etcd/etcdserver/etcdserverpb/rpc.proto

syntax = "proto3";
package etcdserverpb;

import "gogoproto/gogo.proto";
import "etcd/storage/storagepb/kv.proto";

option (gogoproto.marshaler_all) = true;
option (gogoproto.unmarshaler_all) = true;

// Interface exported by the server.
service etcd {
  // Range gets the keys in the range from the store.
  rpc Range(RangeRequest) returns (RangeResponse) {}

  // Put puts the given key into the store.
  // A put request increases the revision of the store,
  // and generates one event in the event history.
  rpc Put(PutRequest) returns (PutResponse) {}

  // Delete deletes the given range from the store.
  // A delete request increase the revision of the store,
  // and generates one event in the event history.
  rpc DeleteRange(DeleteRangeRequest) returns (DeleteRangeResponse) {}

  // Txn processes all the requests in one transaction.
  // A txn request increases the revision of the store,
  // and generates events with the same revision in the event history.
  rpc Txn(TxnRequest) returns (TxnResponse) {}

  // Compact compacts the event history in etcd. User should compact the
  // event history periodically, or it will grow infinitely.
  rpc Compact(CompactionRequest) returns (CompactionResponse) {}
}

service watch {
  // Watch watches the events happening or happened. Both input and output
  // are stream. One watch rpc can watch for multiple keys or prefixs and
  // get a stream of events. The whole events history can be watched unless
  // compacted.
  rpc Watch(stream WatchRequest) returns (stream WatchResponse) {}
}

message ResponseHeader {
  uint64 cluster_id = 1;
  uint64 member_id = 2;
  // revision of the store when the request was applied.
  int64 revision = 3;
  // term of raft when the request was applied.
  uint64 raft_term = 4;
}

message RangeRequest {
  // if the range_end is not given, the request returns the key.
  bytes key = 1;
  // if the range_end is given, it gets the keys in range [key, range_end).
  bytes range_end = 2;
  // limit the number of keys returned.
  int64 limit = 3;
  // range over the store at the given revision.
  // if revision is less or equal to zero, range over the newest store.
  // if the revision has been compacted, ErrCompaction will be returned in
  // response.
  int64 revision = 4;
}

message RangeResponse {
  ResponseHeader header = 1;
  repeated storagepb.KeyValue kvs = 2;
  // more indicates if there are more keys to return in the requested range.
  bool more = 3;
}

message PutRequest {
  bytes key = 1;
  bytes value = 2;
}

message PutResponse {
  ResponseHeader header = 1;
}

message DeleteRangeRequest {
  // if the range_end is not given, the request deletes the key.
  bytes key = 1;
  // if the range_end is given, it deletes the keys in range [key, range_end).
  bytes range_end = 2;
}

message DeleteRangeResponse {
  ResponseHeader header = 1;
}

message RequestUnion {
  oneof request {
    RangeRequest request_range = 1;
    PutRequest request_put = 2;
    DeleteRangeRequest request_delete_range = 3;
  }
}

message ResponseUnion {
  oneof response {
    RangeResponse response_range = 1;
    PutResponse response_put = 2;
    DeleteRangeResponse response_delete_range = 3;
  }
}

message Compare {
  enum CompareResult {
    EQUAL = 0;
    GREATER = 1;
    LESS = 2;
  }
  enum CompareTarget {
    VERSION = 0;
    CREATE = 1;
    MOD = 2;
    VALUE= 3;
  }
  CompareResult result = 1;
  CompareTarget target = 2;
  // key path
  bytes key = 3;
  oneof target_union {
    // version of the given key
    int64 version = 4;
    // create revision of the given key
    int64 create_revision = 5;
    // last modified revision of the given key
    int64 mod_revision = 6;
    // value of the given key
    bytes value = 7;
  }
}

// If the comparisons succeed, then the success requests will be processed in order,
// and the response will contain their respective responses in order.
// If the comparisons fail, then the failure requests will be processed in order,
// and the response will contain their respective responses in order.

// From google paxosdb paper:
// Our implementation hinges around a powerful primitive which we call MultiOp. All other database
// operations except for iteration are implemented as a single call to MultiOp. A MultiOp is applied atomically
// and consists of three components:
// 1. A list of tests called guard. Each test in guard checks a single entry in the database. It may check
// for the absence or presence of a value, or compare with a given value. Two different tests in the guard
// may apply to the same or different entries in the database. All tests in the guard are applied and
// MultiOp returns the results. If all tests are true, MultiOp executes t op (see item 2 below), otherwise
// it executes f op (see item 3 below).
// 2. A list of database operations called t op. Each operation in the list is either an insert, delete, or
// lookup operation, and applies to a single database entry. Two different operations in the list may apply
// to the same or different entries in the database. These operations are executed
// if guard evaluates to
// true.
// 3. A list of database operations called f op. Like t op, but executed if guard evaluates to false.
message TxnRequest {
  repeated Compare compare = 1;
  repeated RequestUnion success = 2;
  repeated RequestUnion failure = 3;
}

message TxnResponse {
  ResponseHeader header = 1;
  bool succeeded = 2;
  repeated ResponseUnion responses = 3;
}

// Compaction compacts the kv store upto the given revision (including).
// It removes the old versions of a key. It keeps the newest version of
// the key even if its latest modification revision is smaller than the given
// revision.
message CompactionRequest {
  int64 revision = 1;
}

message CompactionResponse {
  ResponseHeader header = 1;
}

message WatchRequest {
  // the key to be watched
  bytes key = 1;
  // the prefix to be watched.
  bytes prefix = 2;
  // start_revision is an optional revision (including) to watch from. No start_revision is "now".
  int64 start_revision = 3;
  // TODO: support Range watch?
  // TODO: support notification every time interval or revision increase?
  // TODO: support cancel watch if the server cannot reach with majority?
}

message WatchResponse {
  ResponseHeader header = 1;
  // TODO: support batched events response?
  storagepb.Event event = 2;
}