opensearch-project/opensearch-benchmark-workloads
[FEATURE] Adding match query on high cardinality field in Big5
Open
#617 opened on Apr 21, 2025
enhancementgood first issue
Repository metrics
- Stars
- (33 stars)
- PR merge metrics
- (PR metrics pending)
Description
Is your feature request related to a problem?
When upgrading OpenSearch to Lucene 9.12.0 for 2.18 release, we caught a regression in Lucene from the keyword-in-range operation in Big5. It was related to query not reading skip data and treating non-competitive docs as competitive. See https://github.com/apache/lucene/issues/14445 for details.
"query": {
"bool": {
"must": [{
"range": {
"@timestamp": {
"gte": "2023-01-01T00:00:00",
"lt": "2023-01-03T00:00:00"
}
}
}, {
"match": {
"process.name": "kernel"
}
}]
}
}
The regression could have been identified if just the match query was an operation in itself.
What solution would you like?
Add match query on a high cardinality field process.name as a separate operation.
"match": {
"process.name": "kernel"
}