Skip to content

Commit

Permalink
Fix InternalAutoDateHistogramTests (backport of #54602) (#54688)
Browse files Browse the repository at this point in the history 
The test had errors around time units that have different length - think
leap years or months that aren't 30 days. This fixes those errors. In
the proces I've changed a bunch of things to debug the problem:

* Replace `currentTimeMillis` with a random time. Now the test fails
  randomly! Wonderful. Much better than on random days of the month.
* Generate buckets "closer together" to test random reduction. Without
  this we were super frequently getting stuck in the "year of century"
  rounding because *some* of the of the buckets we built were far apart.
  This generates a much greater variety of tests.
* Implement `toString` on `RoundingInfo` so I can debug without going
  crazy.
* Switch keys in the bucket assertions from epoch millis to `Instant`s
  so we can read the failures.

Closes #54540
Closes #39497
  • Loading branch information
@nik9000
nik9000 committed Apr 3, 2020
1 parent 9841b74 commit 38f4b03
Show file tree
Hide file tree
Showing 2 changed files with 60 additions and 43 deletions.
5 changes: 5 additions & 0 deletions ...asticsearch/search/aggregations/bucket/histogram/AutoDateHistogramAggregationBuilder.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -298,5 +298,10 @@ public boolean equals(Object obj) {
&& Objects.equals(dateTimeUnit, other.dateTimeUnit)
;
}

@Override
public String toString() {
return "RoundingInfo[" + rounding + " " + Arrays.toString(innerIntervals) + "]";
}
}
}
98 changes: 55 additions & 43 deletions ...rg/elasticsearch/search/aggregations/bucket/histogram/InternalAutoDateHistogramTests.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -21,6 +21,7 @@

import org.elasticsearch.common.Rounding;
import org.elasticsearch.common.io.stream.Writeable;
import org.elasticsearch.common.time.DateFormatter;
import org.elasticsearch.search.DocValueFormat;
import org.elasticsearch.search.aggregations.InternalAggregations;
import org.elasticsearch.search.aggregations.ParsedMultiBucketAggregation;
Expand All @@ -40,21 +41,23 @@
import java.util.List;
import java.util.Map;
import java.util.TreeMap;
import java.util.concurrent.TimeUnit;

import static org.elasticsearch.common.unit.TimeValue.timeValueHours;
import static org.elasticsearch.common.unit.TimeValue.timeValueMinutes;
import static org.elasticsearch.common.unit.TimeValue.timeValueSeconds;
import static org.hamcrest.Matchers.equalTo;

public class InternalAutoDateHistogramTests extends InternalMultiBucketAggregationTestCase<InternalAutoDateHistogram> {

private DocValueFormat format;
private RoundingInfo[] roundingInfos;
private long defaultStart;
private int roundingIndex;

@Override
public void setUp() throws Exception {
super.setUp();
format = randomNumericDocValueFormat();
defaultStart = randomLongBetween(0, DateFormatter.forPattern("date_optional_time").parseMillis("2050-01-01"));
roundingIndex = between(0, AutoDateHistogramAggregationBuilder.buildRoundings(null, null).length - 1);
}

@Override
Expand All @@ -68,10 +71,13 @@ protected InternalAutoDateHistogram createTestInstance(String name,
int targetBuckets = randomIntBetween(1, nbBuckets * 2 + 1);
List<InternalAutoDateHistogram.Bucket> buckets = new ArrayList<>(nbBuckets);

long startingDate = System.currentTimeMillis();
long startingDate = defaultStart;
if (rarely()) {
startingDate += randomFrom(TimeUnit.MINUTES, TimeUnit.HOURS, TimeUnit.DAYS).toMillis(between(1, 10000));
}

long interval = randomIntBetween(1, 3);
long intervalMillis = randomFrom(timeValueSeconds(interval), timeValueMinutes(interval), timeValueHours(interval)).getMillis();
long intervalMillis = roundingInfos[roundingIndex].roughEstimateDurationMillis * interval;

for (int i = 0; i < nbBuckets; i++) {
long key = startingDate + (intervalMillis * i);
Expand Down Expand Up @@ -108,10 +114,6 @@ public void testGetAppropriateRoundingUsesCorrectIntervals() {
assertThat(result, equalTo(2));
}

public void testReduceRandom() {
super.testReduceRandom();
}

@Override
protected void assertReduced(InternalAutoDateHistogram reduced, List<InternalAutoDateHistogram> inputs) {

Expand All @@ -134,41 +136,49 @@ protected void assertReduced(InternalAutoDateHistogram reduced, List<InternalAut
RoundingInfo roundingInfo = roundingInfos[roundingIndex];

long normalizedDuration = (highest - lowest) / roundingInfo.getRoughEstimateDurationMillis();
long innerIntervalToUse = roundingInfo.innerIntervals[0];
int innerIntervalIndex = 0;

// First, try to calculate the correct innerInterval using the normalizedDuration.
// This handles cases where highest and lowest are further apart than the interval being used.
/*
* Guess the interval to use based on the roughly estimated
* duration. It'll be accurate or it'll produce more buckets
* than we need but it is quick.
*/
if (normalizedDuration != 0) {
for (int j = roundingInfo.innerIntervals.length-1; j >= 0; j--) {
int interval = roundingInfo.innerIntervals[j];
if (normalizedDuration / interval < reduced.getBuckets().size()) {
innerIntervalToUse = interval;
innerIntervalIndex = j;
}
}
}

long intervalInMillis = innerIntervalToUse * roundingInfo.getRoughEstimateDurationMillis();
int bucketCount = getBucketCount(lowest, highest, roundingInfo, intervalInMillis);

//Next, if our bucketCount is still above what we need, we'll go back and determine the interval
// based on a size calculation.
if (bucketCount > reduced.getBuckets().size()) {
for (int i = innerIntervalIndex; i < roundingInfo.innerIntervals.length; i++) {
long newIntervalMillis = roundingInfo.innerIntervals[i] * roundingInfo.getRoughEstimateDurationMillis();
if (getBucketCount(lowest, highest, roundingInfo, newIntervalMillis) <= reduced.getBuckets().size()) {
innerIntervalToUse = roundingInfo.innerIntervals[i];
intervalInMillis = innerIntervalToUse * roundingInfo.getRoughEstimateDurationMillis();
}
/*
* Next pick smaller intervals until we find the one that makes the right
* number of buckets.
*/
int innerIntervalToUse;
do {
innerIntervalToUse = roundingInfo.innerIntervals[innerIntervalIndex];
int bucketCount = getBucketCount(lowest, highest, roundingInfo.rounding, innerIntervalToUse);
if (bucketCount == reduced.getBuckets().size()) {
break;
}
}

Map<Long, Long> expectedCounts = new TreeMap<>();
for (long keyForBucket = roundingInfo.rounding.round(lowest);
keyForBucket <= roundingInfo.rounding.round(highest);
keyForBucket = keyForBucket + intervalInMillis) {
expectedCounts.put(keyForBucket, 0L);
if (bucketCount < reduced.getBuckets().size()) {
innerIntervalToUse = roundingInfo.innerIntervals[Math.max(0, innerIntervalIndex - 1)];
break;
}
} while (++innerIntervalIndex < roundingInfo.innerIntervals.length);

assertThat(reduced.getInterval().toString(), equalTo(innerIntervalToUse + roundingInfo.unitAbbreviation));
Map<Instant, Long> expectedCounts = new TreeMap<>();
long keyForBucket = roundingInfo.rounding.round(lowest);
while (keyForBucket <= roundingInfo.rounding.round(highest)) {
long nextKey = keyForBucket;
for (int i = 0; i < innerIntervalToUse; i++) {
nextKey = roundingInfo.rounding.nextRoundingValue(nextKey);
}
Instant key = Instant.ofEpochMilli(keyForBucket);
expectedCounts.put(key, 0L);

// Iterate through the input buckets, and for each bucket, determine if it's inside
// the range of the bucket in the outer loop. if it is, add the doc count to the total
Expand All @@ -178,26 +188,26 @@ protected void assertReduced(InternalAutoDateHistogram reduced, List<InternalAut
for (Histogram.Bucket bucket : histogram.getBuckets()) {
long roundedBucketKey = roundingInfo.rounding.round(((ZonedDateTime) bucket.getKey()).toInstant().toEpochMilli());
long docCount = bucket.getDocCount();
if (roundedBucketKey >= keyForBucket
&& roundedBucketKey < keyForBucket + intervalInMillis) {
expectedCounts.compute(keyForBucket,
(key, oldValue) -> (oldValue == null ? 0 : oldValue) + docCount);
if (roundedBucketKey >= keyForBucket && roundedBucketKey < nextKey) {
expectedCounts.compute(key,
(k, oldValue) -> (oldValue == null ? 0 : oldValue) + docCount);
}
}
}
keyForBucket = nextKey;
}

// If there is only a single bucket, and we haven't added it above, add a bucket with no documents.
// this step is necessary because of the roundedBucketKey < keyForBucket + intervalInMillis above.
if (roundingInfo.rounding.round(lowest) == roundingInfo.rounding.round(highest) && expectedCounts.isEmpty()) {
expectedCounts.put(roundingInfo.rounding.round(lowest), 0L);
expectedCounts.put(Instant.ofEpochMilli(roundingInfo.rounding.round(lowest)), 0L);
}


// pick out the actual reduced values to the make the assertion more readable
Map<Long, Long> actualCounts = new TreeMap<>();
Map<Instant, Long> actualCounts = new TreeMap<>();
for (Histogram.Bucket bucket : reduced.getBuckets()) {
actualCounts.compute(((ZonedDateTime) bucket.getKey()).toInstant().toEpochMilli(),
actualCounts.compute(((ZonedDateTime) bucket.getKey()).toInstant(),
(key, oldValue) -> (oldValue == null ? 0 : oldValue) + bucket.getDocCount());
}
assertEquals(expectedCounts, actualCounts);
Expand All @@ -211,11 +221,13 @@ protected void assertReduced(InternalAutoDateHistogram reduced, List<InternalAut
assertThat(reduced.getInterval(), equalTo(expectedInterval));
}

private int getBucketCount(long lowest, long highest, RoundingInfo roundingInfo, long intervalInMillis) {
private int getBucketCount(long min, long max, Rounding rounding, int interval) {
int bucketCount = 0;
for (long keyForBucket = roundingInfo.rounding.round(lowest);
keyForBucket <= roundingInfo.rounding.round(highest);
keyForBucket = keyForBucket + intervalInMillis) {
long key = rounding.round(min);
while (key < max) {
for (int i = 0; i < interval; i++) {
key = rounding.nextRoundingValue(key);
}
bucketCount++;
}
return bucketCount;
Expand Down

0 comments on commit 38f4b03

Please sign in to comment.