# Is this a bug at DateHistogram aggregation?
**URL:** https://discuss.elastic.co/t/is-this-a-bug-at-datehistogram-aggregation/383069
**Category:** Elasticsearch
**Created:** [October 29, 2025, 10:20am UTC](https://discuss.elastic.co/t/is-this-a-bug-at-datehistogram-aggregation/383069 "2025-10-29T10:20:45Z")
**Posts on this page:** 1
**Page:** 1
### Author:  [@vicety](https://discuss.elastic.co/u/vicety)
#### Post date: [October 29, 2025, 10:20am UTC](https://discuss.elastic.co/t/is-this-a-bug-at-datehistogram-aggregation/383069/1 "2025-10-29T10:20:46Z")
I was reading the source code of histogram aggregation and found this.
```auto
// server/src/main/java/org/elasticsearch/search/aggregations/bucket/histogram/InternalDateHistogram.java
// (line 358)
private void addEmptyBuckets(List list, AggregationReduceContext reduceContext) {
/*
* Make sure we have space for the empty buckets we're going to add by
* counting all of the empties we plan to add and firing them into
* consumeBucketsAndMaybeBreak.
*/
class Counter implements LongConsumer {
private int size;
@Override
public void accept(long key) {
size++;
if (size >= REPORT_EMPTY_EVERY) {
reduceContext.consumeBucketsAndMaybeBreak(size);
size = 0;
}
}
}
Counter counter = new Counter();
iterateEmptyBuckets(list, list.listIterator(), counter);
reduceContext.consumeBucketsAndMaybeBreak(counter.size);
InternalAggregations reducedEmptySubAggs = InternalAggregations.reduce(emptyBucketInfo.subAggregations, reduceContext);
ListIterator iter = list.listIterator();
iterateEmptyBuckets(list, iter, new LongConsumer() {
private int size = 0;
@Override
public void accept(long key) {
size++;
if (size >= REPORT_EMPTY_EVERY) {
reduceContext.consumeBucketsAndMaybeBreak(size);
size = 0;
}
iter.add(new InternalDateHistogram.Bucket(key, 0, format, reducedEmptySubAggs));
}
});
}
```
It seems that the first `iterateEmptyBuckets` call is to make sure we won’t have too many empty buckets that causes OOM (by calling `reduceContext.consumeBucketsAndMaybeBreak`), and the second `iterateEmptyBuckets` call is the one that actually puts empty buckets into the result.
My question is, why is `reduceContext.consumeBucketsAndMaybeBreak` called in both the first and second call to `iterateEmptyBuckets`? Shouldn’t we only keep this in the first call to `iterateEmptyBuckets` but not the second one?