Spring Data Elasticsearch返回空聚合桶及命中数异常问题求助

Let's break down the possible issues and fixes for your problem—since your logged query matches what works in Postman but returns no results in Java, we can narrow down the root causes step by step:

1. Confirm Critical Query Configurations Are Enabled #

Looking at your shared code, these key lines are commented out:

// .setQuery(bools)
// .setSize(0)
// .addAggregation(streetAggregation);

If you ran the code with these commented, you're sending an empty search request (no filter, no aggregation) which would only return 0 hits if the index is empty—but you said Postman gets results, so this is likely a typo in your shared code. Assuming you have these uncommented in your actual implementation, let's move to other potential issues.

2. Verify Your Java Client Connects to the Correct ES Cluster #

This is a surprisingly common gotcha: your Java app might be pointing to a different Elasticsearch cluster (e.g., staging vs production) than what you're hitting with Postman. Double-check your elasticsearchClient configuration—ensure node URLs, ports, and any authentication credentials match exactly what you use in Postman.

3. Fix the Document Type Specification #

You commented out .setTypes("pallc"), but in Elasticsearch 6.x (your version), each index can only have one document type. If your pallc_index uses pallc as its type (instead of the default _doc), omitting this line means your Java query is searching for the non-existent _doc type. Uncomment that line to align with your index's structure:

.setTypes("pallc")

4. Ensure Query Values Match Field Data Types #

Your query uses numeric values for zip code fields:

QueryBuilder bools = QueryBuilders.boolQuery()
    .must(matchQuery("zip-code", 90402))
    .must(matchQuery("zip-plus-4", 2411));

If zip-code and zip-plus-4 are mapped as text or keyword types (standard for zip codes), passing integer values might not match the stored string values. Postman is likely sending these values as strings ("90402", "2411"), which works correctly. Update your query to use string literals:

QueryBuilder bools = QueryBuilders.boolQuery()
    .must(matchQuery("zip-code", "90402"))
    .must(matchQuery("zip-plus-4", "2411"));

5. Validate Raw Request Serialization #

Sometimes Spring Data Elasticsearch's client serializes requests slightly differently than manual Postman calls. Enable debug logging for the ES client to see the exact raw JSON request being sent—add this to your application.properties:

logging.level.org.elasticsearch.client=DEBUG

Compare the raw JSON from Java with your Postman request—look for subtle differences like missing type declarations, incorrect value formatting, or misspelled field names.

6. Check Index Permissions or Alias Issues #

• If pallc_index is an alias, confirm the Java client has access to the underlying index it points to.
• Verify the user your Java app uses to connect to ES has read permissions on pallc_index.

Try these steps one by one—most likely, the issue stems from a missing document type, mismatched value types in the query, or connecting to the wrong cluster.

内容的提问来源于stack exchange，提问作者Steve Davis