneMode.ALL_BY_MODTIME Clear any metadata older than a specified mod_time from the store. Note that this modification time is the S3 modification time from the object's metadata - from the object store. Implementations MUST clear file metadata, and MAY clear directory metadata (s3a itself does not track modification time for directories). Implementations may also choose to throw UnsupportedOperationException instead. Note that modification times must be in UTC, as returned by System.currentTimeMillis at the time of modification.

• MetadataStore.PruneMode.TOMBSTONES_BY_LASTUPDATED Clear any tombstone updated earlier than a specified time from the store. Note that this last_updated is the time when the metadata entry was last updated and maintained by the metadata store. Implementations MUST clear file metadata, and MAY clear directory metadata (s3a itself does not track modification time for directories). Implementations may also choose to throw UnsupportedOperationException instead. Note that last_updated must be in UTC, as returned by System.currentTimeMillis at the time of modification.

Parameters:

pruneMode - Prune Mode

cutoff - Oldest time to allow (UTC)

Throws:

IOException - if there is an error

UnsupportedOperationException - if not implemented

prune

long prune(MetadataStore.PruneMode pruneMode,
           long cutoff,
           String keyPrefix)
    throws IOException,
           UnsupportedOperationException

Same as prune(PruneMode, long), but with an additional keyPrefix parameter to filter the pruned keys with a prefix.

Parameters:

pruneMode - Prune Mode

cutoff - Oldest time in milliseconds to allow (UTC)

keyPrefix - The prefix for the keys that should be removed

Returns:

the number of pruned entries

Throws:

IOException - if there is an error

UnsupportedOperationException - if not implemented

getDiagnostics

Map<String,String> getDiagnostics()
                           throws IOException

Get any diagnostics information from a store, as a list of (key, value) tuples for display. Arbitrary values; no guarantee of stability. These are for debugging and testing only.

Returns:

a map of strings.

Throws:

IOException - if there is an error

updateParameters

void updateParameters(Map<String,String> parameters)
               throws IOException

Tune/update parameters for an existing table.

Parameters:

parameters - map of params to change.

Throws:

IOException - if there is an error

markAsAuthoritative

default int markAsAuthoritative(org.apache.hadoop.fs.Path dest,
                                BulkOperationState operationState)
                         throws IOException

Mark all directories created/touched in an operation as authoritative. The metastore can now update that path with any authoritative flags it chooses. The store may assume that therefore the operation state is complete. This holds for rename and needs to be documented for import.

Parameters:

dest - destination path.

operationState - active state.

Returns:

the number of directories marked.

Throws:

IOException - failure.

initiateRenameOperation

RenameTracker initiateRenameOperation(StoreContext storeContext,
                                      org.apache.hadoop.fs.Path source,
                                      S3AFileStatus sourceStatus,
                                      org.apache.hadoop.fs.Path dest)
                               throws IOException

Start a rename operation.

Parameters:

storeContext - store context.

source - source path

sourceStatus - status of the source file/dir

dest - destination path.

Returns:

the rename tracker

Throws:

IOException - Failure.

initiateBulkWrite

default BulkOperationState initiateBulkWrite(BulkOperationState.OperationType operation,
                                             org.apache.hadoop.fs.Path dest)
                                      throws IOException

Initiate a bulk update and create an operation state for it. This may then be passed into put operations.

Parameters:

operation - the type of the operation.

dest - path under which updates will be explicitly put.

Returns:

null or a store-specific state to pass into the put operations.

Throws:

IOException - failure

setTtlTimeProvider

void setTtlTimeProvider(ITtlTimeProvider ttlTimeProvider)

The TtlTimeProvider has to be set during the initialization for the metadatastore, but this method can be used for testing, and change the instance during runtime.

Parameters:

ttlTimeProvider -

getInstrumentation

default MetastoreInstrumentation getInstrumentation()

Get any S3GuardInstrumentation for this store...must not be null.

Returns:

any store instrumentation.