Using docValues should be the default for all string fields due to performance reasons. But there's an important edge case:

String fields are using SORTED_SET, so multiple identical entries are collapsed into a single value. Thus if I insert values 4, 5, 2, 4, 1, my return will be 1, 2, 4, 5 when enabling docValues.

If you need to preserve the order and duplicate entries, consider to store the values as zm_* (twice). Therefor select the "solr_string_storage" custom field type.

CommentFileSizeAuthor
#2 3024554.patch1.6 KBmkalkbrenner
Support from Acquia helps fund testing for Drupal Acquia logo

Comments

mkalkbrenner created an issue. See original summary.

mkalkbrenner’s picture

mkalkbrenner
Primary language German
Location 🇩🇪
CreditAttribution: mkalkbrenner at bio.logis Genetic Information Management GmbH commented
Status: Active » Needs review
FileSize
3024554.patch1.6 KB

  • mkalkbrenner committed 2676d84 on 8.x-3.x 
    Issue #3024554 by mkalkbrenner: Solr 7.x: use docValues="true" for sm_*
mkalkbrenner’s picture

mkalkbrenner
Primary language German
Location 🇩🇪
CreditAttribution: mkalkbrenner at bio.logis Genetic Information Management GmbH commented
Status: Needs review » Fixed

Status: Fixed » Closed (fixed)

Automatically closed - issue fixed for 2 weeks with no activity.

Jancy Christopher’s picture

Jancy Christopher CreditAttribution: Jancy Christopher commented

Hi mkalkbrenner,

As mentioned above

If you need to preserve the order and duplicate entries, consider to store the values as zm_* (twice). Therefor select the "solr_string_storage" custom field type.

Can you explain select the solr_string_storage custom field type?

Thanks for your feedback

Jancy Christopher’s picture

Jancy Christopher CreditAttribution: Jancy Christopher commented

Hi mkalkbrenner,

I Have upgraded my search Search API Solr module from 8.x-2.0-alpha2 to 8.x-3.3.
I can see there is difference in how the best bet fields are indexed.

Added in indexed field as,
field exclude - data type boolean
field query_text - data type string

On solr 8.x-2.0-alpha2 the best bet values were indexed as below.

"response":{"numFound":1,"start":0,"docs":[
      {
        "bm_exclude":[false,
          false,
          false,
          false,
          false,
          true],
        "sm_query_text":["partners",
          "partner",
          "clients",
          "investors",
          "ecosystem",
          "excludeprop",]}]

and now the bm_exclude values does not presave the duplication and order.
It is displaying as below.

 "response":{"numFound":1,"start":0,"docs":[
      {
        "bm_exclude":[false,
          true],
        "id":"8cdkf4-cgi_index-entity:node/57362:en",
        "sm_query_text":["clients",
          "ecosystem",
          "excludeprop",
          "investors",
          "partner",
          "partners"]}]

Any suggestion on the above will be good thanks.

mkalkbrenner’s picture

mkalkbrenner
Primary language German
Location 🇩🇪
CreditAttribution: mkalkbrenner at bio.logis Genetic Information Management GmbH commented

You must not use docValues in this case. You can modify the field type using the yaml file or even better provide a dedicated type for that. The best way would be if that module provides its dedicated Solr field type.

Jancy Christopher’s picture

Jancy Christopher CreditAttribution: Jancy Christopher commented

@mkalkbrenner, Thanks for your feedback.

To fix I have removed the docValues for boolean from schema.xml file.
For the string field I have changed the field type to storage-only(solr_string_storage custom field type option available in search_api_solr).

This preserves the duplication and order.