feat(unified): add index scoring model config (#116210)
* feat(unified): add bm25 index scoring model We want try BM25 scoring model since they have global scoring which we can probably re-use for fan-in/fan-out logic https://github.com/blevesearch/bleve/blob/32d98823c4b7482c62cc6c847508ed7659c23c37/docs/scoring.md#global-scoring * fix(plugins): update plugin test data
This commit is contained in:
@@ -600,6 +600,7 @@ type Cfg struct {
|
||||
IndexRebuildInterval time.Duration
|
||||
IndexCacheTTL time.Duration
|
||||
IndexMinUpdateInterval time.Duration // Don't update index if it was updated less than this interval ago.
|
||||
IndexScoringModel string // Note: Temporary config to switch the index scoring model and will be removed soon.
|
||||
MaxFileIndexAge time.Duration // Max age of file-based indexes. Index older than this will be rebuilt asynchronously.
|
||||
MinFileIndexBuildVersion string // Minimum version of Grafana that built the file-based index. If index was built with older Grafana, it will be rebuilt asynchronously.
|
||||
EnableSharding bool
|
||||
|
||||
@@ -123,6 +123,10 @@ func (cfg *Cfg) setUnifiedStorageConfig() {
|
||||
cfg.IndexRebuildInterval = section.Key("index_rebuild_interval").MustDuration(24 * time.Hour)
|
||||
cfg.IndexCacheTTL = section.Key("index_cache_ttl").MustDuration(10 * time.Minute)
|
||||
cfg.IndexMinUpdateInterval = section.Key("index_min_update_interval").MustDuration(0)
|
||||
cfg.IndexScoringModel = section.Key("index_scoring_model").MustString("")
|
||||
if cfg.IndexScoringModel != "" {
|
||||
cfg.Logger.Info("Index scoring model set", "model", cfg.IndexScoringModel)
|
||||
}
|
||||
cfg.SprinklesApiServer = section.Key("sprinkles_api_server").String()
|
||||
cfg.SprinklesApiServerPageLimit = section.Key("sprinkles_api_server_page_limit").MustInt(10000)
|
||||
cfg.CACertPath = section.Key("ca_cert_path").String()
|
||||
|
||||
@@ -81,6 +81,11 @@ type BleveOptions struct {
|
||||
// Indexes that are not owned by current instance are eligible for cleanup.
|
||||
// If nil, all indexes are owned by the current instance.
|
||||
OwnsIndex func(key resource.NamespacedResource) (bool, error)
|
||||
|
||||
// ScoringModel defines the scoring model used for the bleve indexes
|
||||
// Default: index.TFIDFScoring
|
||||
// Supported values: index.TFIDFScoring and index.BM25Scoring
|
||||
ScoringModel string
|
||||
}
|
||||
|
||||
type bleveBackend struct {
|
||||
@@ -368,7 +373,7 @@ func (b *bleveBackend) BuildIndex(
|
||||
attribute.String("reason", indexBuildReason),
|
||||
)
|
||||
|
||||
mapper, err := GetBleveMappings(fields)
|
||||
mapper, err := GetBleveMappings(b.opts.ScoringModel, fields)
|
||||
if err != nil {
|
||||
return nil, err
|
||||
}
|
||||
|
||||
@@ -4,6 +4,7 @@ import (
|
||||
"context"
|
||||
"testing"
|
||||
|
||||
index "github.com/blevesearch/bleve_index_api"
|
||||
"github.com/stretchr/testify/require"
|
||||
|
||||
"github.com/grafana/grafana/pkg/storage/unified/resource"
|
||||
@@ -19,6 +20,7 @@ func TestBleveSearchBackend(t *testing.T) {
|
||||
backend, err := NewBleveBackend(BleveOptions{
|
||||
Root: tempDir,
|
||||
FileThreshold: 5,
|
||||
ScoringModel: index.BM25Scoring,
|
||||
}, nil)
|
||||
require.NoError(t, err)
|
||||
require.NotNil(t, backend)
|
||||
@@ -52,3 +54,32 @@ func TestSearchBackendBenchmark(t *testing.T) {
|
||||
|
||||
unitest.BenchmarkSearchBackend(t, backend, opts)
|
||||
}
|
||||
|
||||
func BenchmarkScoringModels(b *testing.B) {
|
||||
models := []string{index.TFIDFScoring, index.BM25Scoring}
|
||||
|
||||
for _, model := range models {
|
||||
b.Run(model, func(b *testing.B) {
|
||||
tempDir := b.TempDir()
|
||||
|
||||
backend, err := NewBleveBackend(BleveOptions{
|
||||
Root: tempDir,
|
||||
ScoringModel: model,
|
||||
}, nil)
|
||||
require.NoError(b, err)
|
||||
require.NotNil(b, backend)
|
||||
|
||||
b.Cleanup(backend.Stop)
|
||||
|
||||
opts := &unitest.BenchmarkOptions{
|
||||
NumResources: 1000,
|
||||
Concurrency: 4,
|
||||
NumNamespaces: 10,
|
||||
NumGroups: 10,
|
||||
NumResourceTypes: 10,
|
||||
}
|
||||
|
||||
unitest.BenchmarkSearchBackend(b, backend, opts)
|
||||
})
|
||||
}
|
||||
}
|
||||
|
||||
@@ -5,13 +5,15 @@ import (
|
||||
"github.com/blevesearch/bleve/v2/analysis/analyzer/keyword"
|
||||
"github.com/blevesearch/bleve/v2/analysis/analyzer/standard"
|
||||
"github.com/blevesearch/bleve/v2/mapping"
|
||||
|
||||
"github.com/grafana/grafana/pkg/storage/unified/resource"
|
||||
"github.com/grafana/grafana/pkg/storage/unified/resourcepb"
|
||||
)
|
||||
|
||||
func GetBleveMappings(fields resource.SearchableDocumentFields) (mapping.IndexMapping, error) {
|
||||
func GetBleveMappings(scoringModel string, fields resource.SearchableDocumentFields) (mapping.IndexMapping, error) {
|
||||
mapper := bleve.NewIndexMapping()
|
||||
if scoringModel != "" {
|
||||
mapper.ScoringModel = scoringModel
|
||||
}
|
||||
|
||||
err := RegisterCustomAnalyzers(mapper)
|
||||
if err != nil {
|
||||
|
||||
@@ -13,7 +13,7 @@ import (
|
||||
)
|
||||
|
||||
func TestDocumentMapping(t *testing.T) {
|
||||
mappings, err := search.GetBleveMappings(nil)
|
||||
mappings, err := search.GetBleveMappings("", nil)
|
||||
require.NoError(t, err)
|
||||
data := resource.IndexableDocument{
|
||||
Title: "title",
|
||||
|
||||
@@ -7,6 +7,7 @@ import (
|
||||
"testing"
|
||||
|
||||
"github.com/blevesearch/bleve/v2"
|
||||
index "github.com/blevesearch/bleve_index_api"
|
||||
"github.com/stretchr/testify/require"
|
||||
|
||||
"github.com/grafana/grafana/pkg/apimachinery/identity"
|
||||
@@ -258,6 +259,7 @@ func newTestDashboardsIndex(t testing.TB, threshold int64, size int64, writer re
|
||||
backend, err := search.NewBleveBackend(search.BleveOptions{
|
||||
Root: t.TempDir(),
|
||||
FileThreshold: threshold, // use in-memory for tests
|
||||
ScoringModel: index.BM25Scoring,
|
||||
}, nil)
|
||||
require.NoError(t, err)
|
||||
|
||||
|
||||
@@ -14,6 +14,7 @@ import (
|
||||
"time"
|
||||
|
||||
"github.com/blevesearch/bleve/v2"
|
||||
index "github.com/blevesearch/bleve_index_api"
|
||||
"github.com/prometheus/client_golang/prometheus"
|
||||
"github.com/prometheus/client_golang/prometheus/testutil"
|
||||
"github.com/stretchr/testify/assert"
|
||||
@@ -50,6 +51,7 @@ func TestBleveBackend(t *testing.T) {
|
||||
backend, err := NewBleveBackend(BleveOptions{
|
||||
Root: tmpdir,
|
||||
FileThreshold: 5, // with more than 5 items we create a file on disk
|
||||
ScoringModel: index.BM25Scoring,
|
||||
}, nil)
|
||||
require.NoError(t, err)
|
||||
t.Cleanup(backend.Stop)
|
||||
@@ -773,6 +775,7 @@ func setupBleveBackend(t *testing.T, options ...setupOption) (*bleveBackend, pro
|
||||
IndexCacheTTL: defaultIndexCacheTTL,
|
||||
Logger: log.NewNopLogger(),
|
||||
BuildVersion: buildVersion,
|
||||
ScoringModel: index.BM25Scoring,
|
||||
}
|
||||
for _, opt := range options {
|
||||
opt(&opts)
|
||||
|
||||
@@ -46,6 +46,7 @@ func NewSearchOptions(
|
||||
BuildVersion: cfg.BuildVersion,
|
||||
OwnsIndex: ownsIndexFn,
|
||||
IndexMinUpdateInterval: cfg.IndexMinUpdateInterval,
|
||||
ScoringModel: cfg.IndexScoringModel,
|
||||
}, indexMetrics)
|
||||
|
||||
if err != nil {
|
||||
|
||||
@@ -6,6 +6,7 @@ import (
|
||||
"testing"
|
||||
"time"
|
||||
|
||||
index "github.com/blevesearch/bleve_index_api"
|
||||
"github.com/go-jose/go-jose/v4/jwt"
|
||||
"github.com/prometheus/client_golang/prometheus"
|
||||
"github.com/stretchr/testify/require"
|
||||
@@ -129,21 +130,28 @@ func TestIntegrationSearchAndStorage(t *testing.T) {
|
||||
|
||||
ctx := context.Background()
|
||||
|
||||
// Create a new bleve backend
|
||||
search, err := search.NewBleveBackend(search.BleveOptions{
|
||||
FileThreshold: 0,
|
||||
Root: t.TempDir(),
|
||||
}, nil)
|
||||
require.NoError(t, err)
|
||||
require.NotNil(t, search)
|
||||
t.Cleanup(search.Stop)
|
||||
scoringModels := []string{index.TFIDFScoring, index.BM25Scoring}
|
||||
|
||||
// Create a new resource backend
|
||||
storage, _ := newTestBackend(t, false, 0)
|
||||
require.NotNil(t, storage)
|
||||
for _, model := range scoringModels {
|
||||
t.Run(model, func(t *testing.T) {
|
||||
// Create a new bleve backend
|
||||
search, err := search.NewBleveBackend(search.BleveOptions{
|
||||
FileThreshold: 0,
|
||||
Root: t.TempDir(),
|
||||
ScoringModel: model,
|
||||
}, nil)
|
||||
require.NoError(t, err)
|
||||
require.NotNil(t, search)
|
||||
t.Cleanup(search.Stop)
|
||||
|
||||
// Run the shared storage and search tests
|
||||
unitest.RunTestSearchAndStorage(t, ctx, storage, search)
|
||||
// Create a new resource backend
|
||||
storage, _ := newTestBackend(t, false, 0)
|
||||
require.NotNil(t, storage)
|
||||
|
||||
// Run the shared storage and search tests
|
||||
unitest.RunTestSearchAndStorage(t, ctx, storage, search)
|
||||
})
|
||||
}
|
||||
}
|
||||
|
||||
func TestClientServer(t *testing.T) {
|
||||
|
||||
@@ -209,7 +209,7 @@
|
||||
"path": "public/plugins/grafana-azure-monitor-datasource/img/azure_monitor_cpu.png"
|
||||
}
|
||||
],
|
||||
"version": "12.3.0-pre",
|
||||
"version": "12.4.0-pre",
|
||||
"updated": "",
|
||||
"keywords": [
|
||||
"azure",
|
||||
@@ -589,7 +589,7 @@
|
||||
"hasUpdate": false,
|
||||
"defaultNavUrl": "/plugins/datagrid/",
|
||||
"category": "",
|
||||
"state": "beta",
|
||||
"state": "deprecated",
|
||||
"signature": "internal",
|
||||
"signatureType": "",
|
||||
"signatureOrg": "",
|
||||
@@ -880,7 +880,7 @@
|
||||
},
|
||||
"build": {},
|
||||
"screenshots": null,
|
||||
"version": "12.3.0-pre",
|
||||
"version": "12.4.0-pre",
|
||||
"updated": "",
|
||||
"keywords": null
|
||||
},
|
||||
@@ -934,7 +934,7 @@
|
||||
},
|
||||
"build": {},
|
||||
"screenshots": null,
|
||||
"version": "12.3.0-pre",
|
||||
"version": "12.4.0-pre",
|
||||
"updated": "",
|
||||
"keywords": [
|
||||
"grafana",
|
||||
@@ -1000,7 +1000,7 @@
|
||||
},
|
||||
"build": {},
|
||||
"screenshots": null,
|
||||
"version": "12.3.0-pre",
|
||||
"version": "12.4.0-pre",
|
||||
"updated": "",
|
||||
"keywords": null
|
||||
},
|
||||
@@ -1217,7 +1217,7 @@
|
||||
},
|
||||
"build": {},
|
||||
"screenshots": null,
|
||||
"version": "12.3.0-pre",
|
||||
"version": "12.4.0-pre",
|
||||
"updated": "",
|
||||
"keywords": null
|
||||
},
|
||||
@@ -1325,7 +1325,7 @@
|
||||
},
|
||||
"build": {},
|
||||
"screenshots": null,
|
||||
"version": "12.3.0-pre",
|
||||
"version": "12.4.0-pre",
|
||||
"updated": "",
|
||||
"keywords": null
|
||||
},
|
||||
@@ -1375,7 +1375,7 @@
|
||||
},
|
||||
"build": {},
|
||||
"screenshots": null,
|
||||
"version": "12.3.0-pre",
|
||||
"version": "12.4.0-pre",
|
||||
"updated": "",
|
||||
"keywords": null
|
||||
},
|
||||
@@ -1425,7 +1425,7 @@
|
||||
},
|
||||
"build": {},
|
||||
"screenshots": null,
|
||||
"version": "12.3.0-pre",
|
||||
"version": "12.4.0-pre",
|
||||
"updated": "",
|
||||
"keywords": null
|
||||
},
|
||||
@@ -1575,7 +1575,7 @@
|
||||
},
|
||||
"build": {},
|
||||
"screenshots": null,
|
||||
"version": "",
|
||||
"version": "12.4.0-pre",
|
||||
"updated": "",
|
||||
"keywords": null
|
||||
},
|
||||
@@ -1629,7 +1629,7 @@
|
||||
},
|
||||
"build": {},
|
||||
"screenshots": null,
|
||||
"version": "12.3.0-pre",
|
||||
"version": "12.4.0-pre",
|
||||
"updated": "",
|
||||
"keywords": [
|
||||
"grafana",
|
||||
@@ -1734,7 +1734,7 @@
|
||||
},
|
||||
"build": {},
|
||||
"screenshots": null,
|
||||
"version": "12.3.0-pre",
|
||||
"version": "12.4.0-pre",
|
||||
"updated": "",
|
||||
"keywords": null
|
||||
},
|
||||
@@ -2042,7 +2042,7 @@
|
||||
},
|
||||
"build": {},
|
||||
"screenshots": null,
|
||||
"version": "12.3.0-pre",
|
||||
"version": "12.4.0-pre",
|
||||
"updated": "",
|
||||
"keywords": null
|
||||
},
|
||||
@@ -2092,7 +2092,7 @@
|
||||
},
|
||||
"build": {},
|
||||
"screenshots": null,
|
||||
"version": "12.3.0-pre",
|
||||
"version": "12.4.0-pre",
|
||||
"updated": "",
|
||||
"keywords": null
|
||||
},
|
||||
@@ -2445,7 +2445,7 @@
|
||||
},
|
||||
"build": {},
|
||||
"screenshots": null,
|
||||
"version": "12.3.0-pre",
|
||||
"version": "12.4.0-pre",
|
||||
"updated": "",
|
||||
"keywords": null
|
||||
},
|
||||
|
||||
Reference in New Issue
Block a user