Issue(random) while attaching published articles to a case

[ShankyGrover]

Hi Community Members,

I have recently observed an issue in our ServiceNow Production instance where certain knowledge articles do not consistently appear in search results.

The issue occurs in the following scenarios:

1. Knowledge Homepage: When searching by the article's short description or article number, certain articles do not appear in the search results. However, the same articles can be found when using the global search with the article number.

2. Related Search Results on Case Forms: The same articles are sometimes missing from the Related Search Results when searching for relevant knowledge to attach to an existing case.

The issue appears to affect random articles. Upon verification, the affected articles are in the Published state, and their Valid To dates have not yet expired.

Has anyone encountered similar behavior or can suggest possible causes and troubleshooting steps? 

Any guidance or suggestions would be greatly appreciated.

Thank you!

[prajaktakha]

Hello @ShankyGrover ,

Even if the article's state is Published and its Valid To date has not expired, please check whether the article is active.

If the article's Active field is set to false, it will not appear in the Related Search results.

If the article is inactive, you can activate it using a background script, as list editing is restricted.

If this solution helps, please mark it as helpful and accept it as the solution.

Regards,
Prajakta

[sanjaygupta]

This sounds like a search indexing problem rather than a knowledge visibility problem. The articles are published and still valid, and global search can find them by number, so the likely culprit is that the Knowledge Homepage and Related Search Results use a different search source (AI Search or the legacy Zing text index) that has not fully indexed those articles. Here’s how I’d narrow it down.

First, check whether the articles are actually in the index the knowledge search uses.

• Open AI Search > Search Sources and find the source for the Knowledge table (kb_knowledge). The name varies by release, so check the source attached to your Knowledge search profile.
• Look at the indexed source status and the last indexed time.
• Open AI Search > Indexed Sources and check the document count and any failed or pending documents.
• Try a manual re-index of the affected articles, or a full re-index of the Knowledge source if the count looks low.
• Search again on the Knowledge Homepage using the article number.

If you are on the legacy search engine instead of AI Search, the same idea applies to the text index.

• Go to System Definition > Tables and open Knowledge (kb_knowledge).
• Check that the table is still marked for text indexing in the table’s Controls.
• Open the Text Index Events related list (ts_index_event) and look for stuck or errored events.
• Open Text Search > Indexed Sources and check the knowledge source for errors.

Second, rule out how the articles were published, since random misses often come from this.

• Compare an affected article with a working one. Look at Knowledge Base, Category, Language, Workflow state, Active, and Valid to.
• Check whether the affected articles were imported, bulk updated, or migrated by a script or data load. Direct database updates and some imports skip the index events.
• Check whether the article language matches the language of the user searching. Search profiles often filter by language.
• Confirm the Knowledge Base selected on the search page is the one the article belongs to.

Third, check the access rules, because search quietly hides what the user cannot read.

• Open Knowledge > Knowledge Bases and review the Can Read and Can’t Read user criteria on each knowledge base.
• Look at the Can Read and Can’t Read criteria on the affected articles themselves, since some releases support article-level criteria.
• Test as the same user with impersonation, and then as an admin, to see if results change.
• Note that global search for an exact number can behave differently from relevance-based search, so it may still find articles the other searches skip.

Fourth, check the search configuration itself.

• Open AI Search > Search Profiles and review the profiles used by the Knowledge Homepage and the case form’s Related Search Results.
• Check the search sources, filters, and result limits on each profile. A condition on the source can exclude only some articles.
• Check Knowledge Management Properties (kb_knowledge.properties) for properties that control which article states and dates are shown.
• Review any custom ACLs or business rules on kb_knowledge that might filter records at query time.

As a safe way to prove the cause, take one affected article and make a small edit, such as touching the short description, then save and search again. If it now appears, the index was out of date and a re-index will fix the rest. Run any full re-index in a sub-production instance first, since it can be heavy on a busy Production instance, and do it outside business hours.

If none of this helps, it’s worth opening a case with ServiceNow Support with the article sys_ids, since search index problems can also come from known issues in a specific patch.

A few details would let me be more precise. Which release and patch are you on, and is the Knowledge Homepage using AI Search or the legacy search? Also, do the missing articles share something, like a knowledge base, author, import source, or creation date?