Using the solr_query JSON format

DSE Search supports advanced Solr search features to apply filters to CQL queries using JSON-formatted expressions. Also see Overriding the default TimeZone (UTC) in search queries.

JSON query syntax

The JSON query expression syntax is a JSON string. The JSON-based query expression supports local parameters in addition to the following parameters:

`{
  "q": <query_expression (string)>,
  "fq": <filter_query_expression(s) (string_or_array_of_strings, ...)>,
  "facet": <facet_query_expression (object)>
  "sort": <sort_expression (string)>,
  "start": <start_index(number)>,
timeAllowed: <search_time_limit_ms>,
  "TZ": <zoneID>),   //  Any valid zone ID in java TimeZone class
  "paging": <"driver" (string)>,
  "distrib.singlePass": true|false (boolean),
  "shards.failover": true|false (boolean),    // Default: true
  "shards.tolerant": true|false (boolean),    // Default: false
  "commit": true|false (boolean),
  "route.partition": <partition_routing_expression (array_of_strings)>,
  "route.range": <range_routing_expression (array_of_strings)>,
  "query.name": <query_name (string)>,
}

For more information: "q", "fq", "facet", "sort", "start", timeAllowed, "TZ" (Any valid zone ID in java TimeZone class), and "paging".

For example:

SELECT id FROM nhanes_ks.nhanes WHERE solr_query='{"q":"ethnicity:Asian"}';
SELECT id FROM nhanes_ks.nhanes WHERE solr_query='{"q":"ethnicity:Mexi*", "sort":"id asc"}' LIMIT 3;
SELECT * FROM mykeyspace.mytable WHERE solr_query='{"q":"{!edismax}quotes:yearning or kills"}';

To use Apache Solr™ Extended DisMax Query Parser (eDisMax) with solr_query, you must include defaultSearchField in your schema.

Making distributed queries tolerant of shard failures

Because distributed queries contact many shards, making queries more tolerant of shard failures ensures more successful completions. Use shards.failover and shards.tolerant parameters to define query failover and tolerance of shard failures during JSON queries:

Valid configurations Description

"shards.failover": true, "shards.tolerant": false,

This default configuration enables query failover and disables fault tolerance. Attempt to retry the failed shard requests when errors indicate that there is a reasonable chance of recovery.

"shards.failover": false, "shards.tolerant": true,

Disable query failover. Enable fault tolerance. Make the query succeed, even if the query only partially succeeded, and did not succeed for all nodes.

"shards.failover": false, "shards.tolerant": false,

Disable query failover. Disable fault tolerance.

Failover and tolerance of partial results cannot coexist in the same JSON query. Queries support enabling tolerance for only one parameter.

The shards.tolerant parameter is not supported when deep paging is on.

Other fault tolerance configuration options include netty_client_request_timeout in dse.yaml and read_request_timeout_in_ms in cassandra.yaml.

JSON queries with literal characters that are Apache Solr™/Apache Lucene® special characters

Solr and Lucene support escaping special characters that are part of the query syntax.

Character Description

+

Plus. Required search term operator.

-

Minus. Prohibited search term operator.

&&

Double ampersand. AND operator. Both terms either side of the operator are required for a match.

||

Double pipe. OR operator.

!

Exclamation mark. NOT operator.

(

Left parenthesis

)

Right parenthesis

"

Double quote

~

Tilde

*

Asterisk

?

Question mark

:

Colon

Using JSON with solr_query requires additional syntax for literal characters that are Lucene special characters.

Syntax for a simple search string:

Simple search string

mytestuser1?

Solr query

name:mytestuser1\?

CQL Solr query

solr_query='{"q":"name:mytestuser1\\?"}'

Syntax for a complex search string:

Complex search string

(1+1):2

Solr query

e:(1\+1)\:2

CQL Solr query

solr_query='{"q":"e:\(1\\+1)\\:2"}'

Field, query, and range faceting with a JSON query

Specify the facet parameters inside a facet JSON object to perform field, query, and range faceting inside Solr queries. Distributed pivot faceting is supported.The query syntax is less verbose to specify facets by:

  • Specifying each facet parameter without the facet prefix that is required by HTTP APIs.

  • Expressing multiple facet fields and queries inside a JSON array.

    Faceted search example
    SELECT * FROM solr WHERE solr_query='{"q":"id:*","facet":{"field":"type"}}';
Query facet example
SELECT * FROM solr WHERE solr_query='{"q":"id:*","facet":{"query":"type:0"}}';
Multiple queries example
SELECT * FROM solr WHERE solr_query='{"q":"id:*","facet":{"query":["type:0","type:1"]}}';
Distributed pivot faceting example
SELECT id FROM <table> WHERE solr_query='{"q":"id:*","facet":{"pivot":"type,value","limit":"10"}}'
Range facet example
SELECT * FROM solr WHERE solr_query='{"q":"id:*","facet":{"range":"type", "f.type.range.start":-10, "f.type.range.end":10, "range.gap":1}}}';

The returned result is formatted as a single row with each column corresponding to the output of a facet (either field, query, or range). The value is represented as a JSON blob because facet results can be complex and nested. For example:

facet_fields           | facet_queries
------------------------+-------------------------
 {"type":{"0":2,"1":1}} | {"type:0":2,"type:1":1}
Range by date facet example
SELECT * FROM solr WHERE solr_query='{"q":"business_date:*","facet":{"range":"business_date", "f.business_date.range.start":"2015-01-01T00:00:00Z", "f.business_date.range.end":"2015-08-01T00:00:00Z", "f.business_date.range.gap":"+1MONTH"}}';

Solr range facets before, after, and between might return incorrect and inconsistent results on multi-node clusters. See SOLR-6187 and SOLR-6375.

Interval facet example
SELECT * FROM solr WHERE solr_query='{"q":"id:*","facet":{"interval":"id", "interval.set":"[*,500]"}';

Tracing distributed queries

During a distributed query, every node is responsible for a set of token ranges. A shard is the node/ranges combination. The shard token range is reported:

  • In the shards.info response for HTTP queries.

  • In the system_traces.events table for HTTP queries that provide cassandra.trace=true and CQL Solr queries that enable tracing at the driver level.

Solr single-pass CQL queries

Single-pass distributed queries are supported in CQL Solr queries.

There are a number of ways to use a single-pass distributed query instead of the standard two-pass query. You can:

  • Specify the distrib.singlePass Boolean parameter in the query expression. Example:

        SELECT * FROM keyspace.table WHERE solr_query = '{"q" : "*:*", "distrib.singlePass" : true}'
  • Use a token() or partition key restriction in the WHERE clause. Example:

    SELECT * FROM keyspace.table WHERE token(id) >= 3074457345618258604 AND solr_query = '{"q" : "*:*"}'
  • Execute a COUNT query. Example:

    SELECT count(*) FROM keyspace.table WHERE solr_query = '{"q" : "*:*"}'
  • Specify the primary key elements in the SELECT clause. Example:

    CREATE TABLE cycling.cyclist_name (
       id UUID,
       lastname text,
       firstname text );
    PRIMARY KEY ((id), lastname);
    CREATE SEARCH INDEX ON cycling.cyclist_name ... ;
    dsetool create_core cycling.cyclist_name ...
    INSERT into cycling.cyclist_name ... ;
    SELECT id, lastname FROM cycling.cyclist_name
    WHERE solr_query='{"q": "*:*","fq":"(id:a6f94417-e27b-444f-9d0c-dccb588e421f)"}';

Using a single-pass distributed query has an operational cost that includes potentially more disk and network overhead. With single-pass queries, each node reads all rows that satisfy the query and returns them to the coordinator node. An advanced feature, a single-pass distributed query saves one network round trip transfer during the retrieval of queried rows. A regular distributed query performs two network round trips:

  1. The first one to retrieve IDs from DSE Search that satisfy the query.

  2. The second one to retrieve only the rows that satisfy the query from the database, based on IDs from the first step.

Single-pass distributed queries are most efficient when most of the documents found are returned in the search results. Single-pass distributed queries are inefficient when most of the documents found are not returned to the coordinator node.

For example, a distributed query that only fans out to a single node from the coordinator node will likely be most efficient as a single-pass query.

[IMPORTANT

With single-pass queries, there is a limitation that returns only the table columns that are defined in the Solr schema.xml in the query results. This limitation also applies to map entries that do not conform to the field mapping.

Consider the cycling.cyclist_name table example above. If the column firstname was not indexed, but needed to be retrievable by queries, the entry in the schema.xml for the firstname field could be:

<field indexed="false" multiValued="false" docValues="false" name="firstname" type="StrField"></field>

If the Solr core was created without a customized schema.xml file and generateResources=true was included, the default schema.xml file includes all fields that exist in the table. For more information, see dsetool create_core.

JSON query name option

Using the following syntax to name your queries to support metrics and monitoring for performance objects. Naming queries can be useful for tagging and JMX operations, for example.

SELECT id FROM nhanes_ks.nhanes WHERE solr_query=' {"query.name":"Asian subjects", "q":"ethnicity:Asia*"}' LIMIT 50;

JSON query commit option

When executing custom queries after bulk document loading, with auto soft commit disabled (or is an extremely infrequent configured value), use the JSON query commit option to ensure that all pending updates are soft-committed before the running a query. The commit makes the latest data visible to a query.

Do not use the JSON commit option for live operations against a production cluster. The commit option is not a replacement for the normal auto soft commit process or the COMMIT SEARCH INDEX command.

By default, the commit option is set to false. The example below sets the commit option to true:

SELECT * FROM wiki.solr
WHERE solr_query='{"q":"title:Asia*", "commit":true}'
LIMIT 50;

Queries that dynamically enable paging using a driver

To dynamically enable paging when cql_solr_query_paging is set to off in dse.yaml, set the Solr paging parameter to driver ("paging":"driver"). For example:

SELECT id FROM wiki.solr
WHERE solr_query= '{"q":"*", "sort":"id asc", "paging":"driver"}' ;

Limiting queries by time

DSE Search supports limiting queries by time by using the Solr timeAllowed parameter.

Restriction: DSE Search differs from native Solr:

  • If timeAllowed is exceeded, an exception is thrown.

  • If timeAllowed is exceeded, and the additional shards.tolerant parameter is set to true, the application returns the partial results collected so far.

When partial results are returned, the CQL custom payload contains the DSESearch.isPartialResults key.

Example with a 30 second timeout:

SELECT * FROM users
WHERE solr_query = '{ "q": "*:*", "timeAllowed":30000}';

The Solr timeAllowed parameter is enabled by default to prevent long running shard queries, such as complex facets and Boolean queries, from using system resources after they have timed out from the DSE Search coordinator.

DSE Search checks the timeout per segment instead of during document or terms iteration. The system property solr.timeAllowed.docsPerSample has been removed.

By default, the value for timeAllowed is the same as the internode_messaging_options.client_request_timeout_seconds parameter in dse.yaml. Queries that breach client_request_timeout_seconds fail by default.

The 50th percentile latency for queries using timeAllowed should be within 5% of the same query that does not use timeAllowed.

The timeAllowed parameter applies to these queries:

  • Standard queries and filtering queries, where the timeout is applied on a match collection.

  • Facet queries, including pivot facets.

The json.facet queries are not supported.

Queries that are terminated due to a timeout are logged with a warning.

Using the Solr timeAllowed parameter may cause a latency cost. If you find the cost for queries is too high in your environment, consider setting the -Ddse.timeAllowed.enabled.default property to false at DSE startup time. Or set timeAllowed.enable to false in the query.

Was this helpful?

Give Feedback

How can we improve the documentation?

© 2024 DataStax | Privacy policy | Terms of use

Apache, Apache Cassandra, Cassandra, Apache Tomcat, Tomcat, Apache Lucene, Apache Solr, Apache Hadoop, Hadoop, Apache Pulsar, Pulsar, Apache Spark, Spark, Apache TinkerPop, TinkerPop, Apache Kafka and Kafka are either registered trademarks or trademarks of the Apache Software Foundation or its subsidiaries in Canada, the United States and/or other countries. Kubernetes is the registered trademark of the Linux Foundation.

General Inquiries: +1 (650) 389-6000, info@datastax.com