Files
Martin DisibioandGitHub 5dc8779127 [Parquet] Add traceid index to vparquet2 and use it when finding trace by id (#2697)
* Add traceid index to vparquet2 and use it when finding trace by id

* Update to detect when trace ID is beyond the last row group and bail

* simplify

* Add index to vparquet3

* go away

* lint

* changelog

* Fix index to not flush duplicate entry, which could happen if the block ended on exactly the configured rowgroup size while flushing

* the opposite of move fast and break things

* more optionalality

* Add new env var VPARQUET_INDEX=0 to disable the index lookup
2023-08-24 06:52:16 -04:00

49 lines
864 B
Go

package vparquet3
import (
"bytes"
"encoding/json"
"sort"
"github.com/grafana/tempo/tempodb/encoding/common"
)
type index struct {
lastID common.ID
dirty bool
RowGroups []common.ID `json:"rowGroups"`
}
func (i *index) Add(id common.ID) {
i.lastID = id
i.dirty = true
}
func (i *index) Flush() {
if i.dirty {
i.RowGroups = append(i.RowGroups, i.lastID)
i.dirty = false
}
}
func (i *index) Marshal() ([]byte, error) {
return json.Marshal(i)
}
func (i *index) Find(id common.ID) int {
n := sort.Search(len(i.RowGroups), func(j int) bool {
return bytes.Compare(id, i.RowGroups[j]) <= 0
})
if n >= len(i.RowGroups) {
// Beyond the last row group. This is the only
// area where presence can be ruled out.
return -1
}
return n
}
func unmarshalIndex(b []byte) (*index, error) {
i := &index{}
return i, json.Unmarshal(b, i)
}