Package org.elasticsearch.index.store

Class Store

  • All Implemented Interfaces:
    Closeable, AutoCloseable, RefCounted, IndexShardComponent
    public class Store
extends AbstractIndexShardComponent
implements Closeable, RefCounted
    A Store provides plain access to files written by an elasticsearch index shard. Each shard has a dedicated store that is uses to access Lucene's Directory which represents the lowest level of file abstraction in Lucene used to read and write Lucene indices. This class also provides access to metadata information like checksums for committed files. A committed file is a file that belongs to a segment written by a Lucene commit. Files that have not been committed ie. created during a merge or a shard refresh / NRT reopen are not considered in the MetadataSnapshot.

    Note: If you use a store it's reference count should be increased before using it by calling #incRef and a corresponding #decRef must be called in a try/finally block to release the store again ie.: 

          store.incRef();
      try {
        // use the store...

      } finally {
          store.decRef();
      }

    • Field Detail

      • FORCE_RAM_TERM_DICT

        public static final Setting<Boolean> FORCE_RAM_TERM_DICT
        This is an escape hatch for lucenes internal optimization that checks if the IndexInput is an instance of ByteBufferIndexInput and if that's the case doesn't load the term dictionary into ram but loads it off disk iff the fields is not an ID like field. Since this optimization has been added very late in the release processes we add this setting to allow users to opt-out of this by exploiting lucene internals and wrapping the IndexInput in a simple delegate.

      • INDEX_STORE_STATS_REFRESH_INTERVAL_SETTING

        public static final Setting<TimeValue> INDEX_STORE_STATS_REFRESH_INTERVAL_SETTING

    • Method Detail

      • readLastCommittedSegmentsInfo

        public SegmentInfos readLastCommittedSegmentsInfo()
                                           throws IOException
        Returns the last committed segments info for this store
        Throws:
        IOException - if the index is corrupted or the segments file is not present

      • loadSeqNoInfo

        public static SequenceNumbers.CommitInfo loadSeqNoInfo​(IndexCommit commit)
                                                throws IOException
        Loads the maximum sequence number and local checkpoint from the given Lucene commit point or the latest if not provided.
        Parameters:
        commit - the commit point to load seqno stats, or the last commit in the store if the parameter is null
        Returns:
        SequenceNumbers.CommitInfo containing information about the last commit
        Throws:
        IOException - if an I/O exception occurred reading the latest Lucene commit point from disk

      • renameTempFilesSafe

        public void renameTempFilesSafe​(Map<String,​String> tempFileMap)
                         throws IOException
        Renames all the given files from the key of the map to the value of the map. All successfully renamed files are removed from the map in-place.
        Throws:
        IOException

      • incRef

        public final void incRef()
        Increments the refCount of this Store instance. RefCounts are used to determine when a Store can be closed safely, i.e. as soon as there are no more references. Be sure to always call a corresponding decRef(), in a finally clause; otherwise the store may never be closed. Note that close() simply calls decRef(), which means that the Store will not really be closed until decRef() has been called for all outstanding references.

        Note: Close can safely be called multiple times.

        Specified by:
        incRef in interface RefCounted
        Throws:
        AlreadyClosedException - iff the reference counter can not be incremented.
        See Also:
        decRef(), tryIncRef()

      • tryIncRef

        public final boolean tryIncRef()
        Tries to increment the refCount of this Store instance. This method will return true iff the refCount was incremented successfully otherwise false. RefCounts are used to determine when a Store can be closed safely, i.e. as soon as there are no more references. Be sure to always call a corresponding decRef(), in a finally clause; otherwise the store may never be closed. Note that close() simply calls decRef(), which means that the Store will not really be closed until decRef() has been called for all outstanding references.

        Note: Close can safely be called multiple times.

        Specified by:
        tryIncRef in interface RefCounted
        See Also:
        decRef(), incRef()

      • decRef

        public final void decRef()
        Decreases the refCount of this Store instance. If the refCount drops to 0, then this store is closed.
        Specified by:
        decRef in interface RefCounted
        See Also:
        incRef()

      • createVerifyingOutput

        public IndexOutput createVerifyingOutput​(String fileName,
                                         StoreFileMetaData metadata,
                                         IOContext context)
                                  throws IOException
        The returned IndexOutput validates the files checksum.

        Note: Checksums are calculated by default since version 4.8.0. This method only adds the verification against the checksum in the given metadata and does not add any significant overhead.

        Throws:
        IOException

      • checkIntegrityNoException

        public boolean checkIntegrityNoException​(StoreFileMetaData md)

      • removeCorruptionMarker

        public void removeCorruptionMarker()
                            throws IOException
        Deletes all corruption markers from this store.
        Throws:
        IOException

      • cleanupAndVerify

        public void cleanupAndVerify​(String reason,
                             Store.MetadataSnapshot sourceMetaData)
                      throws IOException
        This method deletes every file in this store that is not contained in the given source meta data or is a legacy checksum file. After the delete it pulls the latest metadata snapshot from the store and compares it to the given snapshot. If the snapshots are inconsistent an illegal state exception is thrown.
        Parameters:
        reason - the reason for this cleanup operation logged for each deleted file
        sourceMetaData - the metadata used for cleanup. all files in this metadata should be kept around.
        Throws:
        IOException - if an IOException occurs
        IllegalStateException - if the latest snapshot in this store differs from the given one after the cleanup.

      • refCount

        public int refCount()
        Returns the current reference count.

      • isAutogenerated

        public static boolean isAutogenerated​(String name)
        Returns true if the file is auto-generated by the store and shouldn't be deleted during cleanup. This includes write lock and checksum files

      • digestToString

        public static String digestToString​(long digest)
        Produces a string representation of the given digest value.

      • deleteQuiet

        public void deleteQuiet​(String... files)

      • markStoreCorrupted

        public void markStoreCorrupted​(IOException exception)
                        throws IOException
        Marks this store as corrupted. This method writes a corrupted_${uuid} file containing the given exception message. If a store contains a corrupted_${uuid} file isMarkedCorrupted() will return true.
        Throws:
        IOException

      • createEmpty

        public void createEmpty​(Version luceneVersion)
                 throws IOException
        creates an empty lucene index and a corresponding empty translog. Any existing data will be deleted.
        Throws:
        IOException

      • bootstrapNewHistory

        public void bootstrapNewHistory()
                         throws IOException
        Marks an existing lucene index with a new history uuid. This is used to make sure no existing shard will recovery from this index using ops based recovery.
        Throws:
        IOException

      • bootstrapNewHistory

        public void bootstrapNewHistory​(long localCheckpoint,
                                long maxSeqNo)
                         throws IOException
        Marks an existing lucene index with a new history uuid and sets the given local checkpoint as well as the maximum sequence number. This is used to make sure no existing shard will recover from this index using ops based recovery.
        Throws:
        IOException
        See Also:
        SequenceNumbers.LOCAL_CHECKPOINT_KEY, SequenceNumbers.MAX_SEQ_NO

      • associateIndexWithNewTranslog

        public void associateIndexWithNewTranslog​(String translogUUID)
                                   throws IOException
        Force bakes the given translog generation as recovery information in the lucene index. This is used when recovering from a snapshot or peer file based recovery where a new empty translog is created and the existing lucene index needs should be changed to use it.
        Throws:
        IOException

      • ensureIndexHasHistoryUUID

        public void ensureIndexHasHistoryUUID()
                               throws IOException
        Checks that the Lucene index contains a history uuid marker. If not, a new one is generated and committed.
        Throws:
        IOException

      • trimUnsafeCommits

        public void trimUnsafeCommits​(long lastSyncedGlobalCheckpoint,
                              long minRetainedTranslogGen,
                              Version indexVersionCreated)
                       throws IOException
        Keeping existing unsafe commits when opening an engine can be problematic because these commits are not safe at the recovering time but they can suddenly become safe in the future. The following issues can happen if unsafe commits are kept oninit.

        1. Replica can use unsafe commit in peer-recovery. This happens when a replica with a safe commit c1(max_seqno=1) and an unsafe commit c2(max_seqno=2) recovers from a primary with c1(max_seqno=1). If a new document(seqno=2) is added without flushing, the global checkpoint is advanced to 2; and the replica recovers again, it will use the unsafe commit c2(max_seqno=2 at most gcp=2) as the starting commit for sequenced-based recovery even the commit c2 contains a stale operation and the document(with seqno=2) will not be replicated to the replica.

        2. Min translog gen for recovery can go backwards in peer-recovery. This happens when are replica with a safe commit c1(local_checkpoint=1, recovery_translog_gen=1) and an unsafe commit c2(local_checkpoint=2, recovery_translog_gen=2). The replica recovers from a primary, and keeps c2 as the last commit, then sets last_translog_gen to 2. Flushing a new commit on the replica will cause exception as the new last commit c3 will have recovery_translog_gen=1. The recovery translog generation of a commit is calculated based on the current local checkpoint. The local checkpoint of c3 is 1 while the local checkpoint of c2 is 2.

        3. Commit without translog can be used in recovery. An old index, which was created before multiple-commits is introduced (v6.2), may not have a safe commit. If that index has a snapshotted commit without translog and an unsafe commit, the policy can consider the snapshotted commit as a safe commit for recovery even the commit does not have translog.

        Throws:
        IOException