Package io.microraft.persistence

Class NopRaftStore

java.lang.Object
io.microraft.persistence.NopRaftStore
All Implemented Interfaces:
RaftStore
public class NopRaftStore extends Object implements RaftStore
Used when a Raft node works transiently (its state is not persisted).

  • Constructor Summary

    Constructors
    Constructor
    Description
    NopRaftStore()
     

  • Method Summary

    Modifier and Type
    Method
    Description
    void
    deleteSnapshotChunks(long logIndex, int snapshotChunkCount)
    Deletes persisted snapshot chunks at the given log index.
    void
    flush()
    Forces all buffered (in any layer) Raft log changes to be written to the storage and returns after those changes are written.
    void
    persistAndFlushInitialGroupMembers(io.microraft.model.log.RaftGroupMembersView initialGroupMembers)
    Persists and flushes the given initial Raft group members.
    void
    persistAndFlushLocalEndpoint(io.microraft.model.persistence.RaftEndpointPersistentState localEndpointPersistentState)
    Persists and flushes the given local Raft endpoint and its voting flag.
    void
    persistAndFlushTerm(io.microraft.model.persistence.RaftTermPersistentState termPersistentState)
    Persists the term and the Raft endpoint that the local Raft node voted for in the given term.
    void
    persistLogEntry(io.microraft.model.log.LogEntry logEntry)
    Persists the given log entry.
    void
    persistSnapshotChunk(io.microraft.model.log.SnapshotChunk snapshotChunk)
    Persists the given snapshot chunk.
    void
    truncateLogEntriesFrom(long logIndexInclusive)
    Rolls back the log by truncating all entries starting with the given index.

    Methods inherited from class java.lang.Object

    clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

  • Constructor Details

    • NopRaftStore

      public NopRaftStore()

  • Method Details

    • persistAndFlushLocalEndpoint

      public void persistAndFlushLocalEndpoint(@Nonnull io.microraft.model.persistence.RaftEndpointPersistentState localEndpointPersistentState)
      Description copied from interface: RaftStore
      Persists and flushes the given local Raft endpoint and its voting flag.

      When this method returns, all the provided data has become durable.

      Specified by:
      persistAndFlushLocalEndpoint in interface RaftStore
      Parameters:
      localEndpointPersistentState - the local endpoint state to be persisted

    • persistAndFlushInitialGroupMembers

      public void persistAndFlushInitialGroupMembers(@Nonnull io.microraft.model.log.RaftGroupMembersView initialGroupMembers)
      Description copied from interface: RaftStore
      Persists and flushes the given initial Raft group members.

      When this method returns, all the provided data has become durable.

      Specified by:
      persistAndFlushInitialGroupMembers in interface RaftStore
      Parameters:
      initialGroupMembers - the initial Raft group member list to persist

    • persistAndFlushTerm

      public void persistAndFlushTerm(@Nonnull io.microraft.model.persistence.RaftTermPersistentState termPersistentState)
      Description copied from interface: RaftStore
      Persists the term and the Raft endpoint that the local Raft node voted for in the given term.

      When this method returns, all the provided data has become durable.

      Specified by:
      persistAndFlushTerm in interface RaftStore
      Parameters:
      termPersistentState - the term state to be persisted

    • persistLogEntry

      public void persistLogEntry(@Nonnull io.microraft.model.log.LogEntry logEntry)
      Description copied from interface: RaftStore
      Persists the given log entry.

      Log entries are appended to the Raft log with sequential log indices. The first log index is 1.

      A block of consecutive log entries has no gaps in the indices, but a gap can appear between a snapshot entry and its preceding regular log entry. This happens in an edge case where a follower has fallen so far behind that the missing entries are no longer available from the leader. In that case the leader will send its snapshot entry instead.

      In another rare failure scenario, MicroRaft must delete a range of the highest entries, rolling back the index of the next persisted entry. Consider the following case where Raft persists 3 log entries and then deletes entries from index=2:

      • persistLogEntry(1)
      • persistLogEntry(2)
      • persistLogEntry(3)
      • truncateLogEntriesFrom(2)
      After this call sequence log indices will remain sequential and the next persistLogEntry() call will be for index=2.
      Specified by:
      persistLogEntry in interface RaftStore
      Parameters:
      logEntry - the log entry object to persist
      See Also:

    • persistSnapshotChunk

      public void persistSnapshotChunk(@Nonnull io.microraft.model.log.SnapshotChunk snapshotChunk)
      Description copied from interface: RaftStore
      Persists the given snapshot chunk.

      A snapshot is persisted with at least 1 chunk. The number of chunks in a snapshot is provided via SnapshotChunk.getSnapshotChunkCount(). A snapshot is considered to be complete when all of its chunks are provided to this method in any order, and RaftStore.flush() will be called afterwards.

      After a snapshot is persisted at index=i and RaftStore.flush() is called, the log entry at index=i, all the preceding log entries, and all the preceding snapshots are no longer needed and can be evicted from storage. Failing to evict stale entries and snapshots do not cause a consistency problem, but can increase the time to recover after a crash or restart. Therefore eviction can be done in a background task.

      MicroRaft takes snapshots at a predetermined interval, controlled by RaftConfig.getCommitCountToTakeSnapshot(). For instance, if it is 100, snapshots will occur at indices 100, 200, 300, and so on.

      The snapshot index can lag behind the index of the highest log entry that was already persisted and flushed, but there is an upper bound to this difference, controlled by RaftConfig.getMaxPendingLogEntryCount(). For instance, if it is 10, and a persistSnapshot() call is made with snapshotIndex=100, the index of the preceding persistLogEntry() call can be at most 110.

      On the other hand, the snapshot index can also be ahead of the highest log entry. This can happen when a Raft follower has fallen so far behind the leader and the leader no longer holds the missing entries. In that case, the follower receives a snapshot from the leader. There is no upper-bound on the gap between the highest log entry and the index of the received snapshot.

      Specified by:
      persistSnapshotChunk in interface RaftStore
      Parameters:
      snapshotChunk - the snapshot chunk object to persist
      See Also:

    • truncateLogEntriesFrom

      public void truncateLogEntriesFrom(long logIndexInclusive)
      Description copied from interface: RaftStore
      Rolls back the log by truncating all entries starting with the given index. A truncated log entry is no longer valid and must not be restored (or at least must be ignored during the restore process).

      There is an upper-bound on the number of persisted log entries that can be truncated afterwards, which is specified by RaftConfig.getMaxPendingLogEntryCount() + 1. Say that it is 5 and the highest persisted log entry index is 20. Then, at most 5 highest entries can be truncated, hence truncation can start at index=16 or higher.

      Specified by:
      truncateLogEntriesFrom in interface RaftStore
      Parameters:
      logIndexInclusive - the log index value from which the log entries must be truncated
      See Also:

    • deleteSnapshotChunks

      public void deleteSnapshotChunks(long logIndex, int snapshotChunkCount)
      Description copied from interface: RaftStore
      Deletes persisted snapshot chunks at the given log index. MicroRaft calls this method when it detects that it needs to start installing a new snapshot while there is a snapshot being persisted. Those snapshot chunks are no longer valid and must not be restored (or at least must be ignored during the restore process). This is merely an optimization method and its side-effects do not need to be flushed when this method returns.
      Specified by:
      deleteSnapshotChunks in interface RaftStore
      Parameters:
      logIndex - the log index value at which some snapshot chunks are persisted.
      snapshotChunkCount - the number of snapshot chunks that could have been persisted.

    • flush

      public void flush()
      Description copied from interface: RaftStore
      Forces all buffered (in any layer) Raft log changes to be written to the storage and returns after those changes are written.

      When this method returns, all the changes done via the other methods have become durable.

      Specified by:
      flush in interface RaftStore
      See Also: