Changing auto-generated search index settings
Using dsetool, you can customize the default settings for auto-generated search indexes by providing a YAML-formatted file with these options:
- auto_soft_commit_max_time:ms
-
The maximum auto soft commit time in milliseconds.
- default_query_field:field
-
The query field to use when no field is specified in queries.
- distributed=(true|false)
-
Whether to distribute and apply the operation to all nodes in the local datacenter.
-
True applies the operation to all nodes in the local datacenter.
-
False applies the operation only to the node it was sent to. False works only when recovery=true.
-
Default: true
|
Distributing a re-index to an entire datacenter degrades performance severely in that datacenter. |
- enable_tokenized_text_copy_fields:( true | false )
-
Whether to generate tokenized text.
Default: false
- exclude_columns: col1, col2, col3, …
-
A comma-separated (CSV) list of columns to exclude.
- generate_DocValues_for_fields:( * | field1, field2, …)
-
The fields to automatically configure DocValues in the generated search index schema. Specify '*' to add all possible fields:
generate_DocValues_for_fields: '*'or specify a comma-separated list of fields, for example:
generate_DocValues_for_fields: uuidfield, bigintfieldDue to SOLR-7264, setting docValues to true on a boolean field in the Solr schema does not work. A workaround for boolean docValues is to use 0 and 1 with a TrieIntField.
- generateResources=(true|false)
-
Whether to automatically generate search index resources based on the existing CQL table metadata. Cannot be used with
schema=andsolrconfig=.Valid values:
-
true - Automatically generate search index schema and configuration resources if resources do not already exist.
-
false - Default. Do not automatically generate search index resources.
-
- include_columns=col1, col2, col3, …
-
A comma-separated (CSV) list of columns to include. Empty = includes all columns.
- index_merge_factor:segments
-
How many segments of equal size to build before merging them into a single segment.
- lenient=( true | false )
-
Ignore non-supported type columns and continue to generate resources, instead of erroring out when non-supported type columns are encountered.
Default: false
- resource_generation_profiles
-
To minimize index size, specify a CSV list of profiles to apply while generating resources.
Resource generation profiles Profile name Description spaceSavingAll
Applies spaceSavingNoJoin and spaceSavingSlowTriePrecision profiles.
spaceSavingNoJoin
Do not index a hidden primary key field. Prevents joins across cores.
spaceSavingSlowTriePrecision
Sets trie fields precisionStep to '0', allowing for greater space saving but slower querying.
Using spaceSavings profiles disables auto generation of DocValues.
For example:
resource_generation_profiles: spaceSavingNoJoin, spaceSavingSlowTriePrecision - rt=true
-
Whether to enable live indexing to increase indexing throughput. Enable live indexing on only one search index per cluster.
rt=true
CQL index management command examples
For example:
CREATE SEARCH INDEX CONFIG ON wiki.solr SET defaultQueryField='last_name';Using dsetool
Customize the search index config with YAML-formatted files
Create a config.yaml file that lists the following options to customize the config and schema files:
default_query_field: name
auto_soft_commit_max_time: 1000
generate_DocValues_for_fields: '*'
enable_string_copy_fields: falseUse the dsetool command to generate the search index with these options to customize the config and schema generation.
Use coreOptions to specify the config.yaml file:
dsetool create_core demo.health_data coreOptions=config.yamlCustomize the search index with options inline
Use the dsetool command to generate the search index and customize the schema generation. Use coreOptions to turn on live indexing (also called RT):
dsetool create_core udt_ks.users generateResources=true reindex=true coreOptions=rt.yamlYou can verify that DSE Search created the solrconfig and schema by reading core resources using dsetool.
Enable encryption for a new search index
Specify the class for directoryFactory to solr.EncryptedFSDirectoryFactory with coreOptionsInline:
dsetool create_core <keyspace_name>.<<table_name>> generateResources=true coreOptionsInline="directory_factory_class:solr.EncryptedFSDirectoryFactory"