distsqlrun

package
v1.0.1 Latest Latest
Warning

This package is not in the latest version of its module.

Go to latest
Published: May 18, 2017 License: Apache-2.0 Imports: 45 Imported by: 0

Documentation

Overview

Package distsqlrun is a generated protocol buffer package.

It is generated from these files:

cockroach/pkg/sql/distsqlrun/api.proto
cockroach/pkg/sql/distsqlrun/data.proto
cockroach/pkg/sql/distsqlrun/processors.proto

It has these top-level messages:

SetupFlowRequest
EvalContext
SimpleResponse
ConsumerSignal
DrainRequest
Error
Expression
Ordering
StreamEndpointSpec
InputSyncSpec
OutputRouterSpec
DatumInfo
ProducerHeader
ProducerData
ProducerMessage
RemoteProducerMetadata
ProcessorSpec
PostProcessSpec
ProcessorCoreUnion
NoopCoreSpec
ValuesCoreSpec
TableReaderSpan
TableReaderSpec
JoinReaderSpec
SorterSpec
DistinctSpec
MergeJoinerSpec
HashJoinerSpec
AggregatorSpec
BackfillerSpec
FlowSpec
AlgebraicSetOpSpec

Index

Constants

View Source
const (
	FlowNotStarted flowStatus = iota
	FlowRunning
	FlowFinished
)

Flow status indicators.

View Source
const MinAcceptedVersion = 3

MinAcceptedVersion is the oldest version that the server is compatible with; see above.

View Source
const Version = 3

Version identifies the distsqlrun protocol version.

This version is separate from the main CockroachDB version numbering; it is only changed when the distsqlrun API changes.

The planner populates the version in SetupFlowRequest. A server only accepts requests with versions in the range MinAcceptedVersion to Version.

Is is possible used to provide a "window" of compatibility when new features are added. Example:

  • we start with Version=1; distsqlrun servers with version 1 only accept requests with version 1.
  • a new distsqlrun feature is added; Version is bumped to 2. The planner does not yet use this feature by default; it still issues requests with version 1.
  • MinAcceptedVersion is still 1, i.e. servers with version 2 accept both versions 1 and 2.
  • after an upgrade cycle, we can enable the feature in the planner, requiring version 2.
  • at some later point, we can choose to deprecate version 1 and have servers only accept versions >= 2 (by setting MinAcceptedVersion to 2).

Variables

View Source
var (
	ErrInvalidLengthApi = fmt.Errorf("proto: negative length found during unmarshaling")
	ErrIntOverflowApi   = fmt.Errorf("proto: integer overflow")
)
View Source
var (
	ErrInvalidLengthData = fmt.Errorf("proto: negative length found during unmarshaling")
	ErrIntOverflowData   = fmt.Errorf("proto: integer overflow")
)
View Source
var (
	ErrInvalidLengthProcessors = fmt.Errorf("proto: negative length found during unmarshaling")
	ErrIntOverflowProcessors   = fmt.Errorf("proto: integer overflow")
)
View Source
var AggregatorSpec_Func_name = map[int32]string{
	0:  "IDENT",
	1:  "AVG",
	2:  "BOOL_AND",
	3:  "BOOL_OR",
	4:  "CONCAT_AGG",
	5:  "COUNT",
	7:  "MAX",
	8:  "MIN",
	9:  "STDDEV",
	10: "SUM",
	11: "SUM_INT",
	12: "VARIANCE",
}
View Source
var AggregatorSpec_Func_value = map[string]int32{
	"IDENT":      0,
	"AVG":        1,
	"BOOL_AND":   2,
	"BOOL_OR":    3,
	"CONCAT_AGG": 4,
	"COUNT":      5,
	"MAX":        7,
	"MIN":        8,
	"STDDEV":     9,
	"SUM":        10,
	"SUM_INT":    11,
	"VARIANCE":   12,
}
View Source
var AlgebraicSetOpSpec_SetOpType_name = map[int32]string{
	0: "Except_all",
}
View Source
var AlgebraicSetOpSpec_SetOpType_value = map[string]int32{
	"Except_all": 0,
}
View Source
var BackfillerSpec_Type_name = map[int32]string{
	0: "Invalid",
	1: "Column",
	2: "Index",
}
View Source
var BackfillerSpec_Type_value = map[string]int32{
	"Invalid": 0,
	"Column":  1,
	"Index":   2,
}
View Source
var InputSyncSpec_Type_name = map[int32]string{
	0: "UNORDERED",
	1: "ORDERED",
}
View Source
var InputSyncSpec_Type_value = map[string]int32{
	"UNORDERED": 0,
	"ORDERED":   1,
}
View Source
var JoinType_name = map[int32]string{
	0: "INNER",
	1: "LEFT_OUTER",
	2: "RIGHT_OUTER",
	3: "FULL_OUTER",
}
View Source
var JoinType_value = map[string]int32{
	"INNER":       0,
	"LEFT_OUTER":  1,
	"RIGHT_OUTER": 2,
	"FULL_OUTER":  3,
}
View Source
var Ordering_Column_Direction_name = map[int32]string{
	0: "ASC",
	1: "DESC",
}
View Source
var Ordering_Column_Direction_value = map[string]int32{
	"ASC":  0,
	"DESC": 1,
}
View Source
var OutputRouterSpec_Type_name = map[int32]string{
	0: "PASS_THROUGH",
	1: "MIRROR",
	2: "BY_HASH",
	3: "BY_RANGE",
}
View Source
var OutputRouterSpec_Type_value = map[string]int32{
	"PASS_THROUGH": 0,
	"MIRROR":       1,
	"BY_HASH":      2,
	"BY_RANGE":     3,
}
View Source
var StreamEndpointSpec_Type_name = map[int32]string{
	0: "LOCAL",
	1: "REMOTE",
	2: "SYNC_RESPONSE",
}
View Source
var StreamEndpointSpec_Type_value = map[string]int32{
	"LOCAL":         0,
	"REMOTE":        1,
	"SYNC_RESPONSE": 2,
}

Functions

func ColumnMutationFilter

func ColumnMutationFilter(m sqlbase.DescriptorMutation) bool

ColumnMutationFilter is a filter that allows mutations that add or drop columns.

func CompareEncDatumRowForMerge

func CompareEncDatumRowForMerge(
	lhs, rhs sqlbase.EncDatumRow,
	leftOrdering, rightOrdering sqlbase.ColumnOrdering,
	da *sqlbase.DatumAlloc,
) (int, error)

CompareEncDatumRowForMerge EncDatumRow compares two EncDatumRows for merging. When merging two streams and preserving the order (as in a MergeSort or a MergeJoin) compare the head of the streams, emitting the one that sorts first. It allows for the EncDatumRow to be nil if one of the streams is exhausted (and hence nil). CompareEncDatumRowForMerge returns 0 when both rows are nil, and a nil row is considered greater than any non-nil row. CompareEncDatumRowForMerge assumes that the two rows have the same columns in the same orders, but can handle different ordering directions. It takes a DatumAlloc which is used for decoding if any underlying EncDatum is not yet decoded.

func ConvertBackfillError

func ConvertBackfillError(tableDesc *sqlbase.TableDescriptor, b *client.Batch) error

ConvertBackfillError returns a cleaner SQL error for a failed Batch.

func DrainAndClose

func DrainAndClose(ctx context.Context, dst RowReceiver, cause error, srcs ...RowSource)

DrainAndClose is a version of DrainAndForwardMetadata that drains multiple sources. These sources are assumed to be the only producers left for dst, so dst is closed once they're all exhausted (this is different from DrainAndForwardMetadata).

If cause is specified, it is forwarded to the consumer before all the drain metadata. This is intended to have been the error, if any, that caused the draining.

srcs can be nil.

All errors are forwarded to the producer.

func DrainAndForwardMetadata

func DrainAndForwardMetadata(ctx context.Context, src RowSource, dst RowReceiver)

DrainAndForwardMetadata calls src.ConsumerDone() (thus asking src for draining metadata) and then forwards all the metadata to dst.

When this returns, src has been properly closed (regardless of the presence or absence of an error). dst, however, has not been closed; someone else must call dst.ProducerDone() when all producers have finished draining.

It is OK to call DrainAndForwardMetadata() multiple times concurrently on the same dst (as RowReceiver.Push() is guaranteed to be thread safe).

TODO(andrei): errors seen while draining should be reported to the gateway, but they shouldn't fail a SQL query.

func GeneratePlanDiagram

func GeneratePlanDiagram(flows map[roachpb.NodeID]FlowSpec, w io.Writer) error

GeneratePlanDiagram generates the json data for a flow diagram. There should be one FlowSpec per node. The function assumes that StreamIDs are unique across all flows.

func GeneratePlanDiagramWithURL

func GeneratePlanDiagramWithURL(flows map[roachpb.NodeID]FlowSpec) (string, url.URL, error)

GeneratePlanDiagramWithURL generates the json data for a flow diagram and a URL which encodes the diagram. There should be one FlowSpec per node. The function assumes that StreamIDs are unique across all flows.

func GetAggregateInfo

func GetAggregateInfo(
	fn AggregatorSpec_Func, inputType sqlbase.ColumnType,
) (
	aggregateConstructor func(*parser.EvalContext) parser.AggregateFunc,
	returnType sqlbase.ColumnType,
	err error,
)

GetAggregateInfo returns the aggregate constructor and the return type for the given aggregate function when applied on the given type.

func IndexMutationFilter

func IndexMutationFilter(m sqlbase.DescriptorMutation) bool

IndexMutationFilter is a filter that allows mutations that add indexes.

func ProcessInboundStream

func ProcessInboundStream(
	ctx context.Context, stream DistSQL_FlowStreamServer, firstMsg *ProducerMessage, dst RowReceiver,
) error

ProcessInboundStream receives rows from a DistSQL_FlowStreamServer and sends them to a RowReceiver. Optionally processes an initial StreamMessage that was already received (because the first message contains the flow and stream IDs, it needs to be received before we can get here).

func RegisterDistSQLServer

func RegisterDistSQLServer(s *grpc.Server, srv DistSQLServer)

func SetFlowRequestTrace

func SetFlowRequestTrace(ctx context.Context, req *SetupFlowRequest) error

SetFlowRequestTrace populates req.Trace with the context of the current Span in the context (if any).

func WriteResumeSpan

func WriteResumeSpan(
	ctx context.Context,
	db *client.DB,
	id sqlbase.ID,
	origSpan roachpb.Span,
	resume roachpb.Span,
	mutationIdx int,
) error

WriteResumeSpan writes a checkpoint for the backfill work on origSpan. origSpan is the span of keys that were assigned to be backfilled, resume is the left over work from origSpan.

Types

type AggregatorSpec

type AggregatorSpec struct {
	// The group key is a subset of the columns in the input stream schema on the
	// basis of which we define our groups.
	GroupCols    []uint32                     `protobuf:"varint,2,rep,packed,name=group_cols,json=groupCols" json:"group_cols,omitempty"`
	Aggregations []AggregatorSpec_Aggregation `protobuf:"bytes,3,rep,name=aggregations" json:"aggregations"`
}

AggregatorSpec is the specification for an "aggregator" (processor core type, not the logical plan computation stage). An aggregator performs 'aggregation' in the SQL sense in that it groups rows and computes an aggregate for each group. The group is configured using the group key. The aggregator can be configured with one or more aggregation functions.

The "internal columns" of an Aggregator map 1-1 to the aggregations.

func (*AggregatorSpec) Descriptor

func (*AggregatorSpec) Descriptor() ([]byte, []int)

func (*AggregatorSpec) Marshal

func (m *AggregatorSpec) Marshal() (dAtA []byte, err error)

func (*AggregatorSpec) MarshalTo

func (m *AggregatorSpec) MarshalTo(dAtA []byte) (int, error)

func (*AggregatorSpec) ProtoMessage

func (*AggregatorSpec) ProtoMessage()

func (*AggregatorSpec) Reset

func (m *AggregatorSpec) Reset()

func (*AggregatorSpec) Size

func (m *AggregatorSpec) Size() (n int)

func (*AggregatorSpec) String

func (m *AggregatorSpec) String() string

func (*AggregatorSpec) Unmarshal

func (m *AggregatorSpec) Unmarshal(dAtA []byte) error

type AggregatorSpec_Aggregation

type AggregatorSpec_Aggregation struct {
	Func AggregatorSpec_Func `protobuf:"varint,1,opt,name=func,enum=cockroach.sql.distsqlrun.AggregatorSpec_Func" json:"func"`
	// Aggregation functions with distinct = true functions like you would
	// expect '<FUNC> DISTINCT' to operate, the default behaviour would be
	// the '<FUNC> ALL' operation.
	Distinct bool `protobuf:"varint,2,opt,name=distinct" json:"distinct"`
	// The column index specifies the argument to the aggregator function.
	ColIdx uint32 `protobuf:"varint,3,opt,name=col_idx,json=colIdx" json:"col_idx"`
}

func (*AggregatorSpec_Aggregation) Descriptor

func (*AggregatorSpec_Aggregation) Descriptor() ([]byte, []int)

func (*AggregatorSpec_Aggregation) Marshal

func (m *AggregatorSpec_Aggregation) Marshal() (dAtA []byte, err error)

func (*AggregatorSpec_Aggregation) MarshalTo

func (m *AggregatorSpec_Aggregation) MarshalTo(dAtA []byte) (int, error)

func (*AggregatorSpec_Aggregation) ProtoMessage

func (*AggregatorSpec_Aggregation) ProtoMessage()

func (*AggregatorSpec_Aggregation) Reset

func (m *AggregatorSpec_Aggregation) Reset()

func (*AggregatorSpec_Aggregation) Size

func (m *AggregatorSpec_Aggregation) Size() (n int)

func (*AggregatorSpec_Aggregation) String

func (m *AggregatorSpec_Aggregation) String() string

func (*AggregatorSpec_Aggregation) Unmarshal

func (m *AggregatorSpec_Aggregation) Unmarshal(dAtA []byte) error

type AggregatorSpec_Func

type AggregatorSpec_Func int32

These mirror the aggregate functions supported by sql/parser. See sql/parser/aggregate_builtins.go.

const (
	// The identity function is set to be the default zero-value function,
	// returning the last value added.
	AggregatorSpec_IDENT      AggregatorSpec_Func = 0
	AggregatorSpec_AVG        AggregatorSpec_Func = 1
	AggregatorSpec_BOOL_AND   AggregatorSpec_Func = 2
	AggregatorSpec_BOOL_OR    AggregatorSpec_Func = 3
	AggregatorSpec_CONCAT_AGG AggregatorSpec_Func = 4
	AggregatorSpec_COUNT      AggregatorSpec_Func = 5
	AggregatorSpec_MAX        AggregatorSpec_Func = 7
	AggregatorSpec_MIN        AggregatorSpec_Func = 8
	AggregatorSpec_STDDEV     AggregatorSpec_Func = 9
	AggregatorSpec_SUM        AggregatorSpec_Func = 10
	AggregatorSpec_SUM_INT    AggregatorSpec_Func = 11
	AggregatorSpec_VARIANCE   AggregatorSpec_Func = 12
)

func (AggregatorSpec_Func) Enum

func (AggregatorSpec_Func) EnumDescriptor

func (AggregatorSpec_Func) EnumDescriptor() ([]byte, []int)

func (AggregatorSpec_Func) String

func (x AggregatorSpec_Func) String() string

func (*AggregatorSpec_Func) UnmarshalJSON

func (x *AggregatorSpec_Func) UnmarshalJSON(data []byte) error

type AlgebraicSetOpSpec

type AlgebraicSetOpSpec struct {
	// If the two input streams are both ordered by a common column ordering,
	// that ordering can be used to optimize resource usage in the processor.
	Ordering Ordering                     `protobuf:"bytes,1,opt,name=ordering" json:"ordering"`
	OpType   AlgebraicSetOpSpec_SetOpType `protobuf:"varint,2,opt,name=op_type,json=opType,enum=cockroach.sql.distsqlrun.AlgebraicSetOpSpec_SetOpType" json:"op_type"`
}

AlgebraicSetOpSpec is a specification for algebraic set operations currently only the EXCEPT ALL set operation, but extensible to other set operations. INTERSECT ALL is implemented with HashJoinerSpec, and UNION ALL with a no-op processor. EXCEPT/INTERSECT/UNION use a DISTINCT processor at the end. The two input streams should have the same schema. The ordering of the left stream will be preserved in the output stream.

func (*AlgebraicSetOpSpec) Descriptor

func (*AlgebraicSetOpSpec) Descriptor() ([]byte, []int)

func (*AlgebraicSetOpSpec) Marshal

func (m *AlgebraicSetOpSpec) Marshal() (dAtA []byte, err error)

func (*AlgebraicSetOpSpec) MarshalTo

func (m *AlgebraicSetOpSpec) MarshalTo(dAtA []byte) (int, error)

func (*AlgebraicSetOpSpec) ProtoMessage

func (*AlgebraicSetOpSpec) ProtoMessage()

func (*AlgebraicSetOpSpec) Reset

func (m *AlgebraicSetOpSpec) Reset()

func (*AlgebraicSetOpSpec) Size

func (m *AlgebraicSetOpSpec) Size() (n int)

func (*AlgebraicSetOpSpec) String

func (m *AlgebraicSetOpSpec) String() string

func (*AlgebraicSetOpSpec) Unmarshal

func (m *AlgebraicSetOpSpec) Unmarshal(dAtA []byte) error

type AlgebraicSetOpSpec_SetOpType

type AlgebraicSetOpSpec_SetOpType int32
const (
	AlgebraicSetOpSpec_Except_all AlgebraicSetOpSpec_SetOpType = 0
)

func (AlgebraicSetOpSpec_SetOpType) Enum

func (AlgebraicSetOpSpec_SetOpType) EnumDescriptor

func (AlgebraicSetOpSpec_SetOpType) EnumDescriptor() ([]byte, []int)

func (AlgebraicSetOpSpec_SetOpType) String

func (*AlgebraicSetOpSpec_SetOpType) UnmarshalJSON

func (x *AlgebraicSetOpSpec_SetOpType) UnmarshalJSON(data []byte) error

type BackfillerSpec

type BackfillerSpec struct {
	Type  BackfillerSpec_Type                    `protobuf:"varint,1,opt,name=type,enum=cockroach.sql.distsqlrun.BackfillerSpec_Type" json:"type"`
	Table cockroach_sql_sqlbase1.TableDescriptor `protobuf:"bytes,2,opt,name=table" json:"table"`
	// Sections of the table to be backfilled.
	Spans []TableReaderSpan `protobuf:"bytes,3,rep,name=spans" json:"spans"`
	// Run the backfill for approximately this duration.
	// The backfill will always process at least one backfill chunk.
	Duration time.Duration `protobuf:"varint,4,opt,name=duration,casttype=time.Duration" json:"duration"`
	// The backfill involves a complete table scan in chunks,
	// where each chunk is a transactional read of a set of rows
	// along with a backfill for the rows. This is the maximum number
	// of entries backfilled per chunk.
	ChunkSize int64 `protobuf:"varint,5,opt,name=chunk_size,json=chunkSize" json:"chunk_size"`
	// Any other (leased) table descriptors necessary for the
	// backfiller to do its job, such as the descriptors for tables with fk
	// relationships to the table being modified.
	OtherTables []cockroach_sql_sqlbase1.TableDescriptor `protobuf:"bytes,6,rep,name=other_tables,json=otherTables" json:"other_tables"`
}

BackfillerSpec is the specification for a "schema change backfiller". The created backfill processor runs a backfill for the first mutations in the table descriptor mutation list with the same mutation id and type. A backfiller processor performs KV operations to retrieve rows for a table and backfills the new indexes/columns contained in the table descriptor. It checkpoints its progress by updating the table descriptor in the database, and doesn't emit any rows nor support any post-processing.

func (*BackfillerSpec) Descriptor

func (*BackfillerSpec) Descriptor() ([]byte, []int)

func (*BackfillerSpec) Marshal

func (m *BackfillerSpec) Marshal() (dAtA []byte, err error)

func (*BackfillerSpec) MarshalTo

func (m *BackfillerSpec) MarshalTo(dAtA []byte) (int, error)

func (*BackfillerSpec) ProtoMessage

func (*BackfillerSpec) ProtoMessage()

func (*BackfillerSpec) Reset

func (m *BackfillerSpec) Reset()

func (*BackfillerSpec) Size

func (m *BackfillerSpec) Size() (n int)

func (*BackfillerSpec) String

func (m *BackfillerSpec) String() string

func (*BackfillerSpec) Unmarshal

func (m *BackfillerSpec) Unmarshal(dAtA []byte) error

type BackfillerSpec_Type

type BackfillerSpec_Type int32
const (
	BackfillerSpec_Invalid BackfillerSpec_Type = 0
	BackfillerSpec_Column  BackfillerSpec_Type = 1
	BackfillerSpec_Index   BackfillerSpec_Type = 2
)

func (BackfillerSpec_Type) Enum

func (BackfillerSpec_Type) EnumDescriptor

func (BackfillerSpec_Type) EnumDescriptor() ([]byte, []int)

func (BackfillerSpec_Type) String

func (x BackfillerSpec_Type) String() string

func (*BackfillerSpec_Type) UnmarshalJSON

func (x *BackfillerSpec_Type) UnmarshalJSON(data []byte) error

type BufferedRecord

type BufferedRecord struct {
	Row  sqlbase.EncDatumRow
	Meta ProducerMetadata
}

BufferedRecord represents a row or metadata record that has been buffered inside a RowBuffer.

type ConsumerSignal

type ConsumerSignal struct {
	// The consumer is done (doesn't need to consume any more rows) and is asking
	// the producer to push whatever trailing metadata it has and close its
	// stream.
	DrainRequest *DrainRequest `protobuf:"bytes,1,opt,name=drain_request,json=drainRequest" json:"drain_request,omitempty"`
	// Used in the RunSyncFlow case; the first message on the client stream must
	// contain this message.
	SetupFlowRequest *SetupFlowRequest `protobuf:"bytes,2,opt,name=setup_flow_request,json=setupFlowRequest" json:"setup_flow_request,omitempty"`
}

func (*ConsumerSignal) Descriptor

func (*ConsumerSignal) Descriptor() ([]byte, []int)

func (*ConsumerSignal) GetValue

func (this *ConsumerSignal) GetValue() interface{}

func (*ConsumerSignal) Marshal

func (m *ConsumerSignal) Marshal() (dAtA []byte, err error)

func (*ConsumerSignal) MarshalTo

func (m *ConsumerSignal) MarshalTo(dAtA []byte) (int, error)

func (*ConsumerSignal) ProtoMessage

func (*ConsumerSignal) ProtoMessage()

func (*ConsumerSignal) Reset

func (m *ConsumerSignal) Reset()

func (*ConsumerSignal) SetValue

func (this *ConsumerSignal) SetValue(value interface{}) bool

func (*ConsumerSignal) Size

func (m *ConsumerSignal) Size() (n int)

func (*ConsumerSignal) String

func (m *ConsumerSignal) String() string

func (*ConsumerSignal) Unmarshal

func (m *ConsumerSignal) Unmarshal(dAtA []byte) error

type ConsumerStatus

type ConsumerStatus uint32

ConsumerStatus is the type returned by RowReceiver.Push(), informing a producer of a consumer's state.

const (
	// NeedMoreRows indicates that the consumer is still expecting more rows.
	NeedMoreRows ConsumerStatus = iota
	// DrainRequested indicates that the consumer will not process any more data
	// rows, but will accept trailing metadata from the producer.
	DrainRequested
	// ConsumerClosed indicates that the consumer will not process any more data
	// rows or metadata. This is also commonly returned in case the consumer has
	// encountered an error.
	ConsumerClosed
)

type DatumInfo

type DatumInfo struct {
	Encoding cockroach_sql_sqlbase2.DatumEncoding `protobuf:"varint,1,opt,name=encoding,enum=cockroach.sql.sqlbase.DatumEncoding" json:"encoding"`
	Type     cockroach_sql_sqlbase1.ColumnType    `protobuf:"bytes,2,opt,name=type" json:"type"`
}

func (*DatumInfo) Descriptor

func (*DatumInfo) Descriptor() ([]byte, []int)

func (*DatumInfo) Marshal

func (m *DatumInfo) Marshal() (dAtA []byte, err error)

func (*DatumInfo) MarshalTo

func (m *DatumInfo) MarshalTo(dAtA []byte) (int, error)

func (*DatumInfo) ProtoMessage

func (*DatumInfo) ProtoMessage()

func (*DatumInfo) Reset

func (m *DatumInfo) Reset()

func (*DatumInfo) Size

func (m *DatumInfo) Size() (n int)

func (*DatumInfo) String

func (m *DatumInfo) String() string

func (*DatumInfo) Unmarshal

func (m *DatumInfo) Unmarshal(dAtA []byte) error

type DistSQLClient

type DistSQLClient interface {
	// RunSyncFlow instantiates a flow and streams back results of that flow.
	// The request must contain one flow, and that flow must have a single mailbox
	// of the special sync response type.
	RunSyncFlow(ctx context.Context, opts ...grpc.CallOption) (DistSQL_RunSyncFlowClient, error)
	// SetupFlow instantiates a flow (subgraphs of a distributed SQL
	// computation) on the receiving node.
	SetupFlow(ctx context.Context, in *SetupFlowRequest, opts ...grpc.CallOption) (*SimpleResponse, error)
	// FlowStream is used to push a stream of messages that is part of a flow. The
	// first message will have a StreamHeader which identifies the flow and the
	// stream (mailbox).
	//
	// The response is a stream because the consumer can signal the producer at
	// any point to start draining.
	FlowStream(ctx context.Context, opts ...grpc.CallOption) (DistSQL_FlowStreamClient, error)
}

func NewDistSQLClient

func NewDistSQLClient(cc *grpc.ClientConn) DistSQLClient

type DistSQLServer

type DistSQLServer interface {
	// RunSyncFlow instantiates a flow and streams back results of that flow.
	// The request must contain one flow, and that flow must have a single mailbox
	// of the special sync response type.
	RunSyncFlow(DistSQL_RunSyncFlowServer) error
	// SetupFlow instantiates a flow (subgraphs of a distributed SQL
	// computation) on the receiving node.
	SetupFlow(context.Context, *SetupFlowRequest) (*SimpleResponse, error)
	// FlowStream is used to push a stream of messages that is part of a flow. The
	// first message will have a StreamHeader which identifies the flow and the
	// stream (mailbox).
	//
	// The response is a stream because the consumer can signal the producer at
	// any point to start draining.
	FlowStream(DistSQL_FlowStreamServer) error
}

type DistSQL_FlowStreamClient

type DistSQL_FlowStreamClient interface {
	Send(*ProducerMessage) error
	Recv() (*ConsumerSignal, error)
	grpc.ClientStream
}

type DistSQL_FlowStreamServer

type DistSQL_FlowStreamServer interface {
	Send(*ConsumerSignal) error
	Recv() (*ProducerMessage, error)
	grpc.ServerStream
}

type DistSQL_RunSyncFlowClient

type DistSQL_RunSyncFlowClient interface {
	Send(*ConsumerSignal) error
	Recv() (*ProducerMessage, error)
	grpc.ClientStream
}

type DistSQL_RunSyncFlowServer

type DistSQL_RunSyncFlowServer interface {
	Send(*ProducerMessage) error
	Recv() (*ConsumerSignal, error)
	grpc.ServerStream
}

type DistinctSpec

type DistinctSpec struct {
	// The ordered columns in the input stream can be optionally specified for
	// possible optimizations. The specific ordering (ascending/descending) of
	// the column itself is not important nor is the order in which the columns
	// are specified.
	OrderedColumns []uint32 `protobuf:"varint,1,rep,name=ordered_columns,json=orderedColumns" json:"ordered_columns,omitempty"`
	// The distinct columns in the input stream are those columns on which we
	// check for distinct rows. If A,B,C are in distinct_columns and there is a
	// 4th column D which is not included in distinct_columns, its values are not
	// considered, so rows A1,B1,C1,D1 and A1,B1,C1,D2 are considered equal and
	// only one of them (the first) is output.
	DistinctColumns []uint32 `protobuf:"varint,2,rep,name=distinct_columns,json=distinctColumns" json:"distinct_columns,omitempty"`
}

func (*DistinctSpec) Descriptor

func (*DistinctSpec) Descriptor() ([]byte, []int)

func (*DistinctSpec) Marshal

func (m *DistinctSpec) Marshal() (dAtA []byte, err error)

func (*DistinctSpec) MarshalTo

func (m *DistinctSpec) MarshalTo(dAtA []byte) (int, error)

func (*DistinctSpec) ProtoMessage

func (*DistinctSpec) ProtoMessage()

func (*DistinctSpec) Reset

func (m *DistinctSpec) Reset()

func (*DistinctSpec) Size

func (m *DistinctSpec) Size() (n int)

func (*DistinctSpec) String

func (m *DistinctSpec) String() string

func (*DistinctSpec) Unmarshal

func (m *DistinctSpec) Unmarshal(dAtA []byte) error

type DrainRequest

type DrainRequest struct {
}

func (*DrainRequest) Descriptor

func (*DrainRequest) Descriptor() ([]byte, []int)

func (*DrainRequest) Marshal

func (m *DrainRequest) Marshal() (dAtA []byte, err error)

func (*DrainRequest) MarshalTo

func (m *DrainRequest) MarshalTo(dAtA []byte) (int, error)

func (*DrainRequest) ProtoMessage

func (*DrainRequest) ProtoMessage()

func (*DrainRequest) Reset

func (m *DrainRequest) Reset()

func (*DrainRequest) Size

func (m *DrainRequest) Size() (n int)

func (*DrainRequest) String

func (m *DrainRequest) String() string

func (*DrainRequest) Unmarshal

func (m *DrainRequest) Unmarshal(dAtA []byte) error

type Error

type Error struct {
	// Types that are valid to be assigned to Detail:
	//	*Error_PGError
	//	*Error_RetryableTxnError
	Detail isError_Detail `protobuf_oneof:"detail"`
}

Error is a generic representation including a string message.

func NewError

func NewError(err error) *Error

NewError creates an Error from an error, to be sent on the wire. It will recognize certain errors and marshall them accordingly, and everything unrecognized is turned into a PGError with code "internal".

func (*Error) Descriptor

func (*Error) Descriptor() ([]byte, []int)

func (*Error) ErrorDetail

func (e *Error) ErrorDetail() error

ErrorDetail returns the payload as a Go error.

func (*Error) GetDetail

func (m *Error) GetDetail() isError_Detail

func (*Error) GetPGError

func (m *Error) GetPGError() *cockroach_pgerror.Error

func (*Error) GetRetryableTxnError

func (m *Error) GetRetryableTxnError() *cockroach_roachpb2.UnhandledRetryableError

func (*Error) Marshal

func (m *Error) Marshal() (dAtA []byte, err error)

func (*Error) MarshalTo

func (m *Error) MarshalTo(dAtA []byte) (int, error)

func (*Error) ProtoMessage

func (*Error) ProtoMessage()

func (*Error) Reset

func (m *Error) Reset()

func (*Error) Size

func (m *Error) Size() (n int)

func (*Error) String

func (e *Error) String() string

String implements fmt.Stringer.

func (*Error) Unmarshal

func (m *Error) Unmarshal(dAtA []byte) error

func (*Error) XXX_OneofFuncs

func (*Error) XXX_OneofFuncs() (func(msg proto.Message, b *proto.Buffer) error, func(msg proto.Message, tag, wire int, b *proto.Buffer) (bool, error), func(msg proto.Message) (n int), []interface{})

XXX_OneofFuncs is for the internal use of the proto package.

type Error_PGError

type Error_PGError struct {
	PGError *cockroach_pgerror.Error `protobuf:"bytes,1,opt,name=pg_error,json=pgError,oneof"`
}

func (*Error_PGError) MarshalTo

func (m *Error_PGError) MarshalTo(dAtA []byte) (int, error)

func (*Error_PGError) Size

func (m *Error_PGError) Size() (n int)

type Error_RetryableTxnError

type Error_RetryableTxnError struct {
	RetryableTxnError *cockroach_roachpb2.UnhandledRetryableError `protobuf:"bytes,2,opt,name=retryableTxnError,oneof"`
}

func (*Error_RetryableTxnError) MarshalTo

func (m *Error_RetryableTxnError) MarshalTo(dAtA []byte) (int, error)

func (*Error_RetryableTxnError) Size

func (m *Error_RetryableTxnError) Size() (n int)

type EvalContext

type EvalContext struct {
	StmtTimestampNanos int64                        `protobuf:"varint,1,opt,name=stmtTimestampNanos" json:"stmtTimestampNanos"`
	TxnTimestampNanos  int64                        `protobuf:"varint,2,opt,name=txnTimestampNanos" json:"txnTimestampNanos"`
	ClusterTimestamp   cockroach_util_hlc.Timestamp `protobuf:"bytes,3,opt,name=clusterTimestamp" json:"clusterTimestamp"`
	// The name of the location according to whose current timezone we're going to
	// parse timestamps. Used to init EvalContext.Location.
	Location   string   `protobuf:"bytes,4,opt,name=location" json:"location"`
	Database   string   `protobuf:"bytes,5,opt,name=database" json:"database"`
	SearchPath []string `protobuf:"bytes,6,rep,name=searchPath" json:"searchPath,omitempty"`
}

EvalContext is used to marshall some planner.EvalContext members.

func MakeEvalContext

func MakeEvalContext(evalCtx parser.EvalContext) EvalContext

MakeEvalContext serializes some of the fields of a parser.EvalContext into a distsqlrun.EvalContext proto.

func (*EvalContext) Descriptor

func (*EvalContext) Descriptor() ([]byte, []int)

func (*EvalContext) Marshal

func (m *EvalContext) Marshal() (dAtA []byte, err error)

func (*EvalContext) MarshalTo

func (m *EvalContext) MarshalTo(dAtA []byte) (int, error)

func (*EvalContext) ProtoMessage

func (*EvalContext) ProtoMessage()

func (*EvalContext) Reset

func (m *EvalContext) Reset()

func (*EvalContext) Size

func (m *EvalContext) Size() (n int)

func (*EvalContext) String

func (m *EvalContext) String() string

func (*EvalContext) Unmarshal

func (m *EvalContext) Unmarshal(dAtA []byte) error

type Expression

type Expression struct {
	// TODO(radu): TBD how this will be used
	Version string `protobuf:"bytes,1,opt,name=version" json:"version"`
	// SQL expressions are passed as a string, with ordinal references
	// (@1, @2, @3 ..) used for "input" variables.
	Expr string `protobuf:"bytes,2,opt,name=expr" json:"expr"`
}

func (*Expression) Descriptor

func (*Expression) Descriptor() ([]byte, []int)

func (*Expression) Marshal

func (m *Expression) Marshal() (dAtA []byte, err error)

func (*Expression) MarshalTo

func (m *Expression) MarshalTo(dAtA []byte) (int, error)

func (*Expression) ProtoMessage

func (*Expression) ProtoMessage()

func (*Expression) Reset

func (m *Expression) Reset()

func (*Expression) Size

func (m *Expression) Size() (n int)

func (*Expression) String

func (m *Expression) String() string

func (*Expression) Unmarshal

func (m *Expression) Unmarshal(dAtA []byte) error

type Flow

type Flow struct {
	FlowCtx
	// contains filtered or unexported fields
}

Flow represents a flow which consists of processors and streams.

func (*Flow) Cleanup

func (f *Flow) Cleanup(ctx context.Context)

Cleanup should be called when the flow completes (after all processors and mailboxes exited).

func (*Flow) RunSync

func (f *Flow) RunSync(ctx context.Context)

RunSync runs the processors in the flow in order (serially), in the same context (no goroutines are spawned).

func (*Flow) Start

func (f *Flow) Start(ctx context.Context, doneFn func())

Start starts the flow (each processor runs in their own goroutine).

func (*Flow) Wait

func (f *Flow) Wait()

Wait waits for all the goroutines for this flow to exit.

type FlowCtx

type FlowCtx struct {
	log.AmbientContext
	// contains filtered or unexported fields
}

FlowCtx encompasses the contexts needed for various flow components.

type FlowID

type FlowID struct {
	uuid.UUID
}

FlowID identifies a flow. It is most importantly used when setting up streams between nodes.

type FlowSpec

type FlowSpec struct {
	FlowID     FlowID          `protobuf:"bytes,1,opt,name=flow_id,json=flowId,customtype=FlowID" json:"flow_id"`
	Processors []ProcessorSpec `protobuf:"bytes,2,rep,name=processors" json:"processors"`
}

FlowSpec describes a "flow" which is a subgraph of a distributed SQL computation consisting of processors and streams.

func (*FlowSpec) Descriptor

func (*FlowSpec) Descriptor() ([]byte, []int)

func (*FlowSpec) Marshal

func (m *FlowSpec) Marshal() (dAtA []byte, err error)

func (*FlowSpec) MarshalTo

func (m *FlowSpec) MarshalTo(dAtA []byte) (int, error)

func (*FlowSpec) ProtoMessage

func (*FlowSpec) ProtoMessage()

func (*FlowSpec) Reset

func (m *FlowSpec) Reset()

func (*FlowSpec) Size

func (m *FlowSpec) Size() (n int)

func (*FlowSpec) String

func (m *FlowSpec) String() string

func (*FlowSpec) Unmarshal

func (m *FlowSpec) Unmarshal(dAtA []byte) error

type HashJoinerSpec

type HashJoinerSpec struct {
	// The join constraints certain columns from the left stream to equal
	// corresponding columns on the right stream. These must have the same length.
	LeftEqColumns  []uint32 `protobuf:"varint,1,rep,packed,name=left_eq_columns,json=leftEqColumns" json:"left_eq_columns,omitempty"`
	RightEqColumns []uint32 `protobuf:"varint,2,rep,packed,name=right_eq_columns,json=rightEqColumns" json:"right_eq_columns,omitempty"`
	// "ON" expression (in addition to the equality constraints captured by the
	// orderings). Assuming that the left stream has N columns and the right
	// stream has M columns, in this expression variables @1 to @N refer to
	// columns of the left stream and variables @N to @(N+M) refer to columns in
	// the right stream.
	OnExpr Expression `protobuf:"bytes,5,opt,name=on_expr,json=onExpr" json:"on_expr"`
	Type   JoinType   `protobuf:"varint,6,opt,name=type,enum=cockroach.sql.distsqlrun.JoinType" json:"type"`
}

HashJoinerSpec is the specification for a hash join processor. The processor has two inputs and one output.

The processor works by reading the entire right input and putting it in a hash table. Thus, there is no guarantee on the ordering of results that stem only from the right input (in the case of RIGHT_OUTER, FULL_OUTER). However, it is guaranteed that results that involve the left stream preserve the ordering; i.e. all results that stem from left row (i) precede results that stem from left row (i+1).

The "internal columns" of a HashJoiner (see ProcessorSpec) are the concatenation of left input columns and right input columns. If the left input has N columns and the right input has M columns, the first N columns contain values from the left side and the following M columns contain values from the right side.

func (*HashJoinerSpec) Descriptor

func (*HashJoinerSpec) Descriptor() ([]byte, []int)

func (*HashJoinerSpec) Marshal

func (m *HashJoinerSpec) Marshal() (dAtA []byte, err error)

func (*HashJoinerSpec) MarshalTo

func (m *HashJoinerSpec) MarshalTo(dAtA []byte) (int, error)

func (*HashJoinerSpec) ProtoMessage

func (*HashJoinerSpec) ProtoMessage()

func (*HashJoinerSpec) Reset

func (m *HashJoinerSpec) Reset()

func (*HashJoinerSpec) Size

func (m *HashJoinerSpec) Size() (n int)

func (*HashJoinerSpec) String

func (m *HashJoinerSpec) String() string

func (*HashJoinerSpec) Unmarshal

func (m *HashJoinerSpec) Unmarshal(dAtA []byte) error

type InputSyncSpec

type InputSyncSpec struct {
	Type     InputSyncSpec_Type   `protobuf:"varint,1,opt,name=type,enum=cockroach.sql.distsqlrun.InputSyncSpec_Type" json:"type"`
	Ordering Ordering             `protobuf:"bytes,2,opt,name=ordering" json:"ordering"`
	Streams  []StreamEndpointSpec `protobuf:"bytes,3,rep,name=streams" json:"streams"`
	// Schema for the streams entering this synchronizer.
	ColumnTypes []cockroach_sql_sqlbase1.ColumnType `protobuf:"bytes,4,rep,name=column_types,json=columnTypes" json:"column_types"`
}

InputSyncSpec is the specification for an input synchronizer; it decides how to interleave rows from multiple input streams.

func (*InputSyncSpec) Descriptor

func (*InputSyncSpec) Descriptor() ([]byte, []int)

func (*InputSyncSpec) Marshal

func (m *InputSyncSpec) Marshal() (dAtA []byte, err error)

func (*InputSyncSpec) MarshalTo

func (m *InputSyncSpec) MarshalTo(dAtA []byte) (int, error)

func (*InputSyncSpec) ProtoMessage

func (*InputSyncSpec) ProtoMessage()

func (*InputSyncSpec) Reset

func (m *InputSyncSpec) Reset()

func (*InputSyncSpec) Size

func (m *InputSyncSpec) Size() (n int)

func (*InputSyncSpec) String

func (m *InputSyncSpec) String() string

func (*InputSyncSpec) Unmarshal

func (m *InputSyncSpec) Unmarshal(dAtA []byte) error

type InputSyncSpec_Type

type InputSyncSpec_Type int32
const (
	// Rows from the input streams are interleaved arbitrarily.
	InputSyncSpec_UNORDERED InputSyncSpec_Type = 0
	// The input streams are guaranteed to be ordered according to the column
	// ordering field; rows from the streams are interleaved to preserve that
	// ordering.
	InputSyncSpec_ORDERED InputSyncSpec_Type = 1
)

func (InputSyncSpec_Type) Enum

func (InputSyncSpec_Type) EnumDescriptor

func (InputSyncSpec_Type) EnumDescriptor() ([]byte, []int)

func (InputSyncSpec_Type) String

func (x InputSyncSpec_Type) String() string

func (*InputSyncSpec_Type) UnmarshalJSON

func (x *InputSyncSpec_Type) UnmarshalJSON(data []byte) error

type JoinReaderSpec

type JoinReaderSpec struct {
	Table cockroach_sql_sqlbase1.TableDescriptor `protobuf:"bytes,1,opt,name=table" json:"table"`
	// If 0, we use the primary index; each row in the input stream has a value
	// for each primary key.
	// TODO(radu): figure out the correct semantics when joining with an index.
	IndexIdx uint32 `protobuf:"varint,2,opt,name=index_idx,json=indexIdx" json:"index_idx"`
}

JoinReaderSpec is the specification for a "join reader". A join reader performs KV operations to retrieve specific rows that correspond to the values in the input stream (join by lookup).

The "internal columns" of a JoinReader (see ProcessorSpec) are all the columns of the table. Internally, only the values for the columns needed by the post-processing stage are be populated.

func (*JoinReaderSpec) Descriptor

func (*JoinReaderSpec) Descriptor() ([]byte, []int)

func (*JoinReaderSpec) Marshal

func (m *JoinReaderSpec) Marshal() (dAtA []byte, err error)

func (*JoinReaderSpec) MarshalTo

func (m *JoinReaderSpec) MarshalTo(dAtA []byte) (int, error)

func (*JoinReaderSpec) ProtoMessage

func (*JoinReaderSpec) ProtoMessage()

func (*JoinReaderSpec) Reset

func (m *JoinReaderSpec) Reset()

func (*JoinReaderSpec) Size

func (m *JoinReaderSpec) Size() (n int)

func (*JoinReaderSpec) String

func (m *JoinReaderSpec) String() string

func (*JoinReaderSpec) Unmarshal

func (m *JoinReaderSpec) Unmarshal(dAtA []byte) error

type JoinType

type JoinType int32
const (
	JoinType_INNER       JoinType = 0
	JoinType_LEFT_OUTER  JoinType = 1
	JoinType_RIGHT_OUTER JoinType = 2
	JoinType_FULL_OUTER  JoinType = 3
)

func (JoinType) Enum

func (x JoinType) Enum() *JoinType

func (JoinType) EnumDescriptor

func (JoinType) EnumDescriptor() ([]byte, []int)

func (JoinType) String

func (x JoinType) String() string

func (*JoinType) UnmarshalJSON

func (x *JoinType) UnmarshalJSON(data []byte) error

type MergeJoinerSpec

type MergeJoinerSpec struct {
	// The streams must be ordered according to the columns that have equality
	// constraints. The first column of the left ordering is constrained to be
	// equal to the first column in the right ordering and so on. The ordering
	// lengths and directions must match.
	// In the example above, left ordering describes C1+,C2- and right ordering
	// describes C5+,C4-.
	LeftOrdering  Ordering `protobuf:"bytes,1,opt,name=left_ordering,json=leftOrdering" json:"left_ordering"`
	RightOrdering Ordering `protobuf:"bytes,2,opt,name=right_ordering,json=rightOrdering" json:"right_ordering"`
	// "ON" expression (in addition to the equality constraints captured by the
	// orderings). Assuming that the left stream has N columns and the right
	// stream has M columns, in this expression ordinal references @1 to @N refer
	// to columns of the left stream and variables @(N+1) to @(N+M) refer to
	// columns in the right stream.
	OnExpr Expression `protobuf:"bytes,5,opt,name=on_expr,json=onExpr" json:"on_expr"`
	Type   JoinType   `protobuf:"varint,6,opt,name=type,enum=cockroach.sql.distsqlrun.JoinType" json:"type"`
}

MergeJoinerSpec is the specification for a merge join processor. The processor has two inputs and one output. The inputs must have the same ordering on the columns that have equality constraints. For example:

SELECT * FROM T1 INNER JOIN T2 ON T1.C1 = T2.C5 AND T1.C2 = T2.C4

To perform a merge join, the streams corresponding to T1 and T2 must have the same ordering on columns C1, C2 and C5, C4 respectively. For example: C1+,C2- and C5+,C4-.

The "internal columns" of a MergeJoiner (see ProcessorSpec) are the concatenation of left input columns and right input columns. If the left input has N columns and the right input has M columns, the first N columns contain values from the left side and the following M columns contain values from the right side.

func (*MergeJoinerSpec) Descriptor

func (*MergeJoinerSpec) Descriptor() ([]byte, []int)

func (*MergeJoinerSpec) Marshal

func (m *MergeJoinerSpec) Marshal() (dAtA []byte, err error)

func (*MergeJoinerSpec) MarshalTo

func (m *MergeJoinerSpec) MarshalTo(dAtA []byte) (int, error)

func (*MergeJoinerSpec) ProtoMessage

func (*MergeJoinerSpec) ProtoMessage()

func (*MergeJoinerSpec) Reset

func (m *MergeJoinerSpec) Reset()

func (*MergeJoinerSpec) Size

func (m *MergeJoinerSpec) Size() (n int)

func (*MergeJoinerSpec) String

func (m *MergeJoinerSpec) String() string

func (*MergeJoinerSpec) Unmarshal

func (m *MergeJoinerSpec) Unmarshal(dAtA []byte) error

type MultiplexedRowChannel

type MultiplexedRowChannel struct {
	// contains filtered or unexported fields
}

MultiplexedRowChannel is a RowChannel wrapper which allows multiple row producers to push rows on the same channel.

func (*MultiplexedRowChannel) ConsumerClosed

func (mrc *MultiplexedRowChannel) ConsumerClosed()

ConsumerClosed is part of the RowSource interface.

func (*MultiplexedRowChannel) ConsumerDone

func (mrc *MultiplexedRowChannel) ConsumerDone()

ConsumerDone is part of the RowSource interface.

func (*MultiplexedRowChannel) Init

func (mrc *MultiplexedRowChannel) Init(numSenders int, types []sqlbase.ColumnType)

Init initializes the MultiplexedRowChannel with the default buffer size.

func (*MultiplexedRowChannel) Next

Next is part of the RowSource interface.

func (*MultiplexedRowChannel) ProducerDone

func (mrc *MultiplexedRowChannel) ProducerDone()

ProducerDone is part of the RowReceiver interface.

func (*MultiplexedRowChannel) Push

Push is part of the RowReceiver interface.

func (*MultiplexedRowChannel) Types

func (mrc *MultiplexedRowChannel) Types() []sqlbase.ColumnType

Types is part of the RowSource interface.

type MutationFilter

type MutationFilter func(sqlbase.DescriptorMutation) bool

MutationFilter is the type of a simple predicate on a mutation.

type NoMetadataRowSource

type NoMetadataRowSource struct {
	// contains filtered or unexported fields
}

NoMetadataRowSource is a wrapper on top of a RowSource that automatically forwards metadata to a RowReceiver. Data rows are returned through an interface similar to RowSource, except that, since metadata is taken care of, only the data rows are returned.

The point of this struct is that it'd be burdensome for some row consumers to have to deal with metadata.

func MakeNoMetadataRowSource

func MakeNoMetadataRowSource(src RowSource, sink RowReceiver) NoMetadataRowSource

MakeNoMetadataRowSource builds a NoMetadataRowSource.

func (*NoMetadataRowSource) NextRow

func (rs *NoMetadataRowSource) NextRow() (sqlbase.EncDatumRow, error)

NextRow is analogous to RowSource.Next. If the producer sends an error, we can't just forward it to metadataSink. We need to let the consumer know so that it's not under the impression that everything is hunky-dory and it can continue consuming rows. So, this interface returns the error. Just like with a raw RowSource, the consumer should generally call ConsumerDone() and drain.

type NoopCoreSpec

type NoopCoreSpec struct {
}

NoopCoreSpec indicates a "no-op" processor core. This is used when we just need post-processing or when only a synchronizer is required (e.g. at the final endpoint).

func (*NoopCoreSpec) Descriptor

func (*NoopCoreSpec) Descriptor() ([]byte, []int)

func (*NoopCoreSpec) Marshal

func (m *NoopCoreSpec) Marshal() (dAtA []byte, err error)

func (*NoopCoreSpec) MarshalTo

func (m *NoopCoreSpec) MarshalTo(dAtA []byte) (int, error)

func (*NoopCoreSpec) ProtoMessage

func (*NoopCoreSpec) ProtoMessage()

func (*NoopCoreSpec) Reset

func (m *NoopCoreSpec) Reset()

func (*NoopCoreSpec) Size

func (m *NoopCoreSpec) Size() (n int)

func (*NoopCoreSpec) String

func (m *NoopCoreSpec) String() string

func (*NoopCoreSpec) Unmarshal

func (m *NoopCoreSpec) Unmarshal(dAtA []byte) error

type Ordering

type Ordering struct {
	Columns []Ordering_Column `protobuf:"bytes,1,rep,name=columns" json:"columns"`
}

Ordering defines an order - specifically a list of column indices and directions. See sqlbase.ColumnOrdering.

func (*Ordering) Descriptor

func (*Ordering) Descriptor() ([]byte, []int)

func (*Ordering) Marshal

func (m *Ordering) Marshal() (dAtA []byte, err error)

func (*Ordering) MarshalTo

func (m *Ordering) MarshalTo(dAtA []byte) (int, error)

func (*Ordering) ProtoMessage

func (*Ordering) ProtoMessage()

func (*Ordering) Reset

func (m *Ordering) Reset()

func (*Ordering) Size

func (m *Ordering) Size() (n int)

func (*Ordering) String

func (m *Ordering) String() string

func (*Ordering) Unmarshal

func (m *Ordering) Unmarshal(dAtA []byte) error

type Ordering_Column

type Ordering_Column struct {
	ColIdx    uint32                    `protobuf:"varint,1,opt,name=col_idx,json=colIdx" json:"col_idx"`
	Direction Ordering_Column_Direction `protobuf:"varint,2,opt,name=direction,enum=cockroach.sql.distsqlrun.Ordering_Column_Direction" json:"direction"`
}

func (*Ordering_Column) Descriptor

func (*Ordering_Column) Descriptor() ([]byte, []int)

func (*Ordering_Column) Marshal

func (m *Ordering_Column) Marshal() (dAtA []byte, err error)

func (*Ordering_Column) MarshalTo

func (m *Ordering_Column) MarshalTo(dAtA []byte) (int, error)

func (*Ordering_Column) ProtoMessage

func (*Ordering_Column) ProtoMessage()

func (*Ordering_Column) Reset

func (m *Ordering_Column) Reset()

func (*Ordering_Column) Size

func (m *Ordering_Column) Size() (n int)

func (*Ordering_Column) String

func (m *Ordering_Column) String() string

func (*Ordering_Column) Unmarshal

func (m *Ordering_Column) Unmarshal(dAtA []byte) error

type Ordering_Column_Direction

type Ordering_Column_Direction int32

The direction of the desired ordering for a column.

const (
	Ordering_Column_ASC  Ordering_Column_Direction = 0
	Ordering_Column_DESC Ordering_Column_Direction = 1
)

func (Ordering_Column_Direction) Enum

func (Ordering_Column_Direction) EnumDescriptor

func (Ordering_Column_Direction) EnumDescriptor() ([]byte, []int)

func (Ordering_Column_Direction) String

func (x Ordering_Column_Direction) String() string

func (*Ordering_Column_Direction) UnmarshalJSON

func (x *Ordering_Column_Direction) UnmarshalJSON(data []byte) error

type OutputRouterSpec

type OutputRouterSpec struct {
	Type    OutputRouterSpec_Type `protobuf:"varint,1,opt,name=type,enum=cockroach.sql.distsqlrun.OutputRouterSpec_Type" json:"type"`
	Streams []StreamEndpointSpec  `protobuf:"bytes,2,rep,name=streams" json:"streams"`
	// Only used for the BY_HASH type; these are the indexes of the columns we are
	// hashing.
	HashColumns []uint32 `protobuf:"varint,3,rep,name=hash_columns,json=hashColumns" json:"hash_columns,omitempty"`
}

OutputRouterSpec is the specification for the output router of a processor; it decides how to send results to multiple output streams.

func (*OutputRouterSpec) Descriptor

func (*OutputRouterSpec) Descriptor() ([]byte, []int)

func (*OutputRouterSpec) Marshal

func (m *OutputRouterSpec) Marshal() (dAtA []byte, err error)

func (*OutputRouterSpec) MarshalTo

func (m *OutputRouterSpec) MarshalTo(dAtA []byte) (int, error)

func (*OutputRouterSpec) ProtoMessage

func (*OutputRouterSpec) ProtoMessage()

func (*OutputRouterSpec) Reset

func (m *OutputRouterSpec) Reset()

func (*OutputRouterSpec) Size

func (m *OutputRouterSpec) Size() (n int)

func (*OutputRouterSpec) String

func (m *OutputRouterSpec) String() string

func (*OutputRouterSpec) Unmarshal

func (m *OutputRouterSpec) Unmarshal(dAtA []byte) error

type OutputRouterSpec_Type

type OutputRouterSpec_Type int32
const (
	// Single output stream.
	OutputRouterSpec_PASS_THROUGH OutputRouterSpec_Type = 0
	// Each row is sent to all output streams.
	OutputRouterSpec_MIRROR OutputRouterSpec_Type = 1
	// Each row is sent to one stream, chosen by hashing certain columns of
	// the row (specified by the hash_columns field).
	OutputRouterSpec_BY_HASH OutputRouterSpec_Type = 2
	// Each row is sent to one stream, chosen according to preset boundaries
	// for the values of certain columns of the row. TODO(radu): an extra
	// optional structure below for the range details.
	OutputRouterSpec_BY_RANGE OutputRouterSpec_Type = 3
)

func (OutputRouterSpec_Type) Enum

func (OutputRouterSpec_Type) EnumDescriptor

func (OutputRouterSpec_Type) EnumDescriptor() ([]byte, []int)

func (OutputRouterSpec_Type) String

func (x OutputRouterSpec_Type) String() string

func (*OutputRouterSpec_Type) UnmarshalJSON

func (x *OutputRouterSpec_Type) UnmarshalJSON(data []byte) error

type PostProcessSpec

type PostProcessSpec struct {
	// A filtering expression which references the internal columns of the
	// processor via ordinal references (@1, @2, etc).
	Filter Expression `protobuf:"bytes,1,opt,name=filter" json:"filter"`
	// If true, output_columns describes a projection. Used to differentiate
	// between an empty projection and no projection.
	//
	// Cannot be set at the same time with render expressions.
	Projection bool `protobuf:"varint,2,opt,name=projection" json:"projection"`
	// The output columns describe a projection on the internal set of columns;
	// only the columns in this list will be emitted.
	//
	// Can only be set if projection is true. Cannot be set at the same time with
	// render expressions.
	OutputColumns []uint32 `protobuf:"varint,3,rep,packed,name=output_columns,json=outputColumns" json:"output_columns,omitempty"`
	// If set, the output is the result of rendering these expressions. The
	// expressions reference the internal columns of the processor.
	//
	// Cannot be set at the same time with output columns.
	RenderExprs []Expression `protobuf:"bytes,4,rep,name=render_exprs,json=renderExprs" json:"render_exprs"`
	// If nonzero, the first <offset> rows will be suppressed.
	Offset uint64 `protobuf:"varint,5,opt,name=offset" json:"offset"`
	// If nonzero, the processor will stop after emitting this many rows.
	Limit uint64 `protobuf:"varint,6,opt,name=limit" json:"limit"`
}

PostProcessSpec describes the processing required to obtain the output (filtering, projection). It operates on the internal schema of the processor (see ProcessorSpec).

func (*PostProcessSpec) Descriptor

func (*PostProcessSpec) Descriptor() ([]byte, []int)

func (*PostProcessSpec) Marshal

func (m *PostProcessSpec) Marshal() (dAtA []byte, err error)

func (*PostProcessSpec) MarshalTo

func (m *PostProcessSpec) MarshalTo(dAtA []byte) (int, error)

func (*PostProcessSpec) ProtoMessage

func (*PostProcessSpec) ProtoMessage()

func (*PostProcessSpec) Reset

func (m *PostProcessSpec) Reset()

func (*PostProcessSpec) Size

func (m *PostProcessSpec) Size() (n int)

func (*PostProcessSpec) String

func (m *PostProcessSpec) String() string

func (*PostProcessSpec) Unmarshal

func (m *PostProcessSpec) Unmarshal(dAtA []byte) error

type ProcessorCoreUnion

type ProcessorCoreUnion struct {
	Noop        *NoopCoreSpec       `protobuf:"bytes,1,opt,name=noop" json:"noop,omitempty"`
	TableReader *TableReaderSpec    `protobuf:"bytes,2,opt,name=tableReader" json:"tableReader,omitempty"`
	JoinReader  *JoinReaderSpec     `protobuf:"bytes,3,opt,name=joinReader" json:"joinReader,omitempty"`
	Sorter      *SorterSpec         `protobuf:"bytes,4,opt,name=sorter" json:"sorter,omitempty"`
	Aggregator  *AggregatorSpec     `protobuf:"bytes,5,opt,name=aggregator" json:"aggregator,omitempty"`
	Distinct    *DistinctSpec       `protobuf:"bytes,7,opt,name=distinct" json:"distinct,omitempty"`
	MergeJoiner *MergeJoinerSpec    `protobuf:"bytes,8,opt,name=mergeJoiner" json:"mergeJoiner,omitempty"`
	HashJoiner  *HashJoinerSpec     `protobuf:"bytes,9,opt,name=hashJoiner" json:"hashJoiner,omitempty"`
	Values      *ValuesCoreSpec     `protobuf:"bytes,10,opt,name=values" json:"values,omitempty"`
	Backfiller  *BackfillerSpec     `protobuf:"bytes,11,opt,name=backfiller" json:"backfiller,omitempty"`
	SetOp       *AlgebraicSetOpSpec `protobuf:"bytes,12,opt,name=setOp" json:"setOp,omitempty"`
}

func (*ProcessorCoreUnion) Descriptor

func (*ProcessorCoreUnion) Descriptor() ([]byte, []int)

func (*ProcessorCoreUnion) GetValue

func (this *ProcessorCoreUnion) GetValue() interface{}

func (*ProcessorCoreUnion) Marshal

func (m *ProcessorCoreUnion) Marshal() (dAtA []byte, err error)

func (*ProcessorCoreUnion) MarshalTo

func (m *ProcessorCoreUnion) MarshalTo(dAtA []byte) (int, error)

func (*ProcessorCoreUnion) ProtoMessage

func (*ProcessorCoreUnion) ProtoMessage()

func (*ProcessorCoreUnion) Reset

func (m *ProcessorCoreUnion) Reset()

func (*ProcessorCoreUnion) SetValue

func (this *ProcessorCoreUnion) SetValue(value interface{}) bool

func (*ProcessorCoreUnion) Size

func (m *ProcessorCoreUnion) Size() (n int)

func (*ProcessorCoreUnion) String

func (m *ProcessorCoreUnion) String() string

func (*ProcessorCoreUnion) Unmarshal

func (m *ProcessorCoreUnion) Unmarshal(dAtA []byte) error

type ProcessorSpec

type ProcessorSpec struct {
	// In most cases, there is one input.
	Input []InputSyncSpec    `protobuf:"bytes,1,rep,name=input" json:"input"`
	Core  ProcessorCoreUnion `protobuf:"bytes,2,opt,name=core" json:"core"`
	Post  PostProcessSpec    `protobuf:"bytes,4,opt,name=post" json:"post"`
	// In most cases, there is one output.
	Output []OutputRouterSpec `protobuf:"bytes,3,rep,name=output" json:"output"`
}

Each processor has the following components:

  • one or more input synchronizers; each one merges rows between one or more input streams;

  • a processor "core" which encapsulates the inner logic of each processor;

  • a post-processing stage which allows "inline" post-processing on results (like projection or filtering);

  • one or more output synchronizers; each one directs rows to one or more output streams.

== Internal columns ==

The core outputs rows of a certain schema to the post-processing stage. We call this the "internal schema" (or "internal columns") and it differs for each type of core. Column indices in a PostProcessSpec refers to these internal columns. Some columns may be unused by the post-processing stage; processor implementations are internally optimized to not produce values for such unneded columns.

func (*ProcessorSpec) Descriptor

func (*ProcessorSpec) Descriptor() ([]byte, []int)

func (*ProcessorSpec) Marshal

func (m *ProcessorSpec) Marshal() (dAtA []byte, err error)

func (*ProcessorSpec) MarshalTo

func (m *ProcessorSpec) MarshalTo(dAtA []byte) (int, error)

func (*ProcessorSpec) ProtoMessage

func (*ProcessorSpec) ProtoMessage()

func (*ProcessorSpec) Reset

func (m *ProcessorSpec) Reset()

func (*ProcessorSpec) Size

func (m *ProcessorSpec) Size() (n int)

func (*ProcessorSpec) String

func (m *ProcessorSpec) String() string

func (*ProcessorSpec) Unmarshal

func (m *ProcessorSpec) Unmarshal(dAtA []byte) error

type ProducerData

type ProducerData struct {
	// A bunch of rows, encoded. Each datum is encoded according to the
	// corresponding DatumInfo.
	RawBytes []byte `protobuf:"bytes,1,opt,name=raw_bytes,json=rawBytes" json:"raw_bytes,omitempty"`
	// In the special case when the stream contains empty rows, the count is
	// passed instead.
	NumEmptyRows int32 `protobuf:"varint,3,opt,name=num_empty_rows,json=numEmptyRows" json:"num_empty_rows"`
	// A bunch of metadata messages.
	Metadata []RemoteProducerMetadata `protobuf:"bytes,2,rep,name=metadata" json:"metadata"`
}

ProducerData is a message that can be sent multiple times as part of a stream from a producer to a consumer. It contains 0 or more rows and/or 0 or more metadata messages.

func (*ProducerData) Descriptor

func (*ProducerData) Descriptor() ([]byte, []int)

func (*ProducerData) Marshal

func (m *ProducerData) Marshal() (dAtA []byte, err error)

func (*ProducerData) MarshalTo

func (m *ProducerData) MarshalTo(dAtA []byte) (int, error)

func (*ProducerData) ProtoMessage

func (*ProducerData) ProtoMessage()

func (*ProducerData) Reset

func (m *ProducerData) Reset()

func (*ProducerData) Size

func (m *ProducerData) Size() (n int)

func (*ProducerData) String

func (m *ProducerData) String() string

func (*ProducerData) Unmarshal

func (m *ProducerData) Unmarshal(dAtA []byte) error

type ProducerHeader

type ProducerHeader struct {
	FlowID   FlowID   `protobuf:"bytes,1,opt,name=flow_id,json=flowId,customtype=FlowID" json:"flow_id"`
	StreamID StreamID `protobuf:"varint,2,opt,name=stream_id,json=streamId,casttype=StreamID" json:"stream_id"`
}

ProducerHeader is a message that is sent once at the beginning of a stream.

func (*ProducerHeader) Descriptor

func (*ProducerHeader) Descriptor() ([]byte, []int)

func (*ProducerHeader) Marshal

func (m *ProducerHeader) Marshal() (dAtA []byte, err error)

func (*ProducerHeader) MarshalTo

func (m *ProducerHeader) MarshalTo(dAtA []byte) (int, error)

func (*ProducerHeader) ProtoMessage

func (*ProducerHeader) ProtoMessage()

func (*ProducerHeader) Reset

func (m *ProducerHeader) Reset()

func (*ProducerHeader) Size

func (m *ProducerHeader) Size() (n int)

func (*ProducerHeader) String

func (m *ProducerHeader) String() string

func (*ProducerHeader) Unmarshal

func (m *ProducerHeader) Unmarshal(dAtA []byte) error

type ProducerMessage

type ProducerMessage struct {
	Header *ProducerHeader `protobuf:"bytes,1,opt,name=header" json:"header,omitempty"`
	// Typing information. There will be one DatumInfo for each element in a row.
	// This field has to be populated on, or before, a ProducerMessage with data
	// in it, and can only be populated once.
	// TODO(andrei): It'd be nice if the typing information for streams would be
	// configured statically at plan creation time, instead of being discovered
	// dynamically through the first rows that flow.
	Typing []DatumInfo  `protobuf:"bytes,2,rep,name=typing" json:"typing"`
	Data   ProducerData `protobuf:"bytes,3,opt,name=data" json:"data"`
}

func (*ProducerMessage) Descriptor

func (*ProducerMessage) Descriptor() ([]byte, []int)

func (*ProducerMessage) Marshal

func (m *ProducerMessage) Marshal() (dAtA []byte, err error)

func (*ProducerMessage) MarshalTo

func (m *ProducerMessage) MarshalTo(dAtA []byte) (int, error)

func (*ProducerMessage) ProtoMessage

func (*ProducerMessage) ProtoMessage()

func (*ProducerMessage) Reset

func (m *ProducerMessage) Reset()

func (*ProducerMessage) Size

func (m *ProducerMessage) Size() (n int)

func (*ProducerMessage) String

func (m *ProducerMessage) String() string

func (*ProducerMessage) Unmarshal

func (m *ProducerMessage) Unmarshal(dAtA []byte) error

type ProducerMetadata

type ProducerMetadata struct {
	// Only one of these fields will be set. If this ever changes, note that
	// there's consumers out there that extract the error and, if there is one,
	// forward it in isolation and drop the rest of the record.
	Ranges []roachpb.RangeInfo
	// TODO(vivek): change to type Error
	Err error
}

ProducerMetadata represents a metadata record flowing through a DistSQL flow.

func (ProducerMetadata) Empty

func (meta ProducerMetadata) Empty() bool

Empty returns true if none of the fields in metadata are populated.

type RemoteProducerMetadata

type RemoteProducerMetadata struct {
	// Types that are valid to be assigned to Value:
	//	*RemoteProducerMetadata_RangeInfo
	//	*RemoteProducerMetadata_Error
	Value isRemoteProducerMetadata_Value `protobuf_oneof:"value"`
}

RemoteProducerMetadata represents records that a producer wants to pass to a consumer, other than data rows. It's named RemoteProducerMetadata to not clash with ProducerMetadata, which is used internally within a node and has a different go error instead of a proto error inside.

func (*RemoteProducerMetadata) Descriptor

func (*RemoteProducerMetadata) Descriptor() ([]byte, []int)

func (*RemoteProducerMetadata) GetError

func (m *RemoteProducerMetadata) GetError() *Error

func (*RemoteProducerMetadata) GetRangeInfo

func (*RemoteProducerMetadata) GetValue

func (m *RemoteProducerMetadata) GetValue() isRemoteProducerMetadata_Value

func (*RemoteProducerMetadata) Marshal

func (m *RemoteProducerMetadata) Marshal() (dAtA []byte, err error)

func (*RemoteProducerMetadata) MarshalTo

func (m *RemoteProducerMetadata) MarshalTo(dAtA []byte) (int, error)

func (*RemoteProducerMetadata) ProtoMessage

func (*RemoteProducerMetadata) ProtoMessage()

func (*RemoteProducerMetadata) Reset

func (m *RemoteProducerMetadata) Reset()

func (*RemoteProducerMetadata) Size

func (m *RemoteProducerMetadata) Size() (n int)

func (*RemoteProducerMetadata) String

func (m *RemoteProducerMetadata) String() string

func (*RemoteProducerMetadata) Unmarshal

func (m *RemoteProducerMetadata) Unmarshal(dAtA []byte) error

func (*RemoteProducerMetadata) XXX_OneofFuncs

func (*RemoteProducerMetadata) XXX_OneofFuncs() (func(msg proto.Message, b *proto.Buffer) error, func(msg proto.Message, tag, wire int, b *proto.Buffer) (bool, error), func(msg proto.Message) (n int), []interface{})

XXX_OneofFuncs is for the internal use of the proto package.

type RemoteProducerMetadata_Error

type RemoteProducerMetadata_Error struct {
	Error *Error `protobuf:"bytes,2,opt,name=error,oneof"`
}

func (*RemoteProducerMetadata_Error) MarshalTo

func (m *RemoteProducerMetadata_Error) MarshalTo(dAtA []byte) (int, error)

func (*RemoteProducerMetadata_Error) Size

func (m *RemoteProducerMetadata_Error) Size() (n int)

type RemoteProducerMetadata_RangeInfo

type RemoteProducerMetadata_RangeInfo struct {
	RangeInfo *RemoteProducerMetadata_RangeInfos `protobuf:"bytes,1,opt,name=range_info,json=rangeInfo,oneof"`
}

func (*RemoteProducerMetadata_RangeInfo) MarshalTo

func (m *RemoteProducerMetadata_RangeInfo) MarshalTo(dAtA []byte) (int, error)

func (*RemoteProducerMetadata_RangeInfo) Size

func (m *RemoteProducerMetadata_RangeInfo) Size() (n int)

type RemoteProducerMetadata_RangeInfos

type RemoteProducerMetadata_RangeInfos struct {
	RangeInfo []cockroach_roachpb3.RangeInfo `protobuf:"bytes,1,rep,name=range_info,json=rangeInfo" json:"range_info"`
}

func (*RemoteProducerMetadata_RangeInfos) Descriptor

func (*RemoteProducerMetadata_RangeInfos) Descriptor() ([]byte, []int)

func (*RemoteProducerMetadata_RangeInfos) Marshal

func (m *RemoteProducerMetadata_RangeInfos) Marshal() (dAtA []byte, err error)

func (*RemoteProducerMetadata_RangeInfos) MarshalTo

func (m *RemoteProducerMetadata_RangeInfos) MarshalTo(dAtA []byte) (int, error)

func (*RemoteProducerMetadata_RangeInfos) ProtoMessage

func (*RemoteProducerMetadata_RangeInfos) ProtoMessage()

func (*RemoteProducerMetadata_RangeInfos) Reset

func (*RemoteProducerMetadata_RangeInfos) Size

func (m *RemoteProducerMetadata_RangeInfos) Size() (n int)

func (*RemoteProducerMetadata_RangeInfos) String

func (*RemoteProducerMetadata_RangeInfos) Unmarshal

func (m *RemoteProducerMetadata_RangeInfos) Unmarshal(dAtA []byte) error

type RowBuffer

type RowBuffer struct {

	// ProducerClosed is used when the RowBuffer is used as a RowReceiver; it is
	// set to true when the sender calls ProducerDone().
	ProducerClosed bool

	// Done is used when the RowBuffer is used as a RowSource; it is set to true
	// when the receiver read all the rows.
	Done bool

	ConsumerStatus ConsumerStatus
	// contains filtered or unexported fields
}

RowBuffer is an implementation of RowReceiver that buffers (accumulates) results in memory, as well as an implementation of RowSource that returns records from a record buffer. Just for tests.

func NewRowBuffer

func NewRowBuffer(
	types []sqlbase.ColumnType, rows sqlbase.EncDatumRows, hooks RowBufferArgs,
) *RowBuffer

NewRowBuffer creates a RowBuffer with the given schema and initial rows. The types are optional if there is at least one row.

func (*RowBuffer) ConsumerClosed

func (rb *RowBuffer) ConsumerClosed()

ConsumerClosed is part of the RowSource interface.

func (*RowBuffer) ConsumerDone

func (rb *RowBuffer) ConsumerDone()

ConsumerDone is part of the RowSource interface.

func (*RowBuffer) Next

Next is part of the RowSource interface.

There's no synchronization here with Push(). The assumption is that these two methods are not called concurrently.

func (*RowBuffer) ProducerDone

func (rb *RowBuffer) ProducerDone()

ProducerDone is part of the interface.

func (*RowBuffer) Push

Push is part of the RowReceiver interface.

func (*RowBuffer) Types

func (rb *RowBuffer) Types() []sqlbase.ColumnType

Types is part of the RowSource interface.

type RowBufferArgs

type RowBufferArgs struct {
	// If not set, then the RowBuffer will behave like a RowChannel and not
	// accumulate rows after it's been put in draining mode. If set, rows will still
	// be accumulated. Useful for tests that want to observe what rows have been
	// pushed after draining.
	AccumulateRowsWhileDraining bool
	// OnConsumerDone, if specified, is called as the first thing in the
	// ConsumerDone() method.
	OnConsumerDone func(*RowBuffer)
	// OnConsumerClose, if specified, is called as the first thing in the
	// ConsumerClosed() method.
	OnConsumerClosed func(*RowBuffer)
	// OnNext, if specified, is called as the first thing in the Next() method.
	// If it returns an empty row and metadata, then RowBuffer.Next() is allowed
	// to run normally. Otherwise, the values are returned from RowBuffer.Next().
	OnNext func(*RowBuffer) (sqlbase.EncDatumRow, ProducerMetadata)
}

RowBufferArgs contains testing-oriented parameters for a RowBuffer.

type RowChannel

type RowChannel struct {

	// The channel on which rows are delivered.
	C <-chan RowChannelMsg
	// contains filtered or unexported fields
}

RowChannel is a thin layer over a RowChannelMsg channel, which can be used to transfer rows between goroutines.

func (*RowChannel) ConsumerClosed

func (rc *RowChannel) ConsumerClosed()

ConsumerClosed is part of the RowSource interface.

func (*RowChannel) ConsumerDone

func (rc *RowChannel) ConsumerDone()

ConsumerDone is part of the RowSource interface.

func (*RowChannel) Init

func (rc *RowChannel) Init(types []sqlbase.ColumnType)

Init initializes the RowChannel with the default buffer size.

func (*RowChannel) InitWithBufSize

func (rc *RowChannel) InitWithBufSize(types []sqlbase.ColumnType, chanBufSize int)

InitWithBufSize initializes the RowChannel with a given buffer size.

func (*RowChannel) Next

Next is part of the RowSource interface.

func (*RowChannel) ProducerDone

func (rc *RowChannel) ProducerDone()

ProducerDone is part of the RowReceiver interface.

func (*RowChannel) Push

Push is part of the RowReceiver interface.

func (*RowChannel) Types

func (rc *RowChannel) Types() []sqlbase.ColumnType

Types is part of the RowSource interface.

type RowChannelMsg

type RowChannelMsg struct {
	// Only one of these fields will be set.
	Row  sqlbase.EncDatumRow
	Meta ProducerMetadata
}

RowChannelMsg is the message used in the channels that implement local physical streams (i.e. the RowChannel's).

type RowReceiver

type RowReceiver interface {
	// Push sends a record to the consumer of this RowReceiver. Exactly one of the
	// row/meta must be specified (i.e. either row needs to be non-nil or meta
	// needs to be non-Empty()). May block.
	//
	// The return value indicates the current status of the consumer. Depending on
	// it, producers are expected to drain or shut down. In all cases,
	// ProducerDone() needs to be called (after draining is done, if draining was
	// requested).
	//
	// The sender must not modify the row after calling this function.
	//
	// After DrainRequested is returned, it is expected that all future calls only
	// carry metadata (however that is not enforced and implementations should be
	// prepared to discard non-metadata rows). If ConsumerClosed is returned,
	// implementations have to ignore further calls to Push() (such calls are
	// allowed because there might be multiple producers for a single RowReceiver
	// and they might not all be aware of the last status returned).
	//
	// Implementations of Push() must be thread-safe.
	Push(row sqlbase.EncDatumRow, meta ProducerMetadata) ConsumerStatus

	// ProducerDone is called when the producer has pushed all the rows and
	// metadata; it causes the RowReceiver to process all rows and clean up.
	//
	// ProducerDone() cannot be called concurrently with Push(), and after it
	// is called, no other method can be called.
	ProducerDone()
}

RowReceiver is any component of a flow that receives rows from another component. It can be an input synchronizer, a router, or a mailbox.

type RowSource

type RowSource interface {
	// Types returns the schema for the rows in this source.
	Types() []sqlbase.ColumnType

	// Next returns the next record that a producer has pushed into this
	// RowSource. At most one of the return values will be non-empty. Both of them
	// can be empty when the RowSource has been exhausted - no more records are
	// coming and any further method calls will be no-ops.
	//
	// A ProducerMetadata record may contain an error. In that case, this
	// interface is oblivious about the semantics: implementers may continue
	// returning different rows on future calls, or may return an empty record
	// (thus asking the consumer to stop asking for rows). In particular,
	// implementers are not required to only return metadata records from this
	// point on (which means, for example, that they're not required to
	// automatically ask every producer to drain, in case there's multiple
	// producers). Therefore, consumers need to be aware that some rows might have
	// been skipped in case they continue to consume rows. Usually a consumer
	// should react to an error by calling ConsumerDone(), thus asking the
	// RowSource to drain, and separately discard any future data rows.
	Next() (sqlbase.EncDatumRow, ProducerMetadata)

	// ConsumerDone lets the producer know that we will not need any more data
	// rows. The producer is expected to start draining and only send metadata
	// rows.
	//
	// May block. If the consumer of the source stops consuming rows before
	// Next indicates that there are no more rows, ConsumerDone() and/or
	// ConsumerClosed() must be called; it is a no-op to call these methods after
	// all the rows were consumed (i.e. after Next() returned an empty row).
	ConsumerDone()

	// ConsumerClosed informs the producer that the consumer will not be reading
	// any more rows. The producer is expected to shut down without sending
	// anything else.
	//
	// Like ConsumerDone(), if the consumer of the source stops consuming rows
	// before Next indicates that there are no more rows, ConsumerDone() and/or
	// ConsumerClosed() must be called; it is a no-op to call these methods after
	// all the rows were consumed (i.e. after Next() returned an empty row).
	ConsumerClosed()
}

RowSource is any component of a flow that produces rows that cam be consumed by another component.

type ServerConfig

type ServerConfig struct {
	log.AmbientContext

	// DB is a handle to the cluster.
	DB *client.DB
	// FlowDB is the DB that flows should use for interacting with the database.
	// This DB has to be set such that it bypasses the local TxnCoordSender. We
	// want only the TxnCoordSender on the gateway to be involved with requests
	// performed by DistSQL.
	FlowDB       *client.DB
	RPCContext   *rpc.Context
	Stopper      *stop.Stopper
	TestingKnobs TestingKnobs

	ParentMemoryMonitor *mon.MemoryMonitor
	Counter             *metric.Counter
	Hist                *metric.Histogram
	// NodeID is the id of the node on which this Server is running.
	NodeID *base.NodeIDContainer
}

ServerConfig encompasses the configuration required to create a DistSQLServer.

type ServerImpl

type ServerImpl struct {
	ServerConfig
	// contains filtered or unexported fields
}

ServerImpl implements the server for the distributed SQL APIs.

func NewServer

func NewServer(ctx context.Context, cfg ServerConfig) *ServerImpl

NewServer instantiates a DistSQLServer.

func (*ServerImpl) FlowStream

func (ds *ServerImpl) FlowStream(stream DistSQL_FlowStreamServer) error

FlowStream is part of the DistSQLServer interface.

func (*ServerImpl) RunSyncFlow

func (ds *ServerImpl) RunSyncFlow(stream DistSQL_RunSyncFlowServer) error

RunSyncFlow is part of the DistSQLServer interface.

func (*ServerImpl) SetupFlow

func (ds *ServerImpl) SetupFlow(_ context.Context, req *SetupFlowRequest) (*SimpleResponse, error)

SetupFlow is part of the DistSQLServer interface.

func (*ServerImpl) SetupSyncFlow

func (ds *ServerImpl) SetupSyncFlow(
	ctx context.Context, req *SetupFlowRequest, output RowReceiver,
) (context.Context, *Flow, error)

SetupSyncFlow sets up a synchoronous flow, connecting the sync response output stream to the given RowReceiver. The flow is not started. The flow will be associated with the given context. Note: the returned context contains a span that must be finished through Flow.Cleanup.

func (*ServerImpl) Start

func (ds *ServerImpl) Start()

Start launches workers for the server.

type SetupFlowRequest

type SetupFlowRequest struct {
	Txn cockroach_roachpb1.Transaction `protobuf:"bytes,1,opt,name=txn" json:"txn"`
	// Version of distsqlrun protocol; a server accepts a certain range of
	// versions, up to its own version. See server.go for more details.
	Version uint32 `protobuf:"varint,5,opt,name=version" json:"version"`
	// If set, the context of an active tracing span.
	TraceContext *cockroach_util_tracing.SpanContextCarrier `protobuf:"bytes,2,opt,name=trace_context,json=traceContext" json:"trace_context,omitempty"`
	Flow         FlowSpec                                   `protobuf:"bytes,3,opt,name=flow" json:"flow"`
	EvalContext  EvalContext                                `protobuf:"bytes,6,opt,name=evalContext" json:"evalContext"`
}

func (*SetupFlowRequest) Descriptor

func (*SetupFlowRequest) Descriptor() ([]byte, []int)

func (*SetupFlowRequest) Marshal

func (m *SetupFlowRequest) Marshal() (dAtA []byte, err error)

func (*SetupFlowRequest) MarshalTo

func (m *SetupFlowRequest) MarshalTo(dAtA []byte) (int, error)

func (*SetupFlowRequest) ProtoMessage

func (*SetupFlowRequest) ProtoMessage()

func (*SetupFlowRequest) Reset

func (m *SetupFlowRequest) Reset()

func (*SetupFlowRequest) Size

func (m *SetupFlowRequest) Size() (n int)

func (*SetupFlowRequest) String

func (m *SetupFlowRequest) String() string

func (*SetupFlowRequest) Unmarshal

func (m *SetupFlowRequest) Unmarshal(dAtA []byte) error

type SimpleResponse

type SimpleResponse struct {
	Error *Error `protobuf:"bytes,1,opt,name=error" json:"error,omitempty"`
}

func (*SimpleResponse) Descriptor

func (*SimpleResponse) Descriptor() ([]byte, []int)

func (*SimpleResponse) Marshal

func (m *SimpleResponse) Marshal() (dAtA []byte, err error)

func (*SimpleResponse) MarshalTo

func (m *SimpleResponse) MarshalTo(dAtA []byte) (int, error)

func (*SimpleResponse) ProtoMessage

func (*SimpleResponse) ProtoMessage()

func (*SimpleResponse) Reset

func (m *SimpleResponse) Reset()

func (*SimpleResponse) Size

func (m *SimpleResponse) Size() (n int)

func (*SimpleResponse) String

func (m *SimpleResponse) String() string

func (*SimpleResponse) Unmarshal

func (m *SimpleResponse) Unmarshal(dAtA []byte) error

type SorterSpec

type SorterSpec struct {
	OutputOrdering Ordering `protobuf:"bytes,1,opt,name=output_ordering,json=outputOrdering" json:"output_ordering"`
	// Ordering match length, specifying that the input is already sorted by the
	// first 'n' output ordering columns, can be optionally specified for
	// possible speed-ups taking advantage of the partial orderings.
	OrderingMatchLen uint32 `protobuf:"varint,2,opt,name=ordering_match_len,json=orderingMatchLen" json:"ordering_match_len"`
}

SorterSpec is the specification for a "sorting aggregator". A sorting aggregator sorts elements in the input stream providing a certain output order guarantee regardless of the input ordering. The output ordering is according to a configurable set of columns.

The "internal columns" of a Sorter (see ProcessorSpec) are the same as the input columns.

func (*SorterSpec) Descriptor

func (*SorterSpec) Descriptor() ([]byte, []int)

func (*SorterSpec) Marshal

func (m *SorterSpec) Marshal() (dAtA []byte, err error)

func (*SorterSpec) MarshalTo

func (m *SorterSpec) MarshalTo(dAtA []byte) (int, error)

func (*SorterSpec) ProtoMessage

func (*SorterSpec) ProtoMessage()

func (*SorterSpec) Reset

func (m *SorterSpec) Reset()

func (*SorterSpec) Size

func (m *SorterSpec) Size() (n int)

func (*SorterSpec) String

func (m *SorterSpec) String() string

func (*SorterSpec) Unmarshal

func (m *SorterSpec) Unmarshal(dAtA []byte) error

type StreamDecoder

type StreamDecoder struct {
	// contains filtered or unexported fields
}

StreamDecoder converts a sequence of ProducerMessage to rows and metadata records.

Sample usage:

sd := StreamDecoder{}
var row sqlbase.EncDatumRow
for each message in stream {
    err := sd.AddMessage(msg)
    if err != nil { ... }
    for {
        row, meta, err := sd.GetRow(row)
        if err != nil { ... }
        if row == nil && meta.Empty() {
            // No more rows in this message.
            break
        }
        // Use <row>
        ...
    }
}

AddMessage can be called multiple times before getting the rows, but this will cause data to accumulate internally.

func (*StreamDecoder) AddMessage

func (sd *StreamDecoder) AddMessage(msg *ProducerMessage) error

AddMessage adds the data in a ProducerMessage to the decoder.

The StreamDecoder may keep a reference to msg.Data.RawBytes and msg.Data.Metadata until all the rows in the message are retrieved with GetRow.

If an error is returned, no records have been buffered in the StreamDecoder.

func (*StreamDecoder) GetRow

GetRow returns a row received in the stream. A row buffer can be provided optionally.

Returns an empty row if there are no more rows received so far.

A decoding error may be returned. Note that these are separate from error coming from the upstream (through ProducerMetadata.Err).

type StreamEncoder

type StreamEncoder struct {
	// contains filtered or unexported fields
}

StreamEncoder converts EncDatum rows into a sequence of ProducerMessage.

Sample usage:

se := StreamEncoder{}

for {
    for ... {
       err := se.AddRow(...)
       ...
    }
    msg := se.FormMessage(false, nil)
    // Send out message.
    ...
}
msg := se.FormMessage(true, nil)
// Send out final message
...

func (*StreamEncoder) AddMetadata

func (se *StreamEncoder) AddMetadata(meta ProducerMetadata)

AddMetadata encodes a metadata message. Unlike AddRow(), it cannot fail. This is important for the caller because a failure to encode a piece of metadata (particularly one that contains an error) would not be recoverable.

Metadata records lose their ordering wrt the data rows. The convention is that the StreamDecoder will return them first, before the data rows, thus ensuring that rows produced _after_ an error are not received _before_ the error.

func (*StreamEncoder) AddRow

func (se *StreamEncoder) AddRow(row sqlbase.EncDatumRow) error

AddRow encodes a message.

func (*StreamEncoder) FormMessage

func (se *StreamEncoder) FormMessage(ctx context.Context) *ProducerMessage

FormMessage populates a message containing the rows added since the last call to FormMessage. The returned ProducerMessage should be treated as immutable.

type StreamEndpointSpec

type StreamEndpointSpec struct {
	Type StreamEndpointSpec_Type `protobuf:"varint,1,opt,name=type,enum=cockroach.sql.distsqlrun.StreamEndpointSpec_Type" json:"type"`
	// The ID of this stream.
	//
	// For LOCAL streams, both ends of the stream are part of the flow on this
	// machine (and there must be a corresponding endpoint with the same ID).
	//
	// For REMOTE streams, this ID is used in the ProducerHeader when connecting to
	// the other host.
	//
	// For SYNC_RESPONSE streams, the ID is unused.
	StreamID StreamID `protobuf:"varint,2,opt,name=stream_id,json=streamId,casttype=StreamID" json:"stream_id"`
	// Serving address for the target host, only used for outgoing REMOTE streams.
	TargetAddr string `protobuf:"bytes,3,opt,name=target_addr,json=targetAddr" json:"target_addr"`
}

StreamEndpointSpec describes one of the endpoints (input or output) of a physical stream.

func (*StreamEndpointSpec) Descriptor

func (*StreamEndpointSpec) Descriptor() ([]byte, []int)

func (*StreamEndpointSpec) Marshal

func (m *StreamEndpointSpec) Marshal() (dAtA []byte, err error)

func (*StreamEndpointSpec) MarshalTo

func (m *StreamEndpointSpec) MarshalTo(dAtA []byte) (int, error)

func (*StreamEndpointSpec) ProtoMessage

func (*StreamEndpointSpec) ProtoMessage()

func (*StreamEndpointSpec) Reset

func (m *StreamEndpointSpec) Reset()

func (*StreamEndpointSpec) Size

func (m *StreamEndpointSpec) Size() (n int)

func (*StreamEndpointSpec) String

func (m *StreamEndpointSpec) String() string

func (*StreamEndpointSpec) Unmarshal

func (m *StreamEndpointSpec) Unmarshal(dAtA []byte) error

type StreamEndpointSpec_Type

type StreamEndpointSpec_Type int32
const (
	// Stream that is part of the local flow.
	StreamEndpointSpec_LOCAL StreamEndpointSpec_Type = 0
	// Stream that has the other endpoint on a different node.
	StreamEndpointSpec_REMOTE StreamEndpointSpec_Type = 1
	// Special stream used when in "sync flow" mode. In this mode, we return
	// results directly as part of the RPC call that set up the flow. This saves
	// overhead (extra RPCs) compared to the normal mode where the RPC just sets
	// up the flow. This type can only be used with outbound endpoints.
	StreamEndpointSpec_SYNC_RESPONSE StreamEndpointSpec_Type = 2
)

func (StreamEndpointSpec_Type) Enum

func (StreamEndpointSpec_Type) EnumDescriptor

func (StreamEndpointSpec_Type) EnumDescriptor() ([]byte, []int)

func (StreamEndpointSpec_Type) String

func (x StreamEndpointSpec_Type) String() string

func (*StreamEndpointSpec_Type) UnmarshalJSON

func (x *StreamEndpointSpec_Type) UnmarshalJSON(data []byte) error

type StreamID

type StreamID int

StreamID identifies a stream; it may be local to a flow or it may cross machine boundaries. The identifier can only be used in the context of a specific flow.

type TableReaderSpan

type TableReaderSpan struct {
	// TODO(radu): the dist_sql APIs should be agnostic to how we map tables to
	// KVs. The span should be described as starting and ending lists of values
	// for a prefix of the index columns, along with inclusive/exclusive flags.
	Span cockroach_roachpb1.Span `protobuf:"bytes,1,opt,name=span" json:"span"`
}

func (*TableReaderSpan) Descriptor

func (*TableReaderSpan) Descriptor() ([]byte, []int)

func (*TableReaderSpan) Marshal

func (m *TableReaderSpan) Marshal() (dAtA []byte, err error)

func (*TableReaderSpan) MarshalTo

func (m *TableReaderSpan) MarshalTo(dAtA []byte) (int, error)

func (*TableReaderSpan) ProtoMessage

func (*TableReaderSpan) ProtoMessage()

func (*TableReaderSpan) Reset

func (m *TableReaderSpan) Reset()

func (*TableReaderSpan) Size

func (m *TableReaderSpan) Size() (n int)

func (*TableReaderSpan) String

func (m *TableReaderSpan) String() string

func (*TableReaderSpan) Unmarshal

func (m *TableReaderSpan) Unmarshal(dAtA []byte) error

type TableReaderSpec

type TableReaderSpec struct {
	Table cockroach_sql_sqlbase1.TableDescriptor `protobuf:"bytes,1,opt,name=table" json:"table"`
	// If 0, we use the primary index. If non-zero, we use the index_idx-th index,
	// i.e. table.indexes[index_idx-1]
	IndexIdx uint32            `protobuf:"varint,2,opt,name=index_idx,json=indexIdx" json:"index_idx"`
	Reverse  bool              `protobuf:"varint,3,opt,name=reverse" json:"reverse"`
	Spans    []TableReaderSpan `protobuf:"bytes,4,rep,name=spans" json:"spans"`
	// A hint for how many rows the consumer of the table reader output might
	// need. This is used to size the initial KV batches to try to avoid reading
	// many more rows than needed by the processor receiving the output.
	//
	// Not used if there is a limit set in the PostProcessSpec of this processor
	// (that value will be used for sizing batches instead).
	LimitHint int64 `protobuf:"varint,5,opt,name=limit_hint,json=limitHint" json:"limit_hint"`
}

TableReaderSpec is the specification for a "table reader". A table reader performs KV operations to retrieve rows for a table and outputs the desired columns of the rows that pass a filter expression.

The "internal columns" of a TableReader (see ProcessorSpec) are all the columns of the table. Internally, only the values for the columns needed by the post-processing stage are be populated.

func (*TableReaderSpec) Descriptor

func (*TableReaderSpec) Descriptor() ([]byte, []int)

func (*TableReaderSpec) Marshal

func (m *TableReaderSpec) Marshal() (dAtA []byte, err error)

func (*TableReaderSpec) MarshalTo

func (m *TableReaderSpec) MarshalTo(dAtA []byte) (int, error)

func (*TableReaderSpec) ProtoMessage

func (*TableReaderSpec) ProtoMessage()

func (*TableReaderSpec) Reset

func (m *TableReaderSpec) Reset()

func (*TableReaderSpec) Size

func (m *TableReaderSpec) Size() (n int)

func (*TableReaderSpec) String

func (m *TableReaderSpec) String() string

func (*TableReaderSpec) Unmarshal

func (m *TableReaderSpec) Unmarshal(dAtA []byte) error

type TestingKnobs

type TestingKnobs struct {
	// RunBeforeBackfillChunk is called before executing each chunk of a
	// backfill during a schema change operation. It is called with the
	// current span and returns an error which eventually is returned to the
	// caller of SchemaChanger.exec(). It is called at the start of the
	// backfill function passed into the transaction executing the chunk.
	RunBeforeBackfillChunk func(sp roachpb.Span) error

	// RunAfterBackfillChunk is called after executing each chunk of a
	// backfill during a schema change operation. It is called just before
	// returning from the backfill function passed into the transaction
	// executing the chunk. It is always called even when the backfill
	// function returns an error, or if the table has already been dropped.
	RunAfterBackfillChunk func()
}

TestingKnobs are the testing knobs.

func (*TestingKnobs) ModuleTestingKnobs

func (*TestingKnobs) ModuleTestingKnobs()

ModuleTestingKnobs is part of the base.ModuleTestingKnobs interface.

type ValuesCoreSpec

type ValuesCoreSpec struct {
	// There is one DatumInfo for each element in a row.
	Columns []DatumInfo `protobuf:"bytes,1,rep,name=columns" json:"columns"`
	// Each raw block encodes one or more data rows; each datum is encoded
	// according to the corresponding DatumInfo.
	RawBytes [][]byte `protobuf:"bytes,2,rep,name=raw_bytes,json=rawBytes" json:"raw_bytes,omitempty"`
}

ValuesCoreSpec is the core of a processor that has no inputs and generates "pre-canned" rows. This is not intended to be used for very large datasets.

func (*ValuesCoreSpec) Descriptor

func (*ValuesCoreSpec) Descriptor() ([]byte, []int)

func (*ValuesCoreSpec) Marshal

func (m *ValuesCoreSpec) Marshal() (dAtA []byte, err error)

func (*ValuesCoreSpec) MarshalTo

func (m *ValuesCoreSpec) MarshalTo(dAtA []byte) (int, error)

func (*ValuesCoreSpec) ProtoMessage

func (*ValuesCoreSpec) ProtoMessage()

func (*ValuesCoreSpec) Reset

func (m *ValuesCoreSpec) Reset()

func (*ValuesCoreSpec) Size

func (m *ValuesCoreSpec) Size() (n int)

func (*ValuesCoreSpec) String

func (m *ValuesCoreSpec) String() string

func (*ValuesCoreSpec) Unmarshal

func (m *ValuesCoreSpec) Unmarshal(dAtA []byte) error

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL