mirror of
https://github.com/elastic/kibana.git
synced 2025-04-23 17:28:26 -04:00
[APM] Optimize anomaly data loading strategy (#87522)
* [APM] Optimize anomaly data loading strategy Closes #86423. * Fix tests/types * Review feedback * Optimize ML calls for latency chart * Optimize ML loading for latency chart * Remove unused optimization * Update snapshots for E2E tests * Make sure area is stacked in correct order * Review feedback + log warning if more than one ML job was found * Review feedback
This commit is contained in:
Dario Gieselaar
GitHub
parent
f2699f9505
commit
a0da8bda04
23 changed files with 617 additions and 608 deletions
|
|
@ -4,9 +4,8 @@
|
|||
* you may not use this file except in compliance with the Elastic License.
|
||||
*/
|
||||
import { isFinite } from 'lodash';
|
||||
import { Maybe } from '../../typings/common';
|
||||
|
||||
// _.isNumber() returns true for NaN, _.isFinite() does not refine
|
||||
export function isFiniteNumber(value: Maybe<number>): value is number {
|
||||
export function isFiniteNumber(value: any): value is number {
|
||||
return isFinite(value);
|
||||
}
|
||||
|
|
|
|||
|
|
@ -100,7 +100,7 @@ export function LatencyChart({ height }: Props) {
|
|||
id="latencyChart"
|
||||
timeseries={latencyTimeseries}
|
||||
yLabelFormat={getResponseTimeTickFormatter(latencyFormatter)}
|
||||
anomalySeries={anomalyTimeseries}
|
||||
anomalyTimeseries={anomalyTimeseries}
|
||||
/>
|
||||
</EuiFlexItem>
|
||||
</EuiFlexGroup>
|
||||
|
|
|
|||
|
|
@ -55,7 +55,7 @@ interface Props {
|
|||
yTickFormat?: (y: number) => string;
|
||||
showAnnotations?: boolean;
|
||||
yDomain?: YDomainRange;
|
||||
anomalySeries?: ReturnType<
|
||||
anomalyTimeseries?: ReturnType<
|
||||
typeof getLatencyChartSelector
|
||||
>['anomalyTimeseries'];
|
||||
}
|
||||
|
|
@ -70,7 +70,7 @@ export function TimeseriesChart({
|
|||
yTickFormat,
|
||||
showAnnotations = true,
|
||||
yDomain,
|
||||
anomalySeries,
|
||||
anomalyTimeseries,
|
||||
}: Props) {
|
||||
const history = useHistory();
|
||||
const { annotations } = useAnnotationsContext();
|
||||
|
|
@ -90,6 +90,8 @@ export function TimeseriesChart({
|
|||
|
||||
const annotationColor = theme.eui.euiColorSecondary;
|
||||
|
||||
const allSeries = [...timeseries, ...(anomalyTimeseries?.boundaries ?? [])];
|
||||
|
||||
return (
|
||||
<ChartContainer hasData={!isEmpty} height={height} status={fetchStatus}>
|
||||
<Chart ref={chartRef} id={id}>
|
||||
|
|
@ -150,7 +152,7 @@ export function TimeseriesChart({
|
|||
/>
|
||||
)}
|
||||
|
||||
{timeseries.map((serie) => {
|
||||
{allSeries.map((serie) => {
|
||||
const Series = serie.type === 'area' ? AreaSeries : LineSeries;
|
||||
|
||||
return (
|
||||
|
|
@ -164,37 +166,28 @@ export function TimeseriesChart({
|
|||
data={isEmpty ? [] : serie.data}
|
||||
color={serie.color}
|
||||
curve={CurveType.CURVE_MONOTONE_X}
|
||||
hideInLegend={serie.hideLegend}
|
||||
fit={serie.fit ?? undefined}
|
||||
filterSeriesInTooltip={
|
||||
serie.hideTooltipValue ? () => false : undefined
|
||||
}
|
||||
stackAccessors={serie.stackAccessors ?? undefined}
|
||||
areaSeriesStyle={serie.areaSeriesStyle}
|
||||
lineSeriesStyle={serie.lineSeriesStyle}
|
||||
/>
|
||||
);
|
||||
})}
|
||||
|
||||
{anomalySeries?.bounderies && (
|
||||
<AreaSeries
|
||||
key={anomalySeries.bounderies.title}
|
||||
id={anomalySeries.bounderies.title}
|
||||
xScaleType={ScaleType.Time}
|
||||
yScaleType={ScaleType.Linear}
|
||||
xAccessor="x"
|
||||
yAccessors={['y']}
|
||||
y0Accessors={['y0']}
|
||||
data={anomalySeries.bounderies.data}
|
||||
color={anomalySeries.bounderies.color}
|
||||
curve={CurveType.CURVE_MONOTONE_X}
|
||||
hideInLegend
|
||||
filterSeriesInTooltip={() => false}
|
||||
/>
|
||||
)}
|
||||
|
||||
{anomalySeries?.scores && (
|
||||
{anomalyTimeseries?.scores && (
|
||||
<RectAnnotation
|
||||
key={anomalySeries.scores.title}
|
||||
key={anomalyTimeseries.scores.title}
|
||||
id="score_anomalies"
|
||||
dataValues={(anomalySeries.scores.data as RectCoordinate[]).map(
|
||||
dataValues={(anomalyTimeseries.scores.data as RectCoordinate[]).map(
|
||||
({ x0, x: x1 }) => ({
|
||||
coordinates: { x0, x1 },
|
||||
})
|
||||
)}
|
||||
style={{ fill: anomalySeries.scores.color }}
|
||||
style={{ fill: anomalyTimeseries.scores.color }}
|
||||
/>
|
||||
)}
|
||||
</Chart>
|
||||
|
|
|
|||
|
|
@ -4,18 +4,22 @@
|
|||
* you may not use this file except in compliance with the Elastic License.
|
||||
*/
|
||||
|
||||
import { Coordinate, TimeSeries } from '../../../../../typings/timeseries';
|
||||
import { isFiniteNumber } from '../../../../../common/utils/is_finite_number';
|
||||
import { APMChartSpec, Coordinate } from '../../../../../typings/timeseries';
|
||||
import { TimeFormatter } from '../../../../../common/utils/formatters';
|
||||
|
||||
export function getResponseTimeTickFormatter(formatter: TimeFormatter) {
|
||||
return (t: number) => formatter(t).formatted;
|
||||
}
|
||||
|
||||
export function getMaxY(timeSeries?: Array<TimeSeries<Coordinate>>) {
|
||||
if (timeSeries) {
|
||||
const coordinates = timeSeries.flatMap((serie) => serie.data);
|
||||
const numbers = coordinates.map((c) => (c.y ? c.y : 0));
|
||||
return Math.max(...numbers, 0);
|
||||
export function getMaxY(specs?: Array<APMChartSpec<Coordinate>>) {
|
||||
const values = specs
|
||||
?.flatMap((spec) => spec.data)
|
||||
.map((coord) => coord.y)
|
||||
.filter(isFiniteNumber);
|
||||
|
||||
if (values?.length) {
|
||||
return Math.max(...values, 0);
|
||||
}
|
||||
return 0;
|
||||
}
|
||||
|
|
|
|||
|
|
@ -25,7 +25,7 @@ const latencyChartData = {
|
|||
latencyTimeseries: [{ x: 1, y: 10 }],
|
||||
anomalyTimeseries: {
|
||||
jobId: '1',
|
||||
anomalyBoundaries: [{ x: 1, y: 2 }],
|
||||
anomalyBoundaries: [{ x: 1, y: 2, y0: 1 }],
|
||||
anomalyScore: [{ x: 1, x0: 2 }],
|
||||
},
|
||||
} as LatencyChartsResponse;
|
||||
|
|
@ -110,30 +110,76 @@ describe('getLatencyChartSelector', () => {
|
|||
latencyAggregationType: LatencyAggregationType.p99,
|
||||
});
|
||||
expect(latencyTimeseries).toEqual({
|
||||
anomalyTimeseries: {
|
||||
boundaries: [
|
||||
{
|
||||
color: 'rgba(0,0,0,0)',
|
||||
areaSeriesStyle: {
|
||||
point: {
|
||||
opacity: 0,
|
||||
},
|
||||
},
|
||||
data: [
|
||||
{
|
||||
x: 1,
|
||||
y: 1,
|
||||
},
|
||||
],
|
||||
fit: 'lookahead',
|
||||
hideLegend: true,
|
||||
hideTooltipValue: true,
|
||||
stackAccessors: ['y'],
|
||||
title: 'anomalyBoundariesLower',
|
||||
type: 'area',
|
||||
},
|
||||
{
|
||||
color: 'rgba(0,0,255,0.5)',
|
||||
areaSeriesStyle: {
|
||||
point: {
|
||||
opacity: 0,
|
||||
},
|
||||
},
|
||||
data: [
|
||||
{
|
||||
x: 1,
|
||||
y: 1,
|
||||
},
|
||||
],
|
||||
fit: 'lookahead',
|
||||
hideLegend: true,
|
||||
hideTooltipValue: true,
|
||||
stackAccessors: ['y'],
|
||||
title: 'anomalyBoundariesUpper',
|
||||
type: 'area',
|
||||
},
|
||||
],
|
||||
scores: {
|
||||
color: 'yellow',
|
||||
data: [
|
||||
{
|
||||
x: 1,
|
||||
x0: 2,
|
||||
},
|
||||
],
|
||||
title: 'anomalyScores',
|
||||
type: 'rectAnnotation',
|
||||
},
|
||||
},
|
||||
latencyTimeseries: [
|
||||
{
|
||||
color: 'black',
|
||||
data: [
|
||||
{
|
||||
x: 1,
|
||||
y: 10,
|
||||
},
|
||||
],
|
||||
title: '99th percentile',
|
||||
titleShort: '99th',
|
||||
data: [{ x: 1, y: 10 }],
|
||||
type: 'linemark',
|
||||
color: 'black',
|
||||
},
|
||||
],
|
||||
mlJobId: '1',
|
||||
anomalyTimeseries: {
|
||||
bounderies: {
|
||||
title: 'Anomaly Boundaries',
|
||||
data: [{ x: 1, y: 2 }],
|
||||
type: 'area',
|
||||
color: 'rgba(0,0,255,0.5)',
|
||||
},
|
||||
scores: {
|
||||
title: 'Anomaly score',
|
||||
data: [{ x: 1, x0: 2 }],
|
||||
type: 'rectAnnotation',
|
||||
color: 'yellow',
|
||||
},
|
||||
},
|
||||
});
|
||||
});
|
||||
});
|
||||
|
|
|
|||
|
|
@ -3,26 +3,20 @@
|
|||
* or more contributor license agreements. Licensed under the Elastic License;
|
||||
* you may not use this file except in compliance with the Elastic License.
|
||||
*/
|
||||
import { Fit } from '@elastic/charts';
|
||||
import { i18n } from '@kbn/i18n';
|
||||
import { rgba } from 'polished';
|
||||
import { EuiTheme } from '../../../observability/public';
|
||||
import { asDuration } from '../../common/utils/formatters';
|
||||
import {
|
||||
Coordinate,
|
||||
RectCoordinate,
|
||||
TimeSeries,
|
||||
} from '../../typings/timeseries';
|
||||
import { APMChartSpec, Coordinate } from '../../typings/timeseries';
|
||||
import { APIReturnType } from '../services/rest/createCallApmApi';
|
||||
|
||||
export type LatencyChartsResponse = APIReturnType<'GET /api/apm/services/{serviceName}/transactions/charts/latency'>;
|
||||
|
||||
interface LatencyChart {
|
||||
latencyTimeseries: Array<TimeSeries<Coordinate>>;
|
||||
interface LatencyChartData {
|
||||
latencyTimeseries: Array<APMChartSpec<Coordinate>>;
|
||||
mlJobId?: string;
|
||||
anomalyTimeseries?: {
|
||||
bounderies: TimeSeries;
|
||||
scores: TimeSeries;
|
||||
};
|
||||
anomalyTimeseries?: { boundaries: APMChartSpec[]; scores: APMChartSpec };
|
||||
}
|
||||
|
||||
export function getLatencyChartSelector({
|
||||
|
|
@ -33,7 +27,7 @@ export function getLatencyChartSelector({
|
|||
latencyChart?: LatencyChartsResponse;
|
||||
theme: EuiTheme;
|
||||
latencyAggregationType?: string;
|
||||
}): LatencyChart {
|
||||
}): LatencyChartData {
|
||||
if (!latencyChart?.latencyTimeseries || !latencyAggregationType) {
|
||||
return {
|
||||
latencyTimeseries: [],
|
||||
|
|
@ -48,7 +42,7 @@ export function getLatencyChartSelector({
|
|||
latencyAggregationType,
|
||||
}),
|
||||
mlJobId: latencyChart.anomalyTimeseries?.jobId,
|
||||
anomalyTimeseries: getAnnomalyTimeseries({
|
||||
anomalyTimeseries: getAnomalyTimeseries({
|
||||
anomalyTimeseries: latencyChart.anomalyTimeseries,
|
||||
theme,
|
||||
}),
|
||||
|
|
@ -114,45 +108,57 @@ function getLatencyTimeseries({
|
|||
return [];
|
||||
}
|
||||
|
||||
function getAnnomalyTimeseries({
|
||||
function getAnomalyTimeseries({
|
||||
anomalyTimeseries,
|
||||
theme,
|
||||
}: {
|
||||
anomalyTimeseries: LatencyChartsResponse['anomalyTimeseries'];
|
||||
theme: EuiTheme;
|
||||
}) {
|
||||
if (anomalyTimeseries) {
|
||||
return {
|
||||
bounderies: getAnomalyBoundariesSeries(
|
||||
anomalyTimeseries.anomalyBoundaries,
|
||||
theme
|
||||
),
|
||||
scores: getAnomalyScoreSeries(anomalyTimeseries.anomalyScore, theme),
|
||||
};
|
||||
}): { boundaries: APMChartSpec[]; scores: APMChartSpec } | undefined {
|
||||
if (!anomalyTimeseries) {
|
||||
return undefined;
|
||||
}
|
||||
}
|
||||
|
||||
export function getAnomalyScoreSeries(data: RectCoordinate[], theme: EuiTheme) {
|
||||
return {
|
||||
title: i18n.translate('xpack.apm.transactions.chart.anomalyScoreLabel', {
|
||||
defaultMessage: 'Anomaly score',
|
||||
}),
|
||||
data,
|
||||
const boundariesConfigBase = {
|
||||
type: 'area',
|
||||
fit: Fit.Lookahead,
|
||||
hideLegend: true,
|
||||
hideTooltipValue: true,
|
||||
stackAccessors: ['y'],
|
||||
areaSeriesStyle: {
|
||||
point: {
|
||||
opacity: 0,
|
||||
},
|
||||
},
|
||||
};
|
||||
|
||||
const boundaries = [
|
||||
{
|
||||
...boundariesConfigBase,
|
||||
title: 'anomalyBoundariesLower',
|
||||
data: anomalyTimeseries.anomalyBoundaries.map((coord) => ({
|
||||
x: coord.x,
|
||||
y: coord.y0,
|
||||
})),
|
||||
color: rgba(0, 0, 0, 0),
|
||||
},
|
||||
{
|
||||
...boundariesConfigBase,
|
||||
title: 'anomalyBoundariesUpper',
|
||||
data: anomalyTimeseries.anomalyBoundaries.map((coord) => ({
|
||||
x: coord.x,
|
||||
y: coord.y - coord.y0,
|
||||
})),
|
||||
color: rgba(theme.eui.euiColorVis1, 0.5),
|
||||
},
|
||||
];
|
||||
|
||||
const scores = {
|
||||
title: 'anomalyScores',
|
||||
type: 'rectAnnotation',
|
||||
data: anomalyTimeseries.anomalyScore,
|
||||
color: theme.eui.euiColorVis9,
|
||||
};
|
||||
}
|
||||
|
||||
function getAnomalyBoundariesSeries(data: Coordinate[], theme: EuiTheme) {
|
||||
return {
|
||||
title: i18n.translate(
|
||||
'xpack.apm.transactions.chart.anomalyBoundariesLabel',
|
||||
{
|
||||
defaultMessage: 'Anomaly Boundaries',
|
||||
}
|
||||
),
|
||||
data,
|
||||
type: 'area',
|
||||
color: rgba(theme.eui.euiColorVis1, 0.5),
|
||||
};
|
||||
return { boundaries, scores };
|
||||
}
|
||||
|
|
|
|||
|
|
@ -4,13 +4,11 @@
|
|||
* you may not use this file except in compliance with the Elastic License.
|
||||
*/
|
||||
import Boom from '@hapi/boom';
|
||||
import { sortBy, uniqBy } from 'lodash';
|
||||
import { ESSearchResponse } from '../../../../../typings/elasticsearch';
|
||||
import { MlPluginSetup } from '../../../../ml/server';
|
||||
import { PromiseReturnType } from '../../../../observability/typings/common';
|
||||
import {
|
||||
getSeverity,
|
||||
ML_ERRORS,
|
||||
ServiceAnomalyStats,
|
||||
} from '../../../common/anomaly_detection';
|
||||
import { getSeverity, ML_ERRORS } from '../../../common/anomaly_detection';
|
||||
import { ENVIRONMENT_ALL } from '../../../common/environment_filter_values';
|
||||
import { getServiceHealthStatus } from '../../../common/service_health_status';
|
||||
import {
|
||||
|
|
@ -20,7 +18,10 @@ import {
|
|||
import { getMlJobsWithAPMGroup } from '../anomaly_detection/get_ml_jobs_with_apm_group';
|
||||
import { Setup, SetupTimeRange } from '../helpers/setup_request';
|
||||
|
||||
export const DEFAULT_ANOMALIES = { mlJobIds: [], serviceAnomalies: {} };
|
||||
export const DEFAULT_ANOMALIES: ServiceAnomaliesResponse = {
|
||||
mlJobIds: [],
|
||||
serviceAnomalies: [],
|
||||
};
|
||||
|
||||
export type ServiceAnomaliesResponse = PromiseReturnType<
|
||||
typeof getServiceAnomalies
|
||||
|
|
@ -39,21 +40,6 @@ export async function getServiceAnomalies({
|
|||
throw Boom.notImplemented(ML_ERRORS.ML_NOT_AVAILABLE);
|
||||
}
|
||||
|
||||
const mlCapabilities = await ml.mlSystem.mlCapabilities();
|
||||
|
||||
if (!mlCapabilities.mlFeatureEnabledInSpace) {
|
||||
throw Boom.forbidden(ML_ERRORS.ML_NOT_AVAILABLE_IN_SPACE);
|
||||
}
|
||||
|
||||
const mlJobIds = await getMLJobIds(ml.anomalyDetectors, environment);
|
||||
|
||||
if (!mlJobIds.length) {
|
||||
return {
|
||||
mlJobIds: [],
|
||||
serviceAnomalies: {},
|
||||
};
|
||||
}
|
||||
|
||||
const params = {
|
||||
body: {
|
||||
size: 0,
|
||||
|
|
@ -61,7 +47,6 @@ export async function getServiceAnomalies({
|
|||
bool: {
|
||||
filter: [
|
||||
{ terms: { result_type: ['model_plot', 'record'] } },
|
||||
{ terms: { job_id: mlJobIds } },
|
||||
{
|
||||
range: {
|
||||
timestamp: {
|
||||
|
|
@ -83,19 +68,25 @@ export async function getServiceAnomalies({
|
|||
},
|
||||
aggs: {
|
||||
services: {
|
||||
terms: { field: 'partition_field_value' },
|
||||
composite: {
|
||||
size: 5000,
|
||||
sources: [
|
||||
{ serviceName: { terms: { field: 'partition_field_value' } } },
|
||||
{ jobId: { terms: { field: 'job_id' } } },
|
||||
],
|
||||
},
|
||||
aggs: {
|
||||
top_score: {
|
||||
top_hits: {
|
||||
sort: { record_score: 'desc' },
|
||||
_source: [
|
||||
'actual',
|
||||
'job_id',
|
||||
'by_field_value',
|
||||
'result_type',
|
||||
'record_score',
|
||||
],
|
||||
size: 1,
|
||||
metrics: {
|
||||
top_metrics: {
|
||||
metrics: [
|
||||
{ field: 'actual' },
|
||||
{ field: 'by_field_value' },
|
||||
{ field: 'result_type' },
|
||||
{ field: 'record_score' },
|
||||
] as const,
|
||||
sort: {
|
||||
record_score: 'desc' as const,
|
||||
},
|
||||
},
|
||||
},
|
||||
},
|
||||
|
|
@ -104,80 +95,54 @@ export async function getServiceAnomalies({
|
|||
},
|
||||
};
|
||||
|
||||
const response = await ml.mlSystem.mlAnomalySearch(params, mlJobIds);
|
||||
const [anomalyResponse, jobIds] = await Promise.all([
|
||||
// pass an empty array of job ids to anomaly search
|
||||
// so any validation is skipped
|
||||
ml.mlSystem.mlAnomalySearch(params, []),
|
||||
getMLJobIds(ml.anomalyDetectors, environment),
|
||||
]);
|
||||
|
||||
const typedAnomalyResponse: ESSearchResponse<
|
||||
unknown,
|
||||
typeof params
|
||||
> = anomalyResponse as any;
|
||||
const relevantBuckets = uniqBy(
|
||||
sortBy(
|
||||
// make sure we only return data for jobs that are available in this space
|
||||
typedAnomalyResponse.aggregations?.services.buckets.filter((bucket) =>
|
||||
jobIds.includes(bucket.key.jobId as string)
|
||||
) ?? [],
|
||||
// sort by job ID in case there are multiple jobs for one service to
|
||||
// ensure consistent results
|
||||
(bucket) => bucket.key.jobId
|
||||
),
|
||||
// return one bucket per service
|
||||
(bucket) => bucket.key.serviceName
|
||||
);
|
||||
|
||||
return {
|
||||
mlJobIds,
|
||||
serviceAnomalies: transformResponseToServiceAnomalies(
|
||||
response as ServiceAnomaliesAggResponse
|
||||
),
|
||||
};
|
||||
}
|
||||
|
||||
interface ServiceAnomaliesAggResponse {
|
||||
aggregations: {
|
||||
services: {
|
||||
buckets: Array<{
|
||||
key: string;
|
||||
top_score: {
|
||||
hits: {
|
||||
hits: Array<{
|
||||
sort: [number];
|
||||
_source: {
|
||||
job_id: string;
|
||||
by_field_value: string;
|
||||
} & (
|
||||
| {
|
||||
record_score: number | null;
|
||||
result_type: 'record';
|
||||
actual: number[];
|
||||
}
|
||||
| {
|
||||
result_type: 'model_plot';
|
||||
actual?: number;
|
||||
}
|
||||
);
|
||||
}>;
|
||||
};
|
||||
};
|
||||
}>;
|
||||
};
|
||||
};
|
||||
}
|
||||
|
||||
function transformResponseToServiceAnomalies(
|
||||
response: ServiceAnomaliesAggResponse
|
||||
) {
|
||||
const serviceAnomaliesMap = (
|
||||
response.aggregations?.services.buckets ?? []
|
||||
).reduce<Record<string, ServiceAnomalyStats>>(
|
||||
(statsByServiceName, { key: serviceName, top_score: topScoreAgg }) => {
|
||||
const mlResult = topScoreAgg.hits.hits[0]._source;
|
||||
mlJobIds: jobIds,
|
||||
serviceAnomalies: relevantBuckets.map((bucket) => {
|
||||
const metrics = bucket.metrics.top[0].metrics;
|
||||
|
||||
const anomalyScore =
|
||||
(mlResult.result_type === 'record' && mlResult.record_score) || 0;
|
||||
metrics.result_type === 'record' && metrics.record_score
|
||||
? (metrics.record_score as number)
|
||||
: 0;
|
||||
|
||||
const severity = getSeverity(anomalyScore);
|
||||
const healthStatus = getServiceHealthStatus({ severity });
|
||||
|
||||
return {
|
||||
...statsByServiceName,
|
||||
[serviceName]: {
|
||||
transactionType: mlResult.by_field_value,
|
||||
jobId: mlResult.job_id,
|
||||
actualValue:
|
||||
mlResult.result_type === 'record'
|
||||
? mlResult.actual[0]
|
||||
: mlResult.actual,
|
||||
anomalyScore,
|
||||
healthStatus,
|
||||
},
|
||||
serviceName: bucket.key.serviceName as string,
|
||||
jobId: bucket.key.jobId as string,
|
||||
transactionType: metrics.by_field_value as string,
|
||||
actualValue: metrics.actual as number | null,
|
||||
anomalyScore,
|
||||
healthStatus,
|
||||
};
|
||||
},
|
||||
{}
|
||||
);
|
||||
|
||||
return serviceAnomaliesMap;
|
||||
}),
|
||||
};
|
||||
}
|
||||
|
||||
export async function getMLJobs(
|
||||
|
|
|
|||
|
|
@ -18,7 +18,6 @@ import { Setup, SetupTimeRange } from '../helpers/setup_request';
|
|||
import {
|
||||
DEFAULT_ANOMALIES,
|
||||
getServiceAnomalies,
|
||||
ServiceAnomaliesResponse,
|
||||
} from './get_service_anomalies';
|
||||
import { getServiceMapFromTraceIds } from './get_service_map_from_trace_ids';
|
||||
import { getTraceSampleIds } from './get_trace_sample_ids';
|
||||
|
|
@ -149,7 +148,7 @@ export type ServiceMapAPIResponse = PromiseReturnType<typeof getServiceMap>;
|
|||
|
||||
export async function getServiceMap(options: IEnvOptions) {
|
||||
const { logger } = options;
|
||||
const anomaliesPromise: Promise<ServiceAnomaliesResponse> = getServiceAnomalies(
|
||||
const anomaliesPromise = getServiceAnomalies(
|
||||
options
|
||||
|
||||
// always catch error to avoid breaking service maps if there is a problem with ML
|
||||
|
|
|
|||
|
|
@ -39,15 +39,16 @@ const javaService = {
|
|||
|
||||
const anomalies = {
|
||||
mlJobIds: ['apm-test-1234-ml-module-name'],
|
||||
serviceAnomalies: {
|
||||
'opbeans-test': {
|
||||
serviceAnomalies: [
|
||||
{
|
||||
serviceName: 'opbeans-test',
|
||||
transactionType: 'request',
|
||||
actualValue: 10000,
|
||||
anomalyScore: 50,
|
||||
jobId: 'apm-test-1234-ml-module-name',
|
||||
healthStatus: ServiceHealthStatus.warning,
|
||||
},
|
||||
},
|
||||
],
|
||||
};
|
||||
|
||||
describe('transformServiceMapResponses', () => {
|
||||
|
|
|
|||
|
|
@ -110,7 +110,9 @@ export function transformServiceMapResponses(response: ServiceMapResponse) {
|
|||
const mergedServiceNode = Object.assign({}, ...matchedServiceNodes);
|
||||
|
||||
const serviceAnomalyStats = serviceName
|
||||
? anomalies.serviceAnomalies[serviceName]
|
||||
? anomalies.serviceAnomalies.find(
|
||||
(item) => item.serviceName === serviceName
|
||||
)
|
||||
: null;
|
||||
|
||||
if (matchedServiceNodes.length) {
|
||||
|
|
|
|||
|
|
@ -6,10 +6,7 @@
|
|||
|
||||
import { getSeverity } from '../../../../common/anomaly_detection';
|
||||
import { getServiceHealthStatus } from '../../../../common/service_health_status';
|
||||
import {
|
||||
getMLJobIds,
|
||||
getServiceAnomalies,
|
||||
} from '../../service_map/get_service_anomalies';
|
||||
import { getServiceAnomalies } from '../../service_map/get_service_anomalies';
|
||||
import {
|
||||
ServicesItemsProjection,
|
||||
ServicesItemsSetup,
|
||||
|
|
@ -29,27 +26,16 @@ export const getHealthStatuses = async (
|
|||
return [];
|
||||
}
|
||||
|
||||
const jobIds = await getMLJobIds(
|
||||
setup.ml.anomalyDetectors,
|
||||
mlAnomaliesEnvironment
|
||||
);
|
||||
if (!jobIds.length) {
|
||||
return [];
|
||||
}
|
||||
|
||||
const anomalies = await getServiceAnomalies({
|
||||
setup,
|
||||
environment: mlAnomaliesEnvironment,
|
||||
});
|
||||
|
||||
return Object.keys(anomalies.serviceAnomalies).map((serviceName) => {
|
||||
const stats = anomalies.serviceAnomalies[serviceName];
|
||||
|
||||
const severity = getSeverity(stats.anomalyScore);
|
||||
return anomalies.serviceAnomalies.map((anomalyStats) => {
|
||||
const severity = getSeverity(anomalyStats.anomalyScore);
|
||||
const healthStatus = getServiceHealthStatus({ severity });
|
||||
|
||||
return {
|
||||
serviceName,
|
||||
serviceName: anomalyStats.serviceName,
|
||||
healthStatus,
|
||||
};
|
||||
});
|
||||
|
|
|
|||
|
|
@ -4,10 +4,9 @@
|
|||
* you may not use this file except in compliance with the Elastic License.
|
||||
*/
|
||||
|
||||
import { Logger } from 'kibana/server';
|
||||
import { ESSearchResponse } from '../../../../../../typings/elasticsearch';
|
||||
import { PromiseReturnType } from '../../../../../observability/typings/common';
|
||||
import { Setup, SetupTimeRange } from '../../helpers/setup_request';
|
||||
import { Setup } from '../../helpers/setup_request';
|
||||
|
||||
export type ESResponse = Exclude<
|
||||
PromiseReturnType<typeof anomalySeriesFetcher>,
|
||||
|
|
@ -18,81 +17,74 @@ export async function anomalySeriesFetcher({
|
|||
serviceName,
|
||||
transactionType,
|
||||
intervalString,
|
||||
mlBucketSize,
|
||||
setup,
|
||||
jobId,
|
||||
logger,
|
||||
ml,
|
||||
start,
|
||||
end,
|
||||
}: {
|
||||
serviceName: string;
|
||||
transactionType: string;
|
||||
intervalString: string;
|
||||
mlBucketSize: number;
|
||||
setup: Setup & SetupTimeRange;
|
||||
jobId: string;
|
||||
logger: Logger;
|
||||
ml: Required<Setup>['ml'];
|
||||
start: number;
|
||||
end: number;
|
||||
}) {
|
||||
const { ml, start, end } = setup;
|
||||
if (!ml) {
|
||||
return;
|
||||
}
|
||||
|
||||
// move the start back with one bucket size, to ensure to get anomaly data in the beginning
|
||||
// this is required because ML has a minimum bucket size (default is 900s) so if our buckets are smaller, we might have several null buckets in the beginning
|
||||
const newStart = start - mlBucketSize * 1000;
|
||||
|
||||
const params = {
|
||||
body: {
|
||||
size: 0,
|
||||
query: {
|
||||
bool: {
|
||||
filter: [
|
||||
{ term: { job_id: jobId } },
|
||||
{ exists: { field: 'bucket_span' } },
|
||||
{ terms: { result_type: ['model_plot', 'record'] } },
|
||||
{ term: { partition_field_value: serviceName } },
|
||||
{ term: { by_field_value: transactionType } },
|
||||
{
|
||||
range: {
|
||||
timestamp: { gte: newStart, lte: end, format: 'epoch_millis' },
|
||||
timestamp: {
|
||||
gte: start,
|
||||
lte: end,
|
||||
format: 'epoch_millis',
|
||||
},
|
||||
},
|
||||
},
|
||||
],
|
||||
},
|
||||
},
|
||||
aggs: {
|
||||
ml_avg_response_times: {
|
||||
date_histogram: {
|
||||
field: 'timestamp',
|
||||
fixed_interval: intervalString,
|
||||
min_doc_count: 0,
|
||||
extended_bounds: { min: newStart, max: end },
|
||||
job_id: {
|
||||
terms: {
|
||||
field: 'job_id',
|
||||
},
|
||||
aggs: {
|
||||
anomaly_score: { max: { field: 'record_score' } },
|
||||
lower: { min: { field: 'model_lower' } },
|
||||
upper: { max: { field: 'model_upper' } },
|
||||
ml_avg_response_times: {
|
||||
date_histogram: {
|
||||
field: 'timestamp',
|
||||
fixed_interval: intervalString,
|
||||
extended_bounds: { min: start, max: end },
|
||||
},
|
||||
aggs: {
|
||||
anomaly_score: {
|
||||
top_metrics: {
|
||||
metrics: [
|
||||
{ field: 'record_score' },
|
||||
{ field: 'timestamp' },
|
||||
{ field: 'bucket_span' },
|
||||
] as const,
|
||||
sort: {
|
||||
record_score: 'desc' as const,
|
||||
},
|
||||
},
|
||||
},
|
||||
lower: { min: { field: 'model_lower' } },
|
||||
upper: { max: { field: 'model_upper' } },
|
||||
},
|
||||
},
|
||||
},
|
||||
},
|
||||
},
|
||||
},
|
||||
};
|
||||
|
||||
try {
|
||||
const response: ESSearchResponse<
|
||||
unknown,
|
||||
typeof params
|
||||
> = (await ml.mlSystem.mlAnomalySearch(params, [jobId])) as any;
|
||||
|
||||
return response;
|
||||
} catch (err) {
|
||||
const isHttpError = 'statusCode' in err;
|
||||
if (isHttpError) {
|
||||
logger.info(
|
||||
`Status code "${err.statusCode}" while retrieving ML anomalies for APM`
|
||||
);
|
||||
return;
|
||||
}
|
||||
logger.error('An error occurred while retrieving ML anomalies for APM');
|
||||
logger.error(err);
|
||||
}
|
||||
return (ml.mlSystem.mlAnomalySearch(params, []) as unknown) as Promise<
|
||||
ESSearchResponse<unknown, typeof params>
|
||||
>;
|
||||
}
|
||||
|
|
|
|||
|
|
@ -1,61 +0,0 @@
|
|||
/*
|
||||
* Copyright Elasticsearch B.V. and/or licensed to Elasticsearch B.V. under one
|
||||
* or more contributor license agreements. Licensed under the Elastic License;
|
||||
* you may not use this file except in compliance with the Elastic License.
|
||||
*/
|
||||
|
||||
import { Logger } from 'kibana/server';
|
||||
import { Setup, SetupTimeRange } from '../../helpers/setup_request';
|
||||
|
||||
interface IOptions {
|
||||
setup: Setup & SetupTimeRange;
|
||||
jobId: string;
|
||||
logger: Logger;
|
||||
}
|
||||
|
||||
interface ESResponse {
|
||||
bucket_span: number;
|
||||
}
|
||||
|
||||
export async function getMlBucketSize({
|
||||
setup,
|
||||
jobId,
|
||||
logger,
|
||||
}: IOptions): Promise<number | undefined> {
|
||||
const { ml, start, end } = setup;
|
||||
if (!ml) {
|
||||
return;
|
||||
}
|
||||
|
||||
const params = {
|
||||
body: {
|
||||
_source: 'bucket_span',
|
||||
size: 1,
|
||||
terminate_after: 1,
|
||||
query: {
|
||||
bool: {
|
||||
filter: [
|
||||
{ term: { job_id: jobId } },
|
||||
{ exists: { field: 'bucket_span' } },
|
||||
{
|
||||
range: {
|
||||
timestamp: { gte: start, lte: end, format: 'epoch_millis' },
|
||||
},
|
||||
},
|
||||
],
|
||||
},
|
||||
},
|
||||
},
|
||||
};
|
||||
|
||||
try {
|
||||
const resp = await ml.mlSystem.mlAnomalySearch<ESResponse>(params, [jobId]);
|
||||
return resp.hits.hits[0]?._source.bucket_span;
|
||||
} catch (err) {
|
||||
const isHttpError = 'statusCode' in err;
|
||||
if (isHttpError) {
|
||||
return;
|
||||
}
|
||||
logger.error(err);
|
||||
}
|
||||
}
|
||||
|
|
@ -3,53 +3,47 @@
|
|||
* or more contributor license agreements. Licensed under the Elastic License;
|
||||
* you may not use this file except in compliance with the Elastic License.
|
||||
*/
|
||||
import { Logger } from 'kibana/server';
|
||||
import { isNumber } from 'lodash';
|
||||
import { compact } from 'lodash';
|
||||
import { Logger } from 'src/core/server';
|
||||
import { isFiniteNumber } from '../../../../common/utils/is_finite_number';
|
||||
import { maybe } from '../../../../common/utils/maybe';
|
||||
import { ENVIRONMENT_ALL } from '../../../../common/environment_filter_values';
|
||||
import { getBucketSize } from '../../helpers/get_bucket_size';
|
||||
import { Setup, SetupTimeRange } from '../../helpers/setup_request';
|
||||
import { anomalySeriesFetcher } from './fetcher';
|
||||
import { getMlBucketSize } from './get_ml_bucket_size';
|
||||
import { anomalySeriesTransform } from './transform';
|
||||
import { getMLJobIds } from '../../service_map/get_service_anomalies';
|
||||
import { getLatencyTimeseries } from '../get_latency_charts';
|
||||
import { PromiseReturnType } from '../../../../../observability/typings/common';
|
||||
import { ANOMALY_THRESHOLD } from '../../../../../ml/common';
|
||||
|
||||
export async function getAnomalySeries({
|
||||
serviceName,
|
||||
transactionType,
|
||||
transactionName,
|
||||
latencyTimeseries,
|
||||
setup,
|
||||
logger,
|
||||
}: {
|
||||
serviceName: string;
|
||||
transactionType: string | undefined;
|
||||
transactionName: string | undefined;
|
||||
latencyTimeseries: PromiseReturnType<
|
||||
typeof getLatencyTimeseries
|
||||
>['latencyTimeseries'];
|
||||
transactionType: string;
|
||||
transactionName?: string;
|
||||
setup: Setup & SetupTimeRange;
|
||||
logger: Logger;
|
||||
}) {
|
||||
const timeseriesDates = latencyTimeseries?.map(({ x }) => x);
|
||||
const { uiFilters, start, end, ml } = setup;
|
||||
const { environment } = uiFilters;
|
||||
|
||||
/*
|
||||
* don't fetch:
|
||||
* - anomalies for transaction details page
|
||||
* - anomalies without a type
|
||||
* - timeseries is empty
|
||||
*/
|
||||
if (transactionName || !transactionType || !timeseriesDates?.length) {
|
||||
return;
|
||||
// don't fetch anomalies if the ML plugin is not setup
|
||||
if (!ml) {
|
||||
return undefined;
|
||||
}
|
||||
|
||||
const { uiFilters, start, end } = setup;
|
||||
const { environment } = uiFilters;
|
||||
// don't fetch anomalies if requested for a specific transaction name
|
||||
// as ML results are not partitioned by transaction name
|
||||
if (!!transactionName) {
|
||||
return undefined;
|
||||
}
|
||||
|
||||
// don't fetch anomalies when no specific environment is selected
|
||||
if (environment === ENVIRONMENT_ALL.value) {
|
||||
return;
|
||||
return undefined;
|
||||
}
|
||||
|
||||
// don't fetch anomalies if unknown uiFilters are applied
|
||||
|
|
@ -60,48 +54,77 @@ export async function getAnomalySeries({
|
|||
.some((uiFilterName) => !knownFilters.includes(uiFilterName));
|
||||
|
||||
if (hasUnknownFiltersApplied) {
|
||||
return;
|
||||
return undefined;
|
||||
}
|
||||
|
||||
// don't fetch anomalies if the ML plugin is not setup
|
||||
if (!setup.ml) {
|
||||
return;
|
||||
}
|
||||
const { intervalString } = getBucketSize({ start, end });
|
||||
|
||||
// don't fetch anomalies if required license is not satisfied
|
||||
const mlCapabilities = await setup.ml.mlSystem.mlCapabilities();
|
||||
if (!mlCapabilities.isPlatinumOrTrialLicense) {
|
||||
return;
|
||||
}
|
||||
// move the start back with one bucket size, to ensure to get anomaly data in the beginning
|
||||
// this is required because ML has a minimum bucket size (default is 900s) so if our buckets
|
||||
// are smaller, we might have several null buckets in the beginning
|
||||
const mlStart = start - 900 * 1000;
|
||||
|
||||
const mlJobIds = await getMLJobIds(setup.ml.anomalyDetectors, environment);
|
||||
const [anomaliesResponse, jobIds] = await Promise.all([
|
||||
anomalySeriesFetcher({
|
||||
serviceName,
|
||||
transactionType,
|
||||
intervalString,
|
||||
ml,
|
||||
start: mlStart,
|
||||
end,
|
||||
}),
|
||||
getMLJobIds(ml.anomalyDetectors, environment),
|
||||
]);
|
||||
|
||||
const jobId = mlJobIds[0];
|
||||
const scoreSeriesCollection = anomaliesResponse?.aggregations?.job_id.buckets
|
||||
.filter((bucket) => jobIds.includes(bucket.key as string))
|
||||
.map((bucket) => {
|
||||
const dateBuckets = bucket.ml_avg_response_times.buckets;
|
||||
|
||||
const mlBucketSize = await getMlBucketSize({ setup, jobId, logger });
|
||||
if (!isNumber(mlBucketSize)) {
|
||||
return;
|
||||
}
|
||||
return {
|
||||
jobId: bucket.key as string,
|
||||
anomalyScore: compact(
|
||||
dateBuckets.map((dateBucket) => {
|
||||
const metrics = maybe(dateBucket.anomaly_score.top[0])?.metrics;
|
||||
const score = metrics?.record_score;
|
||||
|
||||
const { intervalString, bucketSize } = getBucketSize({ start, end });
|
||||
if (
|
||||
!metrics ||
|
||||
!isFiniteNumber(score) ||
|
||||
score < ANOMALY_THRESHOLD.CRITICAL
|
||||
) {
|
||||
return null;
|
||||
}
|
||||
|
||||
const esResponse = await anomalySeriesFetcher({
|
||||
serviceName,
|
||||
transactionType,
|
||||
intervalString,
|
||||
mlBucketSize,
|
||||
setup,
|
||||
jobId,
|
||||
logger,
|
||||
});
|
||||
const anomalyStart = Date.parse(metrics.timestamp as string);
|
||||
const anomalyEnd =
|
||||
anomalyStart + (metrics.bucket_span as number) * 1000;
|
||||
|
||||
if (esResponse && mlBucketSize > 0) {
|
||||
return anomalySeriesTransform(
|
||||
esResponse,
|
||||
mlBucketSize,
|
||||
bucketSize,
|
||||
timeseriesDates,
|
||||
jobId
|
||||
return {
|
||||
x0: anomalyStart,
|
||||
x: anomalyEnd,
|
||||
y: score,
|
||||
};
|
||||
})
|
||||
),
|
||||
anomalyBoundaries: dateBuckets
|
||||
.filter(
|
||||
(dateBucket) =>
|
||||
dateBucket.lower.value !== null && dateBucket.upper.value !== null
|
||||
)
|
||||
.map((dateBucket) => ({
|
||||
x: dateBucket.key,
|
||||
y0: dateBucket.lower.value as number,
|
||||
y: dateBucket.upper.value as number,
|
||||
})),
|
||||
};
|
||||
});
|
||||
|
||||
if ((scoreSeriesCollection?.length ?? 0) > 1) {
|
||||
logger.warn(
|
||||
`More than one ML job was found for ${serviceName} for environment ${environment}. Only showing results from ${scoreSeriesCollection?.[0].jobId}`
|
||||
);
|
||||
}
|
||||
|
||||
return scoreSeriesCollection?.[0];
|
||||
}
|
||||
|
|
|
|||
|
|
@ -1,134 +0,0 @@
|
|||
/*
|
||||
* Copyright Elasticsearch B.V. and/or licensed to Elasticsearch B.V. under one
|
||||
* or more contributor license agreements. Licensed under the Elastic License;
|
||||
* you may not use this file except in compliance with the Elastic License.
|
||||
*/
|
||||
|
||||
import { first, last } from 'lodash';
|
||||
import { Coordinate, RectCoordinate } from '../../../../typings/timeseries';
|
||||
import { ESResponse } from './fetcher';
|
||||
|
||||
type IBucket = ReturnType<typeof getBucket>;
|
||||
function getBucket(
|
||||
bucket: Required<ESResponse>['aggregations']['ml_avg_response_times']['buckets'][0]
|
||||
) {
|
||||
return {
|
||||
x: bucket.key,
|
||||
anomalyScore: bucket.anomaly_score.value,
|
||||
lower: bucket.lower.value,
|
||||
upper: bucket.upper.value,
|
||||
};
|
||||
}
|
||||
|
||||
export type AnomalyTimeSeriesResponse = ReturnType<
|
||||
typeof anomalySeriesTransform
|
||||
>;
|
||||
export function anomalySeriesTransform(
|
||||
response: ESResponse,
|
||||
mlBucketSize: number,
|
||||
bucketSize: number,
|
||||
timeSeriesDates: number[],
|
||||
jobId: string
|
||||
) {
|
||||
const buckets =
|
||||
response.aggregations?.ml_avg_response_times.buckets.map(getBucket) || [];
|
||||
|
||||
const bucketSizeInMillis = Math.max(bucketSize, mlBucketSize) * 1000;
|
||||
|
||||
return {
|
||||
jobId,
|
||||
anomalyScore: getAnomalyScoreDataPoints(
|
||||
buckets,
|
||||
timeSeriesDates,
|
||||
bucketSizeInMillis
|
||||
),
|
||||
anomalyBoundaries: getAnomalyBoundaryDataPoints(buckets, timeSeriesDates),
|
||||
};
|
||||
}
|
||||
|
||||
export function getAnomalyScoreDataPoints(
|
||||
buckets: IBucket[],
|
||||
timeSeriesDates: number[],
|
||||
bucketSizeInMillis: number
|
||||
): RectCoordinate[] {
|
||||
const ANOMALY_THRESHOLD = 75;
|
||||
const firstDate = first(timeSeriesDates);
|
||||
const lastDate = last(timeSeriesDates);
|
||||
|
||||
if (firstDate === undefined || lastDate === undefined) {
|
||||
return [];
|
||||
}
|
||||
|
||||
return buckets
|
||||
.filter(
|
||||
(bucket) =>
|
||||
bucket.anomalyScore !== null && bucket.anomalyScore > ANOMALY_THRESHOLD
|
||||
)
|
||||
.filter(isInDateRange(firstDate, lastDate))
|
||||
.map((bucket) => {
|
||||
return {
|
||||
x0: bucket.x,
|
||||
x: Math.min(bucket.x + bucketSizeInMillis, lastDate), // don't go beyond last date
|
||||
};
|
||||
});
|
||||
}
|
||||
|
||||
export function getAnomalyBoundaryDataPoints(
|
||||
buckets: IBucket[],
|
||||
timeSeriesDates: number[]
|
||||
): Coordinate[] {
|
||||
return replaceFirstAndLastBucket(buckets, timeSeriesDates)
|
||||
.filter((bucket) => bucket.lower !== null)
|
||||
.map((bucket) => {
|
||||
return {
|
||||
x: bucket.x,
|
||||
y0: bucket.lower,
|
||||
y: bucket.upper,
|
||||
};
|
||||
});
|
||||
}
|
||||
|
||||
export function replaceFirstAndLastBucket(
|
||||
buckets: IBucket[],
|
||||
timeSeriesDates: number[]
|
||||
) {
|
||||
const firstDate = first(timeSeriesDates);
|
||||
const lastDate = last(timeSeriesDates);
|
||||
|
||||
if (firstDate === undefined || lastDate === undefined) {
|
||||
return buckets;
|
||||
}
|
||||
|
||||
const preBucketWithValue = buckets
|
||||
.filter((p) => p.x <= firstDate)
|
||||
.reverse()
|
||||
.find((p) => p.lower !== null);
|
||||
|
||||
const bucketsInRange = buckets.filter(isInDateRange(firstDate, lastDate));
|
||||
|
||||
// replace first bucket if it is null
|
||||
const firstBucket = first(bucketsInRange);
|
||||
if (preBucketWithValue && firstBucket && firstBucket.lower === null) {
|
||||
firstBucket.lower = preBucketWithValue.lower;
|
||||
firstBucket.upper = preBucketWithValue.upper;
|
||||
}
|
||||
|
||||
const lastBucketWithValue = [...buckets]
|
||||
.reverse()
|
||||
.find((p) => p.lower !== null);
|
||||
|
||||
// replace last bucket if it is null
|
||||
const lastBucket = last(bucketsInRange);
|
||||
if (lastBucketWithValue && lastBucket && lastBucket.lower === null) {
|
||||
lastBucket.lower = lastBucketWithValue.lower;
|
||||
lastBucket.upper = lastBucketWithValue.upper;
|
||||
}
|
||||
|
||||
return bucketsInRange;
|
||||
}
|
||||
|
||||
// anomaly time series contain one or more buckets extra in the beginning
|
||||
// these extra buckets should be removed
|
||||
function isInDateRange(firstDate: number, lastDate: number) {
|
||||
return (p: IBucket) => p.x >= firstDate && p.x <= lastDate;
|
||||
}
|
||||
|
|
@ -169,23 +169,28 @@ export const transactionLatencyChatsRoute = createRoute({
|
|||
transactionName,
|
||||
setup,
|
||||
searchAggregatedTransactions,
|
||||
logger,
|
||||
};
|
||||
|
||||
const {
|
||||
const [latencyData, anomalyTimeseries] = await Promise.all([
|
||||
getLatencyTimeseries({
|
||||
...options,
|
||||
latencyAggregationType: latencyAggregationType as LatencyAggregationType,
|
||||
}),
|
||||
getAnomalySeries(options).catch((error) => {
|
||||
logger.warn(`Unable to retrieve anomalies for latency charts.`);
|
||||
logger.error(error);
|
||||
return undefined;
|
||||
}),
|
||||
]);
|
||||
|
||||
const { latencyTimeseries, overallAvgDuration } = latencyData;
|
||||
|
||||
return {
|
||||
latencyTimeseries,
|
||||
overallAvgDuration,
|
||||
} = await getLatencyTimeseries({
|
||||
...options,
|
||||
latencyAggregationType: latencyAggregationType as LatencyAggregationType,
|
||||
});
|
||||
|
||||
const anomalyTimeseries = await getAnomalySeries({
|
||||
...options,
|
||||
logger,
|
||||
latencyTimeseries,
|
||||
});
|
||||
|
||||
return { latencyTimeseries, overallAvgDuration, anomalyTimeseries };
|
||||
anomalyTimeseries,
|
||||
};
|
||||
},
|
||||
});
|
||||
|
||||
|
|
|
|||
|
|
@ -3,6 +3,14 @@
|
|||
* or more contributor license agreements. Licensed under the Elastic License;
|
||||
* you may not use this file except in compliance with the Elastic License.
|
||||
*/
|
||||
import {
|
||||
AccessorFn,
|
||||
AreaSeriesStyle,
|
||||
Fit,
|
||||
FitConfig,
|
||||
LineSeriesStyle,
|
||||
} from '@elastic/charts';
|
||||
import { DeepPartial } from 'utility-types';
|
||||
import { Maybe } from '../typings/common';
|
||||
|
||||
export interface Coordinate {
|
||||
|
|
@ -15,7 +23,13 @@ export interface RectCoordinate {
|
|||
x0: number;
|
||||
}
|
||||
|
||||
export interface TimeSeries<
|
||||
type Accessor = Array<string | number | AccessorFn>;
|
||||
|
||||
export type TimeSeries<
|
||||
TCoordinate extends { x: number } = Coordinate | RectCoordinate
|
||||
> = APMChartSpec<TCoordinate>;
|
||||
|
||||
export interface APMChartSpec<
|
||||
TCoordinate extends { x: number } = Coordinate | RectCoordinate
|
||||
> {
|
||||
title: string;
|
||||
|
|
@ -27,6 +41,11 @@ export interface TimeSeries<
|
|||
type: string;
|
||||
color: string;
|
||||
areaColor?: string;
|
||||
fit?: Exclude<Fit, 'explicit'> | FitConfig;
|
||||
stackAccessors?: Accessor;
|
||||
splitSeriesAccessors?: Accessor;
|
||||
lineSeriesStyle?: DeepPartial<LineSeriesStyle>;
|
||||
areaSeriesStyle?: DeepPartial<AreaSeriesStyle>;
|
||||
}
|
||||
|
||||
export type ChartType = 'area' | 'linemark';
|
||||
|
|
|
|||
|
|
@ -5482,8 +5482,6 @@
|
|||
"xpack.apm.transactionOverview.userExperience.calloutTitle": "導入：Elasticユーザーエクスペリエンス",
|
||||
"xpack.apm.transactionOverview.userExperience.linkLabel": "移動",
|
||||
"xpack.apm.transactionRateLabel": "{value} tpm",
|
||||
"xpack.apm.transactions.chart.anomalyBoundariesLabel": "異常境界",
|
||||
"xpack.apm.transactions.chart.anomalyScoreLabel": "異常スコア",
|
||||
"xpack.apm.transactions.latency.chart.95thPercentileLabel": "95 パーセンタイル",
|
||||
"xpack.apm.transactions.latency.chart.99thPercentileLabel": "99 パーセンタイル",
|
||||
"xpack.apm.transactions.latency.chart.averageLabel": "平均",
|
||||
|
|
|
|||
|
|
@ -5492,8 +5492,6 @@
|
|||
"xpack.apm.transactionOverview.userExperience.calloutTitle": "即将引入：Elastic 用户体验",
|
||||
"xpack.apm.transactionOverview.userExperience.linkLabel": "带我前往此处",
|
||||
"xpack.apm.transactionRateLabel": "{value} tpm",
|
||||
"xpack.apm.transactions.chart.anomalyBoundariesLabel": "异常边界",
|
||||
"xpack.apm.transactions.chart.anomalyScoreLabel": "异常分数",
|
||||
"xpack.apm.transactions.latency.chart.95thPercentileLabel": "第 95 个百分位",
|
||||
"xpack.apm.transactions.latency.chart.99thPercentileLabel": "第 99 个百分位",
|
||||
"xpack.apm.transactions.latency.chart.averageLabel": "平均值",
|
||||
|
|
|
|||
File diff suppressed because it is too large
Load diff
|
|
@ -167,9 +167,11 @@ export default function serviceMapsApiTests({ getService }: FtrProviderContext)
|
|||
"id": "opbeans-python",
|
||||
"service.name": "opbeans-python",
|
||||
"serviceAnomalyStats": Object {
|
||||
"actualValue": 24282.2352941176,
|
||||
"anomalyScore": 0,
|
||||
"healthStatus": "healthy",
|
||||
"jobId": "apm-production-1369-high_mean_transaction_duration",
|
||||
"jobId": "apm-environment_not_defined-5626-high_mean_transaction_duration",
|
||||
"serviceName": "opbeans-python",
|
||||
"transactionType": "request",
|
||||
},
|
||||
},
|
||||
|
|
@ -185,6 +187,7 @@ export default function serviceMapsApiTests({ getService }: FtrProviderContext)
|
|||
"anomalyScore": 0,
|
||||
"healthStatus": "healthy",
|
||||
"jobId": "apm-testing-384f-high_mean_transaction_duration",
|
||||
"serviceName": "opbeans-node",
|
||||
"transactionType": "request",
|
||||
},
|
||||
},
|
||||
|
|
@ -200,6 +203,7 @@ export default function serviceMapsApiTests({ getService }: FtrProviderContext)
|
|||
"anomalyScore": 0,
|
||||
"healthStatus": "healthy",
|
||||
"jobId": "apm-testing-384f-high_mean_transaction_duration",
|
||||
"serviceName": "opbeans-rum",
|
||||
"transactionType": "page-load",
|
||||
},
|
||||
},
|
||||
|
|
|
|||
|
|
@ -12,11 +12,6 @@ Array [
|
|||
"y": 0,
|
||||
"y0": 0,
|
||||
},
|
||||
Object {
|
||||
"x": 1607437650000,
|
||||
"y": 0,
|
||||
"y0": 0,
|
||||
},
|
||||
]
|
||||
`;
|
||||
|
||||
|
|
@ -32,11 +27,6 @@ Array [
|
|||
"y": 1660982.24115757,
|
||||
"y0": 5732.00699123528,
|
||||
},
|
||||
Object {
|
||||
"x": 1607437650000,
|
||||
"y": 1660982.24115757,
|
||||
"y0": 5732.00699123528,
|
||||
},
|
||||
]
|
||||
`;
|
||||
|
||||
|
|
@ -52,10 +42,5 @@ Array [
|
|||
"y": 1660982.24115757,
|
||||
"y0": 5732.00699123528,
|
||||
},
|
||||
Object {
|
||||
"x": 1607437650000,
|
||||
"y": 1660982.24115757,
|
||||
"y0": 5732.00699123528,
|
||||
},
|
||||
]
|
||||
`;
|
||||
|
|
|
|||
34
x-pack/typings/elasticsearch/aggregations.d.ts
vendored
34
x-pack/typings/elasticsearch/aggregations.d.ts
vendored
|
|
@ -387,22 +387,24 @@ interface AggregationResponsePart<TAggregationOptionsMap extends AggregationOpti
|
|||
};
|
||||
bucket_sort: undefined;
|
||||
bucket_selector: undefined;
|
||||
top_metrics: [
|
||||
{
|
||||
sort: [string | number];
|
||||
metrics: UnionToIntersection<
|
||||
TAggregationOptionsMap extends {
|
||||
top_metrics: { metrics: { field: infer TFieldName } };
|
||||
}
|
||||
? TopMetricsMap<TFieldName>
|
||||
: TAggregationOptionsMap extends {
|
||||
top_metrics: { metrics: MaybeReadonlyArray<{ field: infer TFieldName }> };
|
||||
}
|
||||
? TopMetricsMap<TFieldName>
|
||||
: TopMetricsMap<string>
|
||||
>;
|
||||
}
|
||||
];
|
||||
top_metrics: {
|
||||
top: [
|
||||
{
|
||||
sort: [string | number];
|
||||
metrics: UnionToIntersection<
|
||||
TAggregationOptionsMap extends {
|
||||
top_metrics: { metrics: { field: infer TFieldName } };
|
||||
}
|
||||
? TopMetricsMap<TFieldName>
|
||||
: TAggregationOptionsMap extends {
|
||||
top_metrics: { metrics: MaybeReadonlyArray<{ field: infer TFieldName }> };
|
||||
}
|
||||
? TopMetricsMap<TFieldName>
|
||||
: TopMetricsMap<string>
|
||||
>;
|
||||
}
|
||||
];
|
||||
};
|
||||
avg_bucket: {
|
||||
value: number | null;
|
||||
};
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue