package raft_serverpb

Get desktop application:
View/edit binary Protocol Buffers messages

Used in: TabletSnapshotResponse

• repeated string file_name = 1

Used in: ExtraMessage

• uint64 from_region_id = 1

• optional metapb.RegionEpoch from_region_epoch = 2

• bool unavailable = 3

• bool trimmed = 4

Used in: ExtraMessage

• uint64 from_region_id = 1

  The region ID who triggers the check and wait for report. It should be the ID of RaftMessage.from.

• uint64 check_region_id = 2

  The region ID to be checked if should be destroyed.

• optional metapb.RegionEpoch check_region_epoch = 3

  The epoch of the region to be checked.

• optional metapb.Peer check_peer = 4

  The peer to be checked.

Used as response type in: tikvpb.Tikv.BatchRaft, tikvpb.Tikv.Raft, tikvpb.Tikv.Snapshot

(message has no fields)

Used in: RaftMessage

• ExtraMessageType type = 1

• uint64 index = 2

  It's merge related index. In `WantRollbackMerge`, it's prepare merge index. In `MsgGcPeerRequest`, it's the commit merge index. In `MsgVoterReplicatedIndexRequest` it's the voter_replicated_index.

• repeated metapb.Peer check_peers = 3

  In `MsgCheckStalePeerResponse`, it's the peers that receiver can continue to query.

• bool wait_data = 4

• bool forcely_awaken = 5

  Flag for forcely wake up hibernate regions if true.

• optional CheckGcPeer check_gc_peer = 6

• optional FlushMemtable flush_memtable = 7

• optional AvailabilityContext availability_context = 8

  Used by `MsgAvailabilityRequest` and `MsgAvailabilityResponse` in v2.

• optional RefreshBuckets refresh_buckets = 9

  notice the peer to refresh buckets version

• bool snap_gen_precheck_passed = 10

  snap_gen_precheck_passed is used to indicate the precheck result when a follower responds to a leader's snapshot gen precheck request.

Used in: ExtraMessage

• MsgRegionWakeUp = 0

• MsgWantRollbackMerge = 1

• MsgCheckStalePeer = 2

• MsgCheckStalePeerResponse = 3

• MsgHibernateRequest = 4

  If leader is going to sleep, it will send requests to all its followers to make sure they all agree to sleep.

• MsgHibernateResponse = 5

• MsgRejectRaftLogCausedByMemoryUsage = 6

• MsgAvailabilityRequest = 7

• MsgAvailabilityResponse = 8

• MsgVoterReplicatedIndexRequest = 9

• MsgVoterReplicatedIndexResponse = 10

• MsgGcPeerRequest = 11

  Message means that `from` is tombstone. Leader can then update removed_records.

• MsgGcPeerResponse = 12

• MsgFlushMemtable = 13

• MsgRefreshBuckets = 14

• MsgSnapGenPrecheckRequest = 15

  Messages for the snapshot gen precheck process.

• MsgSnapGenPrecheckResponse = 16

• MsgPreLoadRegionRequest = 17

  Used in transfer leader process for leader to inform follower to load the region into in-memory engine if the relevant region is cached.

• MsgPreLoadRegionResponse = 18

  Used in transfer leader process for follower to inform leader the completes of the region cache

Used in: ExtraMessage

• uint64 region_id = 1

Used in: enginepb.SnapshotData, RaftSnapshotData

• bytes key = 1

• bytes value = 2

Used in: RegionLocalState

• uint64 min_index = 1

• optional metapb.Region target = 2

• uint64 commit = 3

Used in: RaftSnapshotData, RegionLocalState

• uint64 source_region_id = 1

• optional metapb.RegionEpoch source_epoch = 2

• repeated metapb.Peer source_peers = 3

  Peers of source region when merge is committed.

• repeated metapb.Peer source_removed_records = 9

  Removed peers (by confchange) of source region when merge is committed.

• uint64 target_region_id = 4

• optional metapb.RegionEpoch target_epoch = 5

• repeated metapb.Peer target_peers = 6

• uint64 index = 7

  Commit merge index.

• uint64 source_index = 8

  Prepare merge index.

Used in: RegionLocalState

• Normal = 0

• Applying = 1

• Tombstone = 2

• Merging = 3

• Unavailable = 4

  Currently used for witness to non-witness conversion: When a witness has just become a non-witness, we need to set and persist this state, so that when the service restarts before applying snapshot, we can actively request snapshot when initializing this peer.

Used in: debugpb.RegionInfoResponse, enginepb.CommandResponse, enginepb.SnapshotState, RegionSequenceNumberRelation

• uint64 applied_index = 1

• uint64 last_commit_index = 3

• uint64 commit_index = 4

• uint64 commit_term = 5

• optional RaftTruncatedState truncated_state = 2

Used in: debugpb.RegionInfoResponse, pdpb.PeerReport

• optional eraftpb.HardState hard_state = 1

• uint64 last_index = 2

Used as request type in: tikvpb.Tikv.Raft

Used as field type in: SnapshotChunk, TabletSnapshotHead, tikvpb.BatchRaftMessage

• uint64 region_id = 1

• optional metapb.Peer from_peer = 2

• optional metapb.Peer to_peer = 3

• optional eraftpb.Message message = 4

• optional metapb.RegionEpoch region_epoch = 5

• bool is_tombstone = 6

  true means to_peer is a tombstone peer and it should remove itself.

• bytes start_key = 7

  Region key range [start_key, end_key).

• bytes end_key = 8

• optional metapb.Region merge_target = 9

  If it has value, to_peer should be removed if merge is never going to complete.

• optional ExtraMessage extra_msg = 10

• bytes extra_ctx = 11

• disk_usage.DiskUsage disk_usage = 12

• optional metapb.Region region = 1

• uint64 file_size = 2

• repeated KeyValue data = 3

• uint64 version = 4

• optional SnapshotMeta meta = 5

• repeated metapb.Peer removed_records = 6

• repeated MergedRecord merged_records = 7

Used in: RaftApplyState

• uint64 index = 1

• uint64 term = 2

Used in: ExtraMessage

• uint64 version = 1

• repeated bytes keys = 2

• repeated uint64 sizes = 3

Used in: debugpb.RegionInfoResponse, pdpb.PeerReport, raft_cmdpb.CommitMergeRequest, RegionSequenceNumberRelation

• PeerState state = 1

• optional metapb.Region region = 2

• optional MergeState merge_state = 3

• uint64 tablet_index = 4

  The apply index corresponding to the storage when it's initialized.

• repeated metapb.Peer removed_records = 5

  Raft doesn't guarantee peer will be removed in the end. In v1, peer finds out its destiny by logs or broadcast; in v2, leader is responsible to ensure removed peers are destroyed. Note: only peers who has been part of this region can be in this list.

• repeated MergedRecord merged_records = 6

  Merged peer can't be deleted like gc peers. Instead, leader needs to query target peer to decide whether source peer can be destroyed.

• uint64 region_id = 1

• uint64 sequence_number = 2

• optional RaftApplyState apply_state = 3

• optional RegionLocalState region_state = 4

Used in: SnapshotMeta

• string cf = 1

• uint64 size = 2

• uint32 checksum = 3

Used as request type in: tikvpb.Tikv.Snapshot

• optional RaftMessage message = 1

• bytes data = 2

Used in: RaftSnapshotData

• repeated SnapshotCFFile cf_files = 1

• bool for_balance = 2

  true means this snapshot is triggered for load balance

• bool for_witness = 3

  true means this is an empty snapshot for witness

• uint64 start = 4

  the timestamp second to generate snapshot

• uint64 generate_duration_sec = 5

  the duration of generating snapshot

• string tablet_snap_path = 6

  the path of the tablet snapshot, it should only be used for v1 to receive snapshot from v2

• uint64 commit_index_hint = 7

  A hint of the latest commit index on leader when sending snapshot. It should only be used for v2 to send snapshot to v1. See https://github.com/pingcap/tiflash/issues/7568

• uint64 cluster_id = 1

• uint64 store_id = 2

• kvrpcpb.APIVersion api_version = 3

• uint64 seqno = 1

  Used for TiKV start recovery when WAL of KVDB was disabled. TiKV may read all relations between seqno and raft log index, and replay all raft logs which corresponding seqno smaller than the seqno here. After TiKV replays all raft logs and flushed KV data, the seqno here must be updated.

Used in: TabletSnapshotRequest

• uint64 checksum = 1

  Checksum of all data sent in `TabletSnapshotFileChunk.data` and `TabletSnapshotFileChunk.file_name`.

Used in: TabletSnapshotRequest

• uint64 file_size = 1

• string file_name = 2

• bytes data = 3

  Encrypted.

• bytes iv = 4

  Initial vector if encryption is enabled.

• optional encryptionpb.DataKey key = 5

Used in: TabletSnapshotPreview

• uint64 file_size = 1

• string file_name = 2

• bytes head_chunk = 3

  Some block data. Unencrypted.

• bytes trailing_chunk = 4

  trailing data including checksum. Unencrypted.

Used in: TabletSnapshotRequest

• optional RaftMessage message = 1

• bool use_cache = 2

Snapshot preview for server to decide whether skip some files. Server should send back an `AcceptedSnapshotFile` to let client keep sending specified files. Only SST files can be skipped, all other files should always be sent.

Used in: TabletSnapshotRequest

• repeated TabletSnapshotFileMeta metas = 1

• bool end = 2

  There may be too many metas, use a flag to indicate all metas are sent.

Used as request type in: tikvpb.Tikv.TabletSnapshot

• oneof payload

  • TabletSnapshotHead head = 1

  • TabletSnapshotPreview preview = 2

  • TabletSnapshotFileChunk chunk = 3

  • TabletSnapshotEnd end = 4

Used as response type in: tikvpb.Tikv.TabletSnapshot

• optional AcceptedSnapshotFiles files = 1