Hello, I barely understand what's the core issue here. I want to elaborate regarding end sentences about sql queries. I worked on ETL for structured model before. I found that external merge join (in SQL terms) works pretty awesome for indexers like these. We implemented it as a zipper join in DIH. https://solr.apache.org/guide/6_6/uploading-structured-data-store-data-with-the-data-import-handler.html#entity-processors https://issues.apache.org/jira/browse/SOLR-4799? I don't pitch migration to DIH, I propose to consider external sql join to avoid N+1 pitfalls.
Wish you the best. On Fri, Dec 23, 2022 at 4:56 AM dmitri maziuk <dmitri.maz...@gmail.com> wrote: > Hi all, > > has anyone ever gotten child documents to work in 8.11? I'm trying to > add them in a separate batch w/ setting _root_=$ID_OF_PARENT, > _nest_path_=$SOMETHING and so far I'm unable to get anything from The > Fine Manual's examples > (https://solr.apache.org/guide/8_11/searching-nested-documents.html) -- > at best don't do anything useful. > > At worst, now that I tried making _root_ indexed and stored, the parent > documents actually have _root_:$ID_OF_PARENT??? (There are no > copyfields.) Does this make any sense? > > Any hints? > > I'll try to rewrite the importer script to insert them as nested > documents instead of separately, and see if that makes any difference. > But even if that works as advertised, it's not really a usable option > because of pymssql query overhead: with "only" 2 sub-queries per parent > document I expect "only" a 3x overall slowdown. > > Dima > -- Sincerely yours Mikhail Khludnev
