package schedulingpb

Get desktop application:
View/edit binary Protocol Buffers messages

• rpc StoreHeartbeat (StoreHeartbeatRequest, StoreHeartbeatResponse)

  schedulingpb.proto:21

  message StoreHeartbeatRequest

  schedulingpb.proto:71

  • optional RequestHeader header = 1

  • optional pdpb.StoreStats stats = 2

  message StoreHeartbeatResponse

  schedulingpb.proto:76

  • optional ResponseHeader header = 1

  • string cluster_version = 2

• rpc RegionHeartbeat (stream RegionHeartbeatRequest, stream RegionHeartbeatResponse)

  schedulingpb.proto:23

  message RegionHeartbeatRequest

  schedulingpb.proto:81

  • optional RequestHeader header = 1

  • optional metapb.Region region = 2

  • optional metapb.Peer leader = 3

    Leader Peer sending the heartbeat.

  • uint64 term = 4

    Term is the term of raft group.

  • repeated pdpb.PeerStats down_peers = 5

    Leader considers that these peers are down.

  • repeated metapb.Peer pending_peers = 6

    Pending peers are the peers that the leader can't consider as working followers.

  • uint64 bytes_written = 7

    Bytes read/written during this period.

  • uint64 bytes_read = 8

  • uint64 keys_written = 9

    Keys read/written during this period.

  • uint64 keys_read = 10

  • uint64 approximate_size = 11

    Approximate region size.

  • uint64 approximate_keys = 12

    Approximate number of keys.

  • optional pdpb.QueryStats query_stats = 13

    QueryStats reported write query stats, and there are read query stats in store heartbeat

  • optional pdpb.TimeInterval interval = 14

    Actually reported time interval

  message RegionHeartbeatResponse

  schedulingpb.proto:110

  • optional ResponseHeader header = 1

  • uint64 region_id = 2

    ID of the region

  • optional metapb.RegionEpoch region_epoch = 3

  • optional metapb.Peer target_peer = 4

    Leader of the region at the moment of the corresponding request was made.

  • optional pdpb.ChangePeer change_peer = 5

    Notice, Pd only allows handling reported epoch >= current pd's. Leader peer reports region status with RegionHeartbeatRequest to pd regularly, pd will determine whether this region should do ChangePeer or not. E,g, max peer number is 3, region A, first only peer 1 in A. 1. Pd region state -> Peers (1), ConfVer (1). 2. Leader peer 1 reports region state to pd, pd finds the peer number is < 3, so first changes its current region state -> Peers (1, 2), ConfVer (1), and returns ChangePeer Adding 2. 3. Leader does ChangePeer, then reports Peers (1, 2), ConfVer (2), pd updates its state -> Peers (1, 2), ConfVer (2). 4. Leader may report old Peers (1), ConfVer (1) to pd before ConfChange finished, pd stills responses ChangePeer Adding 2, of course, we must guarantee the second ChangePeer can't be applied in TiKV.

  • optional pdpb.TransferLeader transfer_leader = 6

    Pd can return transfer_leader to let TiKV does leader transfer itself.

  • optional pdpb.Merge merge = 7

  • optional pdpb.SplitRegion split_region = 8

    PD sends split_region to let TiKV split a region into two regions.

  • optional pdpb.ChangePeerV2 change_peer_v2 = 9

    Multiple change peer operations atomically. Note: PD can use both ChangePeer and ChangePeerV2 at the same time (not in the same RegionHeartbeatResponse). Now, PD use ChangePeerV2 in following scenarios: 1. replacing peers 2. demoting voter directly

  • optional pdpb.BatchSwitchWitness switch_witnesses = 10

• rpc SplitRegions (SplitRegionsRequest, SplitRegionsResponse)

  schedulingpb.proto:25

  message SplitRegionsRequest

  schedulingpb.proto:165

  • optional RequestHeader header = 1

  • repeated bytes split_keys = 2

  • uint64 retry_limit = 3

  message SplitRegionsResponse

  schedulingpb.proto:171

  • optional ResponseHeader header = 1

  • uint64 finished_percentage = 2

  • repeated uint64 regions_id = 3

• rpc ScatterRegions (ScatterRegionsRequest, ScatterRegionsResponse)

  schedulingpb.proto:27

  message ScatterRegionsRequest

  schedulingpb.proto:148

  • optional RequestHeader header = 1

  • string group = 2

    If group is defined, the regions with the same group would be scattered as a whole group. If not defined, the regions would be scattered in a cluster level.

  • repeated uint64 regions_id = 3

    If regions_id is defined, the region_id would be ignored.

  • uint64 retry_limit = 4

  • bool skip_store_limit = 5

  message ScatterRegionsResponse

  schedulingpb.proto:160

  • optional ResponseHeader header = 1

  • uint64 finished_percentage = 2

• rpc GetOperator (GetOperatorRequest, GetOperatorResponse)

  schedulingpb.proto:29

  message GetOperatorRequest

  schedulingpb.proto:177

  • optional RequestHeader header = 1

  • uint64 region_id = 2

  message GetOperatorResponse

  schedulingpb.proto:182

  • optional ResponseHeader header = 1

  • uint64 region_id = 2

  • bytes desc = 3

  • pdpb.OperatorStatus status = 4

  • bytes kind = 5

• rpc AskBatchSplit (AskBatchSplitRequest, AskBatchSplitResponse)

  schedulingpb.proto:31

  message AskBatchSplitRequest

  schedulingpb.proto:190

  • optional RequestHeader header = 1

  • optional metapb.Region region = 2

  • uint32 split_count = 3

  message AskBatchSplitResponse

  schedulingpb.proto:197

  • optional ResponseHeader header = 1

  • repeated pdpb.SplitID ids = 2

Used in: ResponseHeader

• ErrorType type = 1

• string message = 2

Used in: Error

• OK = 0

• UNKNOWN = 1

• NOT_BOOTSTRAPPED = 2

• ALREADY_BOOTSTRAPPED = 3

• INVALID_VALUE = 4

• CLUSTER_MISMATCHED = 5

• string name = 1

  name is the unique name of the scheduling participant.

• uint64 id = 2

  id is the unique id of the scheduling participant.

• repeated string listen_urls = 3

  listen_urls is the serivce endpoint list in the url format. listen_urls[0] is primary service endpoint.

Used in: AskBatchSplitRequest, GetOperatorRequest, RegionHeartbeatRequest, ScatterRegionsRequest, SplitRegionsRequest, StoreHeartbeatRequest

• uint64 cluster_id = 1

  cluster_id is the ID of the cluster which be sent to.

• uint64 sender_id = 2

  sender_id is the ID of the sender server.

Used in: AskBatchSplitResponse, GetOperatorResponse, RegionHeartbeatResponse, ScatterRegionsResponse, SplitRegionsResponse, StoreHeartbeatResponse

• uint64 cluster_id = 1

  cluster_id is the ID of the cluster which sent the response.

• optional Error error = 2