Skip to content

Redesigining randomized tests for TS functions #129971

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 17 commits into from
Jul 15, 2025
Merged

Redesigining randomized tests for TS functions #129971

Show file tree
Hide file tree
Changes from 6 commits
Commits
Show all changes
17 commits
Select commit Hold shift + click to select a range
699e755
Working on rate integration test
pabloem Jun 24, 2025
8fae5ab
Sketch of new rate tests. Unsure of the source of variation >10pct
pabloem Jun 24, 2025
091490b
Tuning and improving auto-test for rate-based aggregations
pabloem Jun 28, 2025
3f2bd20
Merge branch 'main' into pabloem-rateit
pabloem Jun 29, 2025
697ef59
fixup
pabloem Jun 30, 2025
fd89c97
Merge branch 'main' into pabloem-rateit
pabloem Jun 30, 2025
c457df1
addressing comments
pabloem Jul 7, 2025
b4bd117
fixup
pabloem Jul 7, 2025
c603572
Merge branch 'main' into pabloem-rateit
pabloem Jul 8, 2025
88e6d22
improve failure rate
pabloem Jul 8, 2025
9c62818
tuned deviation limits to reduce flakiness
pabloem Jul 9, 2025
8bbef85
Merge branch 'main' into pabloem-rateit
pabloem Jul 9, 2025
472875d
fixup - wider error margin
pabloem Jul 10, 2025
6537bfa
Merge branch 'main' into pabloem-rateit
pabloem Jul 11, 2025
4e88241
wider margin
pabloem Jul 14, 2025
c26e5db
wider range for aggs
pabloem Jul 14, 2025
132c536
Merge branch 'main' into pabloem-rateit
pabloem Jul 15, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
316 changes: 0 additions & 316 deletions ...n/esql/src/internalClusterTest/java/org/elasticsearch/xpack/esql/action/TimeSeriesIT.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -9,17 +9,14 @@

import org.elasticsearch.Build;
import org.elasticsearch.common.Randomness;
import org.elasticsearch.common.Rounding;
import org.elasticsearch.common.settings.Settings;
import org.elasticsearch.compute.lucene.TimeSeriesSourceOperator;
import org.elasticsearch.compute.operator.DriverProfile;
import org.elasticsearch.compute.operator.TimeSeriesAggregationOperator;
import org.elasticsearch.core.TimeValue;
import org.elasticsearch.xpack.esql.EsqlTestUtils;
import org.elasticsearch.xpack.esql.core.type.DataType;
import org.junit.Before;

import java.time.ZoneOffset;
import java.util.ArrayList;
import java.util.Comparator;
import java.util.HashMap;
Expand Down Expand Up @@ -321,319 +318,6 @@ record RateKey(String cluster, String host) {
}
}

@AwaitsFix(bugUrl = "removed?")
public void testRateWithTimeBucket() {
var rounding = new Rounding.Builder(TimeValue.timeValueSeconds(60)).timeZone(ZoneOffset.UTC).build().prepareForUnknown();
record RateKey(String host, String cluster, long interval) {}
Map<RateKey, List<RequestCounter>> groups = new HashMap<>();
for (Doc doc : docs) {
RateKey key = new RateKey(doc.host, doc.cluster, rounding.round(doc.timestamp));
groups.computeIfAbsent(key, k -> new ArrayList<>()).add(new RequestCounter(doc.timestamp, doc.requestCount));
}
Map<Long, List<Double>> bucketToRates = new HashMap<>();
for (Map.Entry<RateKey, List<RequestCounter>> e : groups.entrySet()) {
List<Double> values = bucketToRates.computeIfAbsent(e.getKey().interval, k -> new ArrayList<>());
Double rate = computeRate(e.getValue());
if (rate != null) {
values.add(rate);
}
}
List<Long> sortedKeys = bucketToRates.keySet().stream().sorted().limit(5).toList();
try (var resp = run("TS hosts | STATS sum(rate(request_count)) BY ts=bucket(@timestamp, 1 minute) | SORT ts | LIMIT 5")) {
assertThat(
resp.columns(),
equalTo(List.of(new ColumnInfoImpl("sum(rate(request_count))", "double", null), new ColumnInfoImpl("ts", "date", null)))
);
List<List<Object>> values = EsqlTestUtils.getValuesList(resp);
assertThat(values, hasSize(sortedKeys.size()));
for (int i = 0; i < sortedKeys.size(); i++) {
List<Object> row = values.get(i);
assertThat(row, hasSize(2));
long key = sortedKeys.get(i);
assertThat(row.get(1), equalTo(DEFAULT_DATE_TIME_FORMATTER.formatMillis(key)));
List<Double> bucketValues = bucketToRates.get(key);
if (bucketValues.isEmpty()) {
assertNull(row.get(0));
} else {
assertThat((double) row.get(0), closeTo(bucketValues.stream().mapToDouble(d -> d).sum(), 0.1));
}
}
}
try (var resp = run("TS hosts | STATS avg(rate(request_count)) BY ts=bucket(@timestamp, 1minute) | SORT ts | LIMIT 5")) {
assertThat(
resp.columns(),
equalTo(List.of(new ColumnInfoImpl("avg(rate(request_count))", "double", null), new ColumnInfoImpl("ts", "date", null)))
);
List<List<Object>> values = EsqlTestUtils.getValuesList(resp);
assertThat(values, hasSize(sortedKeys.size()));
for (int i = 0; i < sortedKeys.size(); i++) {
List<Object> row = values.get(i);
assertThat(row, hasSize(2));
long key = sortedKeys.get(i);
assertThat(row.get(1), equalTo(DEFAULT_DATE_TIME_FORMATTER.formatMillis(key)));
List<Double> bucketValues = bucketToRates.get(key);
if (bucketValues.isEmpty()) {
assertNull(row.get(0));
} else {
double avg = bucketValues.stream().mapToDouble(d -> d).sum() / bucketValues.size();
assertThat((double) row.get(0), closeTo(avg, 0.1));
}
}
}
try (var resp = run("""
TS hosts
| STATS avg(rate(request_count)), avg(rate(request_count)) BY ts=bucket(@timestamp, 1minute)
| SORT ts
| LIMIT 5
""")) {
assertThat(
resp.columns(),
equalTo(
List.of(
new ColumnInfoImpl("avg(rate(request_count))", "double", null),
new ColumnInfoImpl("avg(rate(request_count))", "double", null),
new ColumnInfoImpl("ts", "date", null)
)
)
);
List<List<Object>> values = EsqlTestUtils.getValuesList(resp);
assertThat(values, hasSize(sortedKeys.size()));
for (int i = 0; i < sortedKeys.size(); i++) {
List<Object> row = values.get(i);
assertThat(row, hasSize(3));
long key = sortedKeys.get(i);
assertThat(row.get(2), equalTo(DEFAULT_DATE_TIME_FORMATTER.formatMillis(key)));
List<Double> bucketValues = bucketToRates.get(key);
if (bucketValues.isEmpty()) {
assertNull(row.get(0));
assertNull(row.get(1));
} else {
double avg = bucketValues.stream().mapToDouble(d -> d).sum() / bucketValues.size();
assertThat((double) row.get(0), closeTo(avg, 0.1));
assertThat((double) row.get(1), closeTo(avg, 0.1));
}
}
}
}

@AwaitsFix(bugUrl = "removed?")
public void testRateWithTimeBucketAndCluster() {
var rounding = new Rounding.Builder(TimeValue.timeValueSeconds(60)).timeZone(ZoneOffset.UTC).build().prepareForUnknown();
record RateKey(String host, String cluster, long interval) {}
Map<RateKey, List<RequestCounter>> groups = new HashMap<>();
for (Doc doc : docs) {
RateKey key = new RateKey(doc.host, doc.cluster, rounding.round(doc.timestamp));
groups.computeIfAbsent(key, k -> new ArrayList<>()).add(new RequestCounter(doc.timestamp, doc.requestCount));
}
record GroupKey(String cluster, long interval) {}
Map<GroupKey, List<Double>> rateBuckets = new HashMap<>();
for (Map.Entry<RateKey, List<RequestCounter>> e : groups.entrySet()) {
RateKey key = e.getKey();
List<Double> values = rateBuckets.computeIfAbsent(new GroupKey(key.cluster, key.interval), k -> new ArrayList<>());
Double rate = computeRate(e.getValue());
if (rate != null) {
values.add(rate);
}
}
Map<GroupKey, List<Double>> cpuBuckets = new HashMap<>();
for (Doc doc : docs) {
GroupKey key = new GroupKey(doc.cluster, rounding.round(doc.timestamp));
cpuBuckets.computeIfAbsent(key, k -> new ArrayList<>()).add(doc.cpu);
}
List<GroupKey> sortedKeys = rateBuckets.keySet()
.stream()
.sorted(Comparator.comparing(GroupKey::interval).thenComparing(GroupKey::cluster))
.limit(5)
.toList();
try (var resp = run("""
TS hosts
| STATS sum(rate(request_count)) BY ts=bucket(@timestamp, 1 minute), cluster
| SORT ts, cluster
| LIMIT 5""")) {
assertThat(
resp.columns(),
equalTo(
List.of(
new ColumnInfoImpl("sum(rate(request_count))", "double", null),
new ColumnInfoImpl("ts", "date", null),
new ColumnInfoImpl("cluster", "keyword", null)
)
)
);
List<List<Object>> values = EsqlTestUtils.getValuesList(resp);
assertThat(values, hasSize(sortedKeys.size()));
for (int i = 0; i < sortedKeys.size(); i++) {
List<Object> row = values.get(i);
assertThat(row, hasSize(3));
var key = sortedKeys.get(i);
assertThat(row.get(1), equalTo(DEFAULT_DATE_TIME_FORMATTER.formatMillis(key.interval)));
assertThat(row.get(2), equalTo(key.cluster));
List<Double> bucketValues = rateBuckets.get(key);
if (bucketValues.isEmpty()) {
assertNull(row.get(0));
} else {
assertThat((double) row.get(0), closeTo(bucketValues.stream().mapToDouble(d -> d).sum(), 0.1));
}
}
}
try (var resp = run("""
TS hosts
| STATS avg(rate(request_count)) BY ts=bucket(@timestamp, 1minute), cluster
| SORT ts, cluster
| LIMIT 5""")) {
assertThat(
resp.columns(),
equalTo(
List.of(
new ColumnInfoImpl("avg(rate(request_count))", "double", null),
new ColumnInfoImpl("ts", "date", null),
new ColumnInfoImpl("cluster", "keyword", null)
)
)
);
List<List<Object>> values = EsqlTestUtils.getValuesList(resp);
assertThat(values, hasSize(sortedKeys.size()));
for (int i = 0; i < sortedKeys.size(); i++) {
List<Object> row = values.get(i);
assertThat(row, hasSize(3));
var key = sortedKeys.get(i);
assertThat(row.get(1), equalTo(DEFAULT_DATE_TIME_FORMATTER.formatMillis(key.interval)));
assertThat(row.get(2), equalTo(key.cluster));
List<Double> bucketValues = rateBuckets.get(key);
if (bucketValues.isEmpty()) {
assertNull(row.get(0));
} else {
double avg = bucketValues.stream().mapToDouble(d -> d).sum() / bucketValues.size();
assertThat((double) row.get(0), closeTo(avg, 0.1));
}
}
}

try (var resp = run("""
TS hosts
| STATS
s = sum(rate(request_count)),
c = count(rate(request_count)),
max(rate(request_count)),
avg(rate(request_count))
BY ts=bucket(@timestamp, 1minute), cluster
| SORT ts, cluster
| LIMIT 5
| EVAL avg_rate= s/c
| KEEP avg_rate, `max(rate(request_count))`, `avg(rate(request_count))`, ts, cluster
""")) {
assertThat(
resp.columns(),
equalTo(
List.of(
new ColumnInfoImpl("avg_rate", "double", null),
new ColumnInfoImpl("max(rate(request_count))", "double", null),
new ColumnInfoImpl("avg(rate(request_count))", "double", null),
new ColumnInfoImpl("ts", "date", null),
new ColumnInfoImpl("cluster", "keyword", null)
)
)
);
List<List<Object>> values = EsqlTestUtils.getValuesList(resp);
assertThat(values, hasSize(sortedKeys.size()));
for (int i = 0; i < sortedKeys.size(); i++) {
List<Object> row = values.get(i);
assertThat(row, hasSize(5));
var key = sortedKeys.get(i);
assertThat(row.get(3), equalTo(DEFAULT_DATE_TIME_FORMATTER.formatMillis(key.interval)));
assertThat(row.get(4), equalTo(key.cluster));
List<Double> bucketValues = rateBuckets.get(key);
if (bucketValues.isEmpty()) {
assertNull(row.get(0));
assertNull(row.get(1));
} else {
double avg = bucketValues.stream().mapToDouble(d -> d).sum() / bucketValues.size();
assertThat((double) row.get(0), closeTo(avg, 0.1));
double max = bucketValues.stream().mapToDouble(d -> d).max().orElse(0.0);
assertThat((double) row.get(1), closeTo(max, 0.1));
}
assertEquals(row.get(0), row.get(2));
}
}
try (var resp = run("""
TS hosts
| STATS sum(rate(request_count)), max(cpu) BY ts=bucket(@timestamp, 1 minute), cluster
| SORT ts, cluster
| LIMIT 5""")) {
assertThat(
resp.columns(),
equalTo(
List.of(
new ColumnInfoImpl("sum(rate(request_count))", "double", null),
new ColumnInfoImpl("max(cpu)", "double", null),
new ColumnInfoImpl("ts", "date", null),
new ColumnInfoImpl("cluster", "keyword", null)
)
)
);
List<List<Object>> values = EsqlTestUtils.getValuesList(resp);
assertThat(values, hasSize(sortedKeys.size()));
for (int i = 0; i < sortedKeys.size(); i++) {
List<Object> row = values.get(i);
assertThat(row, hasSize(4));
var key = sortedKeys.get(i);
assertThat(row.get(2), equalTo(DEFAULT_DATE_TIME_FORMATTER.formatMillis(key.interval)));
assertThat(row.get(3), equalTo(key.cluster));
List<Double> rateBucket = rateBuckets.get(key);
if (rateBucket.isEmpty()) {
assertNull(row.get(0));
} else {
assertThat((double) row.get(0), closeTo(rateBucket.stream().mapToDouble(d -> d).sum(), 0.1));
}
List<Double> cpuBucket = cpuBuckets.get(key);
if (cpuBuckets.isEmpty()) {
assertNull(row.get(1));
} else {
assertThat((double) row.get(1), closeTo(cpuBucket.stream().mapToDouble(d -> d).max().orElse(0.0), 0.1));
}
}
}
try (var resp = run("""
TS hosts
| STATS sum(rate(request_count)), avg(cpu) BY ts=bucket(@timestamp, 1 minute), cluster
| SORT ts, cluster
| LIMIT 5""")) {
assertThat(
resp.columns(),
equalTo(
List.of(
new ColumnInfoImpl("sum(rate(request_count))", "double", null),
new ColumnInfoImpl("avg(cpu)", "double", null),
new ColumnInfoImpl("ts", "date", null),
new ColumnInfoImpl("cluster", "keyword", null)
)
)
);
List<List<Object>> values = EsqlTestUtils.getValuesList(resp);
assertThat(values, hasSize(sortedKeys.size()));
for (int i = 0; i < sortedKeys.size(); i++) {
List<Object> row = values.get(i);
assertThat(row, hasSize(4));
var key = sortedKeys.get(i);
assertThat(row.get(2), equalTo(DEFAULT_DATE_TIME_FORMATTER.formatMillis(key.interval)));
assertThat(row.get(3), equalTo(key.cluster));
List<Double> rateBucket = rateBuckets.get(key);
if (rateBucket.isEmpty()) {
assertNull(row.get(0));
} else {
assertThat((double) row.get(0), closeTo(rateBucket.stream().mapToDouble(d -> d).sum(), 0.1));
}
List<Double> cpuBucket = cpuBuckets.get(key);
if (cpuBuckets.isEmpty()) {
assertNull(row.get(1));
} else {
double avg = cpuBucket.stream().mapToDouble(d -> d).sum() / cpuBucket.size();
assertThat((double) row.get(1), closeTo(avg, 0.1));
}
}
}
}

public void testApplyRateBeforeFinalGrouping() {
record RateKey(String cluster, String host) {

Expand Down
Loading