fix: handle traces intrinsic fields

This commit is contained in:
Tushar Vats
2026-02-03 05:10:09 +05:30
parent f6f8c78aaf
commit 45efda0b93
6 changed files with 173 additions and 41 deletions

View File

@@ -11,5 +11,8 @@
"[go]": {
"editor.formatOnSave": true,
"editor.defaultFormatter": "golang.go"
},
"[sql]": {
"editor.defaultFormatter": "adpyke.vscode-sql-formatter"
}
}

View File

@@ -134,6 +134,12 @@ var (
FieldContext: telemetrytypes.FieldContextSpan,
FieldDataType: telemetrytypes.FieldDataTypeNumber,
},
"timestamp": {
Name: "timestamp",
Signal: telemetrytypes.SignalTraces,
FieldContext: telemetrytypes.FieldContextSpan,
FieldDataType: telemetrytypes.FieldDataTypeNumber,
},
}
CalculatedFields = map[string]telemetrytypes.TelemetryFieldKey{

View File

@@ -187,7 +187,7 @@ func (m *defaultFieldMapper) getColumn(
case telemetrytypes.FieldDataTypeBool:
return indexV3Columns["attributes_bool"], nil
}
case telemetrytypes.FieldContextSpan, telemetrytypes.FieldContextUnspecified:
case telemetrytypes.FieldContextSpan:
// Check if this is a span scope field
if strings.ToLower(key.Name) == SpanSearchScopeRoot || strings.ToLower(key.Name) == SpanSearchScopeEntryPoint {
// The actual SQL will be generated in the condition builder
@@ -196,20 +196,28 @@ func (m *defaultFieldMapper) getColumn(
// TODO(srikanthccv): remove this when it's safe to remove
// issue with CH aliasing
/*
NOTE: There are fields which are deprecated for only to not show up as user suggestion and is possible that
they don't have a mapping in oldToNew map. So we need to look up in indexV3Columns directly for those fields.
For example: kind, timestamp etc.
*/
if _, ok := CalculatedFieldsDeprecated[key.Name]; ok {
return indexV3Columns[oldToNew[key.Name]], nil
// Check if we have a mapping for the deprecated calculated field
if col, ok := indexV3Columns[oldToNew[key.Name]]; ok {
return col, nil
}
}
if _, ok := IntrinsicFieldsDeprecated[key.Name]; ok {
// Check if we have a mapping for the deprecated intrinsic field
if _, ok := indexV3Columns[oldToNew[key.Name]]; ok {
return indexV3Columns[oldToNew[key.Name]], nil
if col, ok := indexV3Columns[oldToNew[key.Name]]; ok {
return col, nil
}
}
if col, ok := indexV3Columns[key.Name]; ok {
return col, nil
}
return nil, qbtypes.ErrColumnNotFound
}
return nil, qbtypes.ErrColumnNotFound
}

View File

@@ -74,6 +74,41 @@ func (b *traceQueryStatementBuilder) Build(
return nil, err
}
/*
Adding a tech debt note here:
This piece of code is a hot fix and should be removed once we close issue: engineering-pod/issues/3622
*/
/*
-------------------------------- Start of tech debt ----------------------------
*/
if requestType == qbtypes.RequestTypeRaw {
selectedFields := query.SelectFields
if len(selectedFields) == 0 {
sortedKeys := maps.Keys(DefaultFields)
slices.Sort(sortedKeys)
for _, key := range sortedKeys {
selectedFields = append(selectedFields, DefaultFields[key])
}
query.SelectFields = selectedFields
}
selectFieldKeys := []string{}
for _, field := range selectedFields {
selectFieldKeys = append(selectFieldKeys, field.Name)
}
for _, x := range []string{"timestamp", "span_id", "trace_id"} {
if !slices.Contains(selectFieldKeys, x) {
query.SelectFields = append(query.SelectFields, DefaultFields[x])
}
}
}
/*
-------------------------------- End of tech debt ----------------------------
*/
query = b.adjustKeys(ctx, keys, query, requestType)
// Check if filter contains trace_id(s) and optimize time range if needed
@@ -167,19 +202,13 @@ func (b *traceQueryStatementBuilder) adjustKeys(ctx context.Context, keys map[st
// 1. to not fail filter expression that use deprecated cols
// 2. this could have been moved to metadata fetching itself, however, that
// would mean, they also show up in suggestions we we don't want to do
// 3. reason for not doing a simple append is to keep intrinsic/calculated field first so that it gets
// priority in multi_if sql expression
for fieldKeyName, fieldKey := range IntrinsicFieldsDeprecated {
if _, ok := keys[fieldKeyName]; !ok {
keys[fieldKeyName] = []*telemetrytypes.TelemetryFieldKey{&fieldKey}
} else {
keys[fieldKeyName] = append(keys[fieldKeyName], &fieldKey)
}
keys[fieldKeyName] = append([]*telemetrytypes.TelemetryFieldKey{&fieldKey}, keys[fieldKeyName]...)
}
for fieldKeyName, fieldKey := range CalculatedFieldsDeprecated {
if _, ok := keys[fieldKeyName]; !ok {
keys[fieldKeyName] = []*telemetrytypes.TelemetryFieldKey{&fieldKey}
} else {
keys[fieldKeyName] = append(keys[fieldKeyName], &fieldKey)
}
keys[fieldKeyName] = append([]*telemetrytypes.TelemetryFieldKey{&fieldKey}, keys[fieldKeyName]...)
}
// Adjust keys for alias expressions in aggregations
@@ -282,29 +311,8 @@ func (b *traceQueryStatementBuilder) buildListQuery(
cteArgs = append(cteArgs, args)
}
selectedFields := query.SelectFields
if len(selectedFields) == 0 {
sortedKeys := maps.Keys(DefaultFields)
slices.Sort(sortedKeys)
for _, key := range sortedKeys {
selectedFields = append(selectedFields, DefaultFields[key])
}
}
selectFieldKeys := []string{}
for _, field := range selectedFields {
selectFieldKeys = append(selectFieldKeys, field.Name)
}
for _, x := range []string{"timestamp", "span_id", "trace_id"} {
if !slices.Contains(selectFieldKeys, x) {
selectedFields = append(selectedFields, DefaultFields[x])
}
}
// TODO: should we deprecate `SelectFields` and return everything from a span like we do for logs?
for _, field := range selectedFields {
for _, field := range query.SelectFields {
colExpr, err := b.fm.ColumnExpressionFor(ctx, &field, keys)
if err != nil {
return nil, err

View File

@@ -359,7 +359,7 @@ func TestStatementBuilder(t *testing.T) {
},
},
expected: qbtypes.Statement{
Query: "WITH __resource_filter AS (SELECT fingerprint FROM signoz_traces.distributed_traces_v3_resource WHERE (simpleJSONExtractString(labels, 'service.name') = ? AND labels LIKE ? AND labels LIKE ?) AND seen_at_ts_bucket_start >= ? AND seen_at_ts_bucket_start <= ?), __limit_cte AS (SELECT toString(multiIf(response_status_code <> ?, response_status_code, NULL)) AS `responseStatusCode`, quantile(0.90)(multiIf(duration_nano <> ?, duration_nano, NULL)) AS __result_0 FROM signoz_traces.distributed_signoz_index_v3 WHERE resource_fingerprint GLOBAL IN (SELECT fingerprint FROM __resource_filter) AND true AND timestamp >= ? AND timestamp < ? AND ts_bucket_start >= ? AND ts_bucket_start <= ? GROUP BY `responseStatusCode` ORDER BY __result_0 DESC LIMIT ?) SELECT toStartOfInterval(timestamp, INTERVAL 30 SECOND) AS ts, toString(multiIf(response_status_code <> ?, response_status_code, NULL)) AS `responseStatusCode`, quantile(0.90)(multiIf(duration_nano <> ?, duration_nano, NULL)) AS __result_0 FROM signoz_traces.distributed_signoz_index_v3 WHERE resource_fingerprint GLOBAL IN (SELECT fingerprint FROM __resource_filter) AND true AND timestamp >= ? AND timestamp < ? AND ts_bucket_start >= ? AND ts_bucket_start <= ? AND (`responseStatusCode`) GLOBAL IN (SELECT `responseStatusCode` FROM __limit_cte) GROUP BY ts, `responseStatusCode`",
Query: "WITH __resource_filter AS (SELECT fingerprint FROM signoz_traces.distributed_traces_v3_resource WHERE (simpleJSONExtractString(labels, 'service.name') = ? AND labels LIKE ? AND labels LIKE ?) AND seen_at_ts_bucket_start >= ? AND seen_at_ts_bucket_start <= ?), __limit_cte AS (SELECT toString(multiIf(response_status_code <> ?, response_status_code, NULL)) AS `responseStatusCode`, quantile(0.90)(multiIf(duration_nano <> ?, toFloat64(duration_nano), NULL)) AS __result_0 FROM signoz_traces.distributed_signoz_index_v3 WHERE resource_fingerprint GLOBAL IN (SELECT fingerprint FROM __resource_filter) AND true AND timestamp >= ? AND timestamp < ? AND ts_bucket_start >= ? AND ts_bucket_start <= ? GROUP BY `responseStatusCode` ORDER BY __result_0 DESC LIMIT ?) SELECT toStartOfInterval(timestamp, INTERVAL 30 SECOND) AS ts, toString(multiIf(response_status_code <> ?, response_status_code, NULL)) AS `responseStatusCode`, quantile(0.90)(multiIf(duration_nano <> ?, toFloat64(duration_nano), NULL)) AS __result_0 FROM signoz_traces.distributed_signoz_index_v3 WHERE resource_fingerprint GLOBAL IN (SELECT fingerprint FROM __resource_filter) AND true AND timestamp >= ? AND timestamp < ? AND ts_bucket_start >= ? AND ts_bucket_start <= ? AND (`responseStatusCode`) GLOBAL IN (SELECT `responseStatusCode` FROM __limit_cte) GROUP BY ts, `responseStatusCode`",
Args: []any{"redis-manual", "%service.name%", "%service.name\":\"redis-manual%", uint64(1747945619), uint64(1747983448), "", 0, "1747947419000000000", "1747983448000000000", uint64(1747945619), uint64(1747983448), 10, "", 0, "1747947419000000000", "1747983448000000000", uint64(1747945619), uint64(1747983448)},
},
expectedErr: nil,
@@ -483,7 +483,7 @@ func TestStatementBuilderListQuery(t *testing.T) {
Limit: 10,
},
expected: qbtypes.Statement{
Query: "WITH __resource_filter AS (SELECT fingerprint FROM signoz_traces.distributed_traces_v3_resource WHERE (simpleJSONExtractString(labels, 'service.name') = ? AND labels LIKE ? AND labels LIKE ?) AND seen_at_ts_bucket_start >= ? AND seen_at_ts_bucket_start <= ?) SELECT duration_nano AS `duration_nano`, name AS `name`, response_status_code AS `response_status_code`, multiIf(resource.`service.name` IS NOT NULL, resource.`service.name`::String, `resource_string_service$$name_exists`==true, `resource_string_service$$name`, NULL) AS `service.name`, span_id AS `span_id`, timestamp AS `timestamp`, trace_id AS `trace_id` FROM signoz_traces.distributed_signoz_index_v3 WHERE resource_fingerprint GLOBAL IN (SELECT fingerprint FROM __resource_filter) AND true AND timestamp >= ? AND timestamp < ? AND ts_bucket_start >= ? AND ts_bucket_start <= ? ORDER BY attributes_string['user.id'] AS `user.id` desc LIMIT ?",
Query: "WITH __resource_filter AS (SELECT fingerprint FROM signoz_traces.distributed_traces_v3_resource WHERE (simpleJSONExtractString(labels, 'service.name') = ? AND labels LIKE ? AND labels LIKE ?) AND seen_at_ts_bucket_start >= ? AND seen_at_ts_bucket_start <= ?) SELECT duration_nano AS `duration_nano`, name AS `name`, response_status_code AS `response_status_code`, multiIf(resource.`service.name` IS NOT NULL, resource.`service.name`::String, mapContains(resources_string, 'service.name'), resources_string['service.name'], NULL) AS `service.name`, span_id AS `span_id`, timestamp AS `timestamp`, trace_id AS `trace_id` FROM signoz_traces.distributed_signoz_index_v3 WHERE resource_fingerprint GLOBAL IN (SELECT fingerprint FROM __resource_filter) AND true AND timestamp >= ? AND timestamp < ? AND ts_bucket_start >= ? AND ts_bucket_start <= ? ORDER BY attributes_string['user.id'] AS `user.id` desc LIMIT ?",
Args: []any{"redis-manual", "%service.name%", "%service.name\":\"redis-manual%", uint64(1747945619), uint64(1747983448), "1747947419000000000", "1747983448000000000", uint64(1747945619), uint64(1747983448), 10},
},
expectedErr: nil,
@@ -698,6 +698,113 @@ func TestStatementBuilderListQuery(t *testing.T) {
}
}
func TestStatementBuilderListQueryWithCorruptData(t *testing.T) {
cases := []struct {
name string
requestType qbtypes.RequestType
query qbtypes.QueryBuilderQuery[qbtypes.TraceAggregation]
keysMap map[string][]*telemetrytypes.TelemetryFieldKey
expected qbtypes.Statement
expectedErr error
}{
{
name: "List query with empty select fields",
requestType: qbtypes.RequestTypeRaw,
keysMap: map[string][]*telemetrytypes.TelemetryFieldKey{
"timestamp": {
{
Name: "timestamp",
Signal: telemetrytypes.SignalTraces,
FieldContext: telemetrytypes.FieldContextAttribute,
FieldDataType: telemetrytypes.FieldDataTypeString,
},
},
},
query: qbtypes.QueryBuilderQuery[qbtypes.TraceAggregation]{
Signal: telemetrytypes.SignalTraces,
StepInterval: qbtypes.Step{Duration: 30 * time.Second},
Filter: &qbtypes.Filter{},
Limit: 10,
},
expected: qbtypes.Statement{
Query: "WITH __resource_filter AS (SELECT fingerprint FROM signoz_traces.distributed_traces_v3_resource WHERE seen_at_ts_bucket_start >= ? AND seen_at_ts_bucket_start <= ?) SELECT duration_nano AS `duration_nano`, name AS `name`, response_status_code AS `response_status_code`, multiIf(resource.`service.name` IS NOT NULL, resource.`service.name`::String, mapContains(resources_string, 'service.name'), resources_string['service.name'], NULL) AS `service.name`, span_id AS `span_id`, timestamp AS `timestamp`, trace_id AS `trace_id` FROM signoz_traces.distributed_signoz_index_v3 WHERE resource_fingerprint GLOBAL IN (SELECT fingerprint FROM __resource_filter) AND timestamp >= ? AND timestamp < ? AND ts_bucket_start >= ? AND ts_bucket_start <= ? LIMIT ?",
Args: []any{uint64(1747945619), uint64(1747983448), "1747947419000000000", "1747983448000000000", uint64(1747945619), uint64(1747983448), 10},
},
expectedErr: nil,
},
{
name: "List query with empty select fields and order by timestamp",
requestType: qbtypes.RequestTypeRaw,
keysMap: map[string][]*telemetrytypes.TelemetryFieldKey{
"timestamp": {
{
Name: "timestamp",
Signal: telemetrytypes.SignalTraces,
FieldContext: telemetrytypes.FieldContextAttribute,
FieldDataType: telemetrytypes.FieldDataTypeString,
},
},
},
query: qbtypes.QueryBuilderQuery[qbtypes.TraceAggregation]{
Signal: telemetrytypes.SignalTraces,
StepInterval: qbtypes.Step{Duration: 30 * time.Second},
Filter: &qbtypes.Filter{},
Limit: 10,
Order: []qbtypes.OrderBy{{
Key: qbtypes.OrderByKey{
TelemetryFieldKey: telemetrytypes.TelemetryFieldKey{
Name: "timestamp",
},
},
Direction: qbtypes.OrderDirectionAsc,
}},
},
expected: qbtypes.Statement{
Query: "WITH __resource_filter AS (SELECT fingerprint FROM signoz_traces.distributed_traces_v3_resource WHERE seen_at_ts_bucket_start >= ? AND seen_at_ts_bucket_start <= ?) SELECT duration_nano AS `duration_nano`, name AS `name`, response_status_code AS `response_status_code`, multiIf(resource.`service.name` IS NOT NULL, resource.`service.name`::String, mapContains(resources_string, 'service.name'), resources_string['service.name'], NULL) AS `service.name`, span_id AS `span_id`, multiIf(toString(timestamp) != '', toString(timestamp), toString(attributes_string['timestamp']) != '', toString(attributes_string['timestamp']), NULL) AS `timestamp`, trace_id AS `trace_id` FROM signoz_traces.distributed_signoz_index_v3 WHERE resource_fingerprint GLOBAL IN (SELECT fingerprint FROM __resource_filter) AND timestamp >= ? AND timestamp < ? AND ts_bucket_start >= ? AND ts_bucket_start <= ? ORDER BY multiIf(toString(timestamp) != '', toString(timestamp), toString(attributes_string['timestamp']) != '', toString(attributes_string['timestamp']), NULL) AS `timestamp` asc LIMIT ?",
Args: []any{uint64(1747945619), uint64(1747983448), "1747947419000000000", "1747983448000000000", uint64(1747945619), uint64(1747983448), 10},
},
expectedErr: nil,
},
}
for _, c := range cases {
t.Run(c.name, func(t *testing.T) {
fm := NewFieldMapper()
cb := NewConditionBuilder(fm)
mockMetadataStore := telemetrytypestest.NewMockMetadataStore()
mockMetadataStore.KeysMap = c.keysMap
if mockMetadataStore.KeysMap == nil {
mockMetadataStore.KeysMap = buildCompleteFieldKeyMap()
}
aggExprRewriter := querybuilder.NewAggExprRewriter(instrumentationtest.New().ToProviderSettings(), nil, fm, cb, nil)
resourceFilterStmtBuilder := resourceFilterStmtBuilder()
statementBuilder := NewTraceQueryStatementBuilder(
instrumentationtest.New().ToProviderSettings(),
mockMetadataStore,
fm,
cb,
resourceFilterStmtBuilder,
aggExprRewriter,
nil,
)
q, err := statementBuilder.Build(context.Background(), 1747947419000, 1747983448000, c.requestType, c.query, nil)
if c.expectedErr != nil {
require.Error(t, err)
require.Contains(t, err.Error(), c.expectedErr.Error())
} else {
require.NoError(t, err)
require.Equal(t, c.expected.Query, q.Query)
require.Equal(t, c.expected.Args, q.Args)
require.Equal(t, c.expected.Warnings, q.Warnings)
}
})
}
}
func TestStatementBuilderTraceQuery(t *testing.T) {
cases := []struct {
name string

View File

@@ -322,8 +322,8 @@ func TestTraceOperatorStatementBuilder(t *testing.T) {
},
},
expected: qbtypes.Statement{
Query: "WITH toDateTime64(1747947419000000000, 9) AS t_from, toDateTime64(1747983448000000000, 9) AS t_to, 1747945619 AS bucket_from, 1747983448 AS bucket_to, all_spans AS (SELECT *, resource_string_service$$name AS `service.name` FROM signoz_traces.distributed_signoz_index_v3 WHERE timestamp >= ? AND timestamp < ? AND ts_bucket_start >= ? AND ts_bucket_start <= ?), __resource_filter_A AS (SELECT fingerprint FROM signoz_traces.distributed_traces_v3_resource WHERE (simpleJSONExtractString(labels, 'service.name') = ? AND labels LIKE ? AND labels LIKE ?) AND seen_at_ts_bucket_start >= ? AND seen_at_ts_bucket_start <= ?), A AS (SELECT * FROM signoz_traces.distributed_signoz_index_v3 WHERE resource_fingerprint GLOBAL IN (SELECT fingerprint FROM __resource_filter_A) AND timestamp >= ? AND timestamp < ? AND ts_bucket_start >= ? AND ts_bucket_start <= ? AND true), __resource_filter_B AS (SELECT fingerprint FROM signoz_traces.distributed_traces_v3_resource WHERE true AND seen_at_ts_bucket_start >= ? AND seen_at_ts_bucket_start <= ?), B AS (SELECT * FROM signoz_traces.distributed_signoz_index_v3 WHERE resource_fingerprint GLOBAL IN (SELECT fingerprint FROM __resource_filter_B) AND timestamp >= ? AND timestamp < ? AND ts_bucket_start >= ? AND ts_bucket_start <= ? AND toFloat64(response_status_code) < ?), A_AND_B AS (SELECT l.* FROM A AS l INNER JOIN B AS r ON l.trace_id = r.trace_id) SELECT toString(multiIf(multiIf(resource.`service.name` IS NOT NULL, resource.`service.name`::String, mapContains(resources_string, 'service.name'), resources_string['service.name'], NULL) IS NOT NULL, multiIf(resource.`service.name` IS NOT NULL, resource.`service.name`::String, mapContains(resources_string, 'service.name'), resources_string['service.name'], NULL), NULL)) AS `service.name`, avg(multiIf(duration_nano <> ?, duration_nano, NULL)) AS __result_0 FROM A_AND_B GROUP BY `service.name` ORDER BY __result_0 desc SETTINGS distributed_product_mode='allow', max_memory_usage=10000000000",
Args: []any{"1747947419000000000", "1747983448000000000", uint64(1747945619), uint64(1747983448), "frontend", "%service.name%", "%service.name\":\"frontend%", uint64(1747945619), uint64(1747983448), "1747947419000000000", "1747983448000000000", uint64(1747945619), uint64(1747983448), uint64(1747945619), uint64(1747983448), "1747947419000000000", "1747983448000000000", uint64(1747945619), uint64(1747983448), float64(400), 0},
Query: "WITH toDateTime64(1747947419000000000, 9) AS t_from, toDateTime64(1747983448000000000, 9) AS t_to, 1747945619 AS bucket_from, 1747983448 AS bucket_to, all_spans AS (SELECT *, resource_string_service$$name AS `service.name` FROM signoz_traces.distributed_signoz_index_v3 WHERE timestamp >= ? AND timestamp < ? AND ts_bucket_start >= ? AND ts_bucket_start <= ?), __resource_filter_A AS (SELECT fingerprint FROM signoz_traces.distributed_traces_v3_resource WHERE (simpleJSONExtractString(labels, 'service.name') = ? AND labels LIKE ? AND labels LIKE ?) AND seen_at_ts_bucket_start >= ? AND seen_at_ts_bucket_start <= ?), A AS (SELECT * FROM signoz_traces.distributed_signoz_index_v3 WHERE resource_fingerprint GLOBAL IN (SELECT fingerprint FROM __resource_filter_A) AND timestamp >= ? AND timestamp < ? AND ts_bucket_start >= ? AND ts_bucket_start <= ? AND true), __resource_filter_B AS (SELECT fingerprint FROM signoz_traces.distributed_traces_v3_resource WHERE true AND seen_at_ts_bucket_start >= ? AND seen_at_ts_bucket_start <= ?), B AS (SELECT * FROM signoz_traces.distributed_signoz_index_v3 WHERE resource_fingerprint GLOBAL IN (SELECT fingerprint FROM __resource_filter_B) AND timestamp >= ? AND timestamp < ? AND ts_bucket_start >= ? AND ts_bucket_start <= ? AND toFloat64(response_status_code) < ?), A_AND_B AS (SELECT l.* FROM A AS l INNER JOIN B AS r ON l.trace_id = r.trace_id) SELECT toString(multiIf(multiIf(resource.`service.name` IS NOT NULL, resource.`service.name`::String, mapContains(resources_string, 'service.name'), resources_string['service.name'], NULL) IS NOT NULL, multiIf(resource.`service.name` IS NOT NULL, resource.`service.name`::String, mapContains(resources_string, 'service.name'), resources_string['service.name'], NULL), NULL)) AS `service.name`, avg(multiIf(duration_nano <> ?, toFloat64(duration_nano), mapContains(attributes_number, 'duration_nano') = ?, toFloat64(attributes_number['duration_nano']), NULL)) AS __result_0 FROM A_AND_B GROUP BY `service.name` ORDER BY __result_0 desc SETTINGS distributed_product_mode='allow', max_memory_usage=10000000000",
Args: []any{"1747947419000000000", "1747983448000000000", uint64(1747945619), uint64(1747983448), "frontend", "%service.name%", "%service.name\":\"frontend%", uint64(1747945619), uint64(1747983448), "1747947419000000000", "1747983448000000000", uint64(1747945619), uint64(1747983448), uint64(1747945619), uint64(1747983448), "1747947419000000000", "1747983448000000000", uint64(1747945619), uint64(1747983448), float64(400), 0, true},
},
expectedErr: nil,
},