Documentation
¶
Overview ¶
Package trace implements trace-specific vectorized query operators: the two-phase pull pipeline that resolves trace IDs (Phase 1) and materializes spans grouped per trace (Phase 2).
Index ¶
- Constants
- func IncrQueryCount()
- func NewPhase1Schema() *vectorized.BatchSchema
- func QueryCount() int64
- type DistinctTraceID
- func (d *DistinctTraceID) Close() error
- func (d *DistinctTraceID) Init(context.Context) error
- func (d *DistinctTraceID) Keys() map[string]int64
- func (d *DistinctTraceID) Order() []string
- func (d *DistinctTraceID) OutputSchema() *vectorized.BatchSchema
- func (d *DistinctTraceID) Process(_ context.Context, batch *vectorized.RecordBatch) error
- func (d *DistinctTraceID) TraceIDsByPart() map[int64][]string
- type IDDecode
- type Limit
- type LimitedDistinctTraceID
- func (l *LimitedDistinctTraceID) Carry() *DistinctTraceID
- func (l *LimitedDistinctTraceID) Close() error
- func (l *LimitedDistinctTraceID) Init(ctx context.Context) error
- func (l *LimitedDistinctTraceID) OutputSchema() *vectorized.BatchSchema
- func (l *LimitedDistinctTraceID) Process(_ context.Context, batch *vectorized.RecordBatch) error
- type MergeItem
- type Phase1Plan
- type SidxResponseIterator
- type SidxRowBatch
- type SortedMerge
- type StaticTraceIDSource
- type VectorizedConfig
Constants ¶
const ( Phase1ColumnNameKey = "key" Phase1ColumnNameSeriesID = "seriesID" Phase1ColumnNamePartID = "partID" Phase1ColumnNamePayload = "payload" Phase1ColumnNameTraceID = "traceID" )
Phase-1 column names.
Variables ¶
This section is empty.
Functions ¶
func IncrQueryCount ¶
func IncrQueryCount()
IncrQueryCount increments the process-wide vectorized trace query counter. Called by banyand/trace when the vectorized path is taken.
func NewPhase1Schema ¶
func NewPhase1Schema() *vectorized.BatchSchema
NewPhase1Schema returns the thin trace-ID resolution batch schema.
func QueryCount ¶
func QueryCount() int64
QueryCount returns the cumulative number of vectorized trace queries successfully started by this process.
Types ¶
type DistinctTraceID ¶
type DistinctTraceID struct {
// contains filtered or unexported fields
}
DistinctTraceID carries ordered trace IDs, part hints, and keys into Phase 2.
func NewDistinctTraceID ¶
func NewDistinctTraceID(schema *vectorized.BatchSchema) *DistinctTraceID
NewDistinctTraceID constructs the Phase-1 carry-forward fusible.
func (*DistinctTraceID) Close ¶
func (d *DistinctTraceID) Close() error
Close is idempotent and a no-op.
func (*DistinctTraceID) Init ¶
func (d *DistinctTraceID) Init(context.Context) error
Init resets carry-forward state.
func (*DistinctTraceID) Keys ¶
func (d *DistinctTraceID) Keys() map[string]int64
Keys returns the traceID to sort-key mapping collected so far.
func (*DistinctTraceID) Order ¶
func (d *DistinctTraceID) Order() []string
Order returns trace IDs in Phase-1 arrival order.
func (*DistinctTraceID) OutputSchema ¶
func (d *DistinctTraceID) OutputSchema() *vectorized.BatchSchema
OutputSchema returns the unchanged input schema.
func (*DistinctTraceID) Process ¶
func (d *DistinctTraceID) Process(_ context.Context, batch *vectorized.RecordBatch) error
Process records active rows without rewriting the selection vector.
func (*DistinctTraceID) TraceIDsByPart ¶
func (d *DistinctTraceID) TraceIDsByPart() map[int64][]string
TraceIDsByPart returns the part-hint mapping collected so far.
type IDDecode ¶
type IDDecode struct {
// contains filtered or unexported fields
}
IDDecode decodes idFormatV1 trace-ID payloads into the traceID column.
func NewIDDecode ¶
func NewIDDecode(schema *vectorized.BatchSchema) *IDDecode
NewIDDecode constructs a trace-ID decode fusible.
func (*IDDecode) OutputSchema ¶
func (d *IDDecode) OutputSchema() *vectorized.BatchSchema
OutputSchema returns the unchanged input schema.
func (*IDDecode) Process ¶
func (d *IDDecode) Process(_ context.Context, batch *vectorized.RecordBatch) error
Process decodes every active row's payload into traceID.
type Limit ¶
type Limit struct {
// contains filtered or unexported fields
}
Limit applies MaxTraceSize to the ordered Phase-1 stream.
func NewLimit ¶
func NewLimit(schema *vectorized.BatchSchema, maxRows uint32) *Limit
NewLimit constructs a trace Phase-1 limit fusible.
func (*Limit) OutputSchema ¶
func (l *Limit) OutputSchema() *vectorized.BatchSchema
OutputSchema returns the unchanged input schema.
func (*Limit) Process ¶
func (l *Limit) Process(_ context.Context, batch *vectorized.RecordBatch) error
Process rewrites selection to keep only the first max active rows.
type LimitedDistinctTraceID ¶
type LimitedDistinctTraceID struct {
// contains filtered or unexported fields
}
LimitedDistinctTraceID applies MaxTraceSize while carrying kept rows.
func NewLimitedDistinctTraceID ¶
func NewLimitedDistinctTraceID(schema *vectorized.BatchSchema, maxRows uint32) *LimitedDistinctTraceID
NewLimitedDistinctTraceID constructs the Phase-1 limit and carry fusible.
func (*LimitedDistinctTraceID) Carry ¶
func (l *LimitedDistinctTraceID) Carry() *DistinctTraceID
Carry returns the embedded carry-forward state.
func (*LimitedDistinctTraceID) Close ¶
func (l *LimitedDistinctTraceID) Close() error
Close is idempotent and closes the embedded carry operator.
func (*LimitedDistinctTraceID) Init ¶
func (l *LimitedDistinctTraceID) Init(ctx context.Context) error
Init resets carry-forward state.
func (*LimitedDistinctTraceID) OutputSchema ¶
func (l *LimitedDistinctTraceID) OutputSchema() *vectorized.BatchSchema
OutputSchema returns the unchanged input schema.
func (*LimitedDistinctTraceID) Process ¶
func (l *LimitedDistinctTraceID) Process(_ context.Context, batch *vectorized.RecordBatch) error
Process rewrites selection to kept rows and records only those rows.
type MergeItem ¶
type MergeItem struct {
TraceID string
Payload []byte
Key int64
SeriesID int64
PartID int64
// contains filtered or unexported fields
}
MergeItem is one sorted Phase-1 trace candidate consumed by SortedMerge.
func NewMergeItem ¶
NewMergeItem builds a merge item whose SortedField preserves numeric int64 order.
func (*MergeItem) SortedField ¶
SortedField returns the order-preserving encoded key for itersort.
type Phase1Plan ¶
type Phase1Plan struct {
Pipeline *vectorized.Pipeline
Carry *DistinctTraceID
}
Phase1Plan holds the executable Phase-1 pipeline and its carry-forward operator.
func BuildMergePhase1 ¶
func BuildMergePhase1(iters []itersort.Iterator[*MergeItem], desc bool, maxTraceSize uint32, batchSize int) (*Phase1Plan, error)
BuildMergePhase1 builds the ordered sidx Phase-1 pipeline.
func BuildStaticPhase1 ¶
func BuildStaticPhase1(traceIDs []string, keys map[string]int64, maxTraceSize uint32, batchSize int) (*Phase1Plan, error)
BuildStaticPhase1 builds the traceID lookup Phase-1 pipeline.
type SidxResponseIterator ¶
type SidxResponseIterator struct {
// contains filtered or unexported fields
}
SidxResponseIterator adapts ordered sidx row batches to merge items.
func NewSidxResponseIterator ¶
func NewSidxResponseIterator(responses []*SidxRowBatch) *SidxResponseIterator
NewSidxResponseIterator constructs a merge iterator over sidx row batches.
func (*SidxResponseIterator) Close ¶
func (i *SidxResponseIterator) Close() error
Close marks the iterator closed.
func (*SidxResponseIterator) Error ¶
func (i *SidxResponseIterator) Error() error
Error returns the first response or shape error observed by Next.
func (*SidxResponseIterator) Next ¶
func (i *SidxResponseIterator) Next() bool
Next advances to the next well-formed row.
func (*SidxResponseIterator) Val ¶
func (i *SidxResponseIterator) Val() *MergeItem
Val returns the current merge item.
type SidxRowBatch ¶
SidxRowBatch is the sidx row shape consumed by vectorized trace Phase 1.
func (*SidxRowBatch) Len ¶
func (b *SidxRowBatch) Len() int
Len returns the number of rows in the batch.
type SortedMerge ¶
type SortedMerge struct {
// contains filtered or unexported fields
}
SortedMerge lazily merges sorted trace candidate iterators and deduplicates payloads.
func NewSortedMerge ¶
NewSortedMerge returns a streaming pull operator backed by itersort.NewItemIter.
func (*SortedMerge) Close ¶
func (s *SortedMerge) Close() error
Close releases the source iterators through the underlying merger.
func (*SortedMerge) Init ¶
func (s *SortedMerge) Init(context.Context) error
Init initializes the merge heap and output pool.
func (*SortedMerge) NextBatch ¶
func (s *SortedMerge) NextBatch(ctx context.Context) (*vectorized.RecordBatch, error)
NextBatch returns the next demand-bounded merged batch.
func (*SortedMerge) OutputSchema ¶
func (s *SortedMerge) OutputSchema() *vectorized.BatchSchema
OutputSchema returns the Phase-1 schema.
type StaticTraceIDSource ¶
type StaticTraceIDSource struct {
// contains filtered or unexported fields
}
StaticTraceIDSource emits Phase-1 rows for direct trace-ID lookup mode.
func NewStaticTraceIDSource ¶
func NewStaticTraceIDSource(traceIDs []string, keys map[string]int64, batchSize int) *StaticTraceIDSource
NewStaticTraceIDSource constructs a static lookup source.
func (*StaticTraceIDSource) Close ¶
func (s *StaticTraceIDSource) Close() error
Close is idempotent and a no-op.
func (*StaticTraceIDSource) Init ¶
func (s *StaticTraceIDSource) Init(context.Context) error
Init initializes the output pool.
func (*StaticTraceIDSource) NextBatch ¶
func (s *StaticTraceIDSource) NextBatch(ctx context.Context) (*vectorized.RecordBatch, error)
NextBatch emits the next static lookup batch.
func (*StaticTraceIDSource) OutputSchema ¶
func (s *StaticTraceIDSource) OutputSchema() *vectorized.BatchSchema
OutputSchema returns the Phase-1 schema.
type VectorizedConfig ¶
type VectorizedConfig struct {
BatchSize int
// QueryMemoryMiB is a soft span-loading threshold: it caps the cumulative
// uncompressed span bytes fetched from disk per query. SIDX responses, tags,
// record-batch overhead, and other per-query allocations are not counted.
// The first block always loads regardless of the budget (first-block exception),
// so a single oversized block may exceed this value.
QueryMemoryMiB int
Enabled bool
}
VectorizedConfig controls the v1 vectorized Trace query path.
func DefaultConfig ¶
func DefaultConfig() VectorizedConfig
DefaultConfig returns the default trace vectorized configuration — enabled, with the shared default batch size and a 256 MiB per-query memory budget.
Enabled also selects the liaison<->data wire format: a flag-on distributed data node emits the native columnar frame instead of protobuf. A liaison decodes both (it dispatches on the frame magic byte per message), but an older liaison has no frame decoder at all, so a cluster must upgrade liaison nodes BEFORE data nodes. See docs/operation/upgrade.md.
To roll back the vec path entirely, pass --trace-vectorized-enabled=false on the standalone or data-node command line and restart; the row path resumes immediately.
func (VectorizedConfig) Validate ¶
func (c VectorizedConfig) Validate() error
Validate rejects invalid trace vectorized configurations.
Source Files