linslee75 commented on PR #5194: URL: https://github.com/apache/solr/pull/5194#issuecomment-6435529283
> FWIW I'm pretty sympathetic to doing an automatic softCommit as the user experience is then closer to the current state. Shouldn't be hard. Thanks, David. A few thoughts for the discussion: SOLR-18350 makes each request use the schema snapshot associated with its searcher. This keeps schema reads stable for in-flight requests, but it does not require Solr to reopen a searcher immediately after every schema update. In SolrCloud, Schemaless auto-add publishes the updated schema without immediately reopening searchers. This PR follows the same model for additive Schema API changes. Until a replica opens a searcher with the new schema, a query or facet that references the new field can fail with undefined field. The difference from current Schema API behavior is intentional: avoiding a core reload prevents a long-running merge from blocking indexing. My preference is to keep this PR aligned with Schemaless and document the visibility window. If we decide that new fields should become queryable immediately, I think we should discuss a mechanism that applies consistently to both Schema API and Schemaless, rather than giving them different reopen behavior. David’s automatic softCommit idea may be worth considering as part of that broader discussion. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
