Invalid facet distribution, attribute `xxx` is not filterable.

Created on 15 September 2023, about 1 year ago
Updated 2 November 2023, about 1 year ago

Problem/Motivation

It seems I can only add "rended HTML" and "title" to my index. When I add other fields, such as "body", or even a custom "content ID", there will be an error in indexing. The error messages look like this:

Invalid facet distribution, attribute `body` is not filterable. The available filterable attributes are `nid, rendered_item, search_api_datasource, search_api_id, search_api_language, title`.

I am not sure if it is a bug or not. Is it because of a misconfiguration, perhaps? Any help is much appreciated. Thank you in advance for your time!

Steps to reproduce

1. Install and activate the module
2. Configure server and add an index
3. Add fields (rendered HTML, body, title)
4. Visit the "View" tab of the manage index page
5. Sees the error message.
6. Reindex
7. Search -- no result. Sees the same error message.

However, when I check the status of the Meilisearch server, it looks like the items are indexed. But I am not sure whether it's an older index, or a current one.

🐛 Bug report
Status

Postponed: needs info

Version

1.1

Component

Code

Created by

🇯🇵Japan hktang

Live updates comments and jobs are added and updated live.
Sign in to follow issues

Comments & Activities

  • Issue created by @hktang
  • 🇯🇵Japan hktang

    I cannot reproduce this on a clean installation, where the facets work fine. Any idea where to look?

    By the way, there are other search indexes on my site, DB search and Solr. I am not sure if they affect how facets work on Meilisearch. Any pointer would be much appreciated!

  • 🇸🇮Slovenia DeaOm

    I was trying to reproduce this, but can't. I do have DB server and an index installed besides the Meilisearch one. The DB and Solr should not affect faceting of Meilisearch. We did however had some bugs regarding the filterable attributes.
    Did you try clearing all the indexed data and re-index again?
    Could you maybe try with the 2.x version, just so we can see if it's happening there too? The 2.x uses a more stable Meilisearch version 1.3.3 and there are some bug fixes applied, that could possible already solve this issue.

  • 🇯🇵Japan hktang

    Thank you! Let me try 2.x version and get back to you with my findings

  • Status changed to Postponed: needs info about 1 year ago
  • @hktang How did it go with the 2.x version? It would also be useful to know a few more information. Did this happen on a fresh install? If not, from what to what version did you upgrade? What version of Meilisearch are you using?

  • 🇯🇵Japan hktang

    Hi @admirlju unfortunately I haven't tried the 2.x yet. I had to keep our current Solr installation for now, but I hope to switch to Meilisearch soon.

    One thing I should report is that, our site has mix language content. The site language is set to English, but a fair amount of text is in Japanese. I am suspecting this has some effect on the errors I expeienced. Not sure if it's related though. It would be nice if there's some kind of language configuration in this plugin.

    Will try to revisit this again!

  • So I've quickly tested random Japanese text in an article content type and the language left to English. I'm not getting that error on 2.x, so either it's fixed in 2.x, or potentially something else is causing it.

  • 🇯🇵Japan hktang

    Hi there, I switched to version 2.x and so far it works perfectly.

    The CJK text was searchable (although not ideal) after enabling tokenizer for CJK.

    Thank you so much for your time building and maintaining this module!

Production build 0.71.5 2024