2021-10-25 Capacity Planning Meeting

Charlotte Whitt

On Friday Michal Kuklisis planning to document what our options are and pass them on to Charlotte.

Matt Connolly, a member of Cornell's usability group, will be talking to folks at Cornell to learn more details about

Two approaches:

1) Steps to move back to the previous search PostgreSQL implementation:

Option a: Rollback changes which introduced Elastic Search and facets:

• • • • Rollback new search box and facets (they are currently coupled together)
      • Rollback all changes related to facets which were done after the integration was in place
    • Time estimate: couple days maybe
• Pros: Same UX 
• Cons: Bad performance

Option b:  Switch search endpoint to use mod-search

• • • • Switch the search endpoint to use Elastic Search instead of mod-inventory (PostgreSQL)
      • Make sure all existing filters work with the new search endpoint (we are not sure if all existing filters are mapped correctly in mod-search)
    • Time estimate: couple days maybe
• Pros: Keep ES performance 
• Cons: Lose some of the capabilities (facets)/Boolean/Exact Phrase possibly add facets in a Kiwi patch release. 

c: Integrate facets with the dropdown search input 

• • • Time estimate: This is the biggest unknown because the UI developers would have to break up facets from the current search box and make sure they work with the previous search box
  • Pros:
  • Cons: TBD 

2) Keep the Elastic Search work (Inventory: UIIN-1566, UIIN-1567, and the new work in the instance look up component) but in the interface we add an option to bring up a drop down menu listing all search options for each segment Instance, Holdings, and Item, and we add a <i> icon with link to more information on how to search using the Elasticsearch (this is easily fixed with a link to the tips and tricks page on the FOLIO Wiki: Search - using Elasticsearch (or OpenSearch).

Go with approach 2 will require a dialogue with Cornell (management of customer expectations) explaining all the advantages we achieve when Inventory is using Elasticsearch as search engine (fast performance, accurate result count, facets on the filtering, possibility to do searches using boolean operators).

Notice that during the past 1-1.5 months Cornell has filed several support tickets about searches taking too long, searches not returning results, searches timing out - when using postgreSQL

Oleksii Petrenko

• Juniper HF#3 (date TBD) 
  • Need resolution for MODDATAIMP-569 - Getting issue details... STATUS (only happens with HF#3 release) 
• Kiwi Bugfixes
  • 26 P1 issues reported so far 
• Kiwi Bugfest stats
  • 54% passed
  • 3% Not claimed
  • 39% Claimed