On 24/08/2024 22:43, Shawn Heisey wrote: > The ConfigSet API is for maintaining configsets. There is a way to use > configsets in standalone mode, but SolrCloud is preferred. > > A configset consists of solrconfig.xml, managed-schema.xml (or > schema.xml if using the classic schema factory), and any other files > referenced by those two configs. You can't just upload the schema, you > must also have the solrconfig. > > https://solr.apache.org/guide/solr/latest/configuration-guide/configsets-api.html > > Each configset stored in ZK has a name, and that name is used when > creating a collection. An existing collection can be altered to use a > different config name, and reloading the collection will activate the > change.
Thanks. So from what I can make out, it looks like the application would be given the name of an existing ConfigSet. It would then download that ConfigSet, replace the schema with its own, and then create a new ConfigSet based on the old one with the new schema. It would then create a collection referencing that new ConfigSet. Does this sound correct? In our case, we'd be treating schemas as immutable once created. If we need to change the schema, we'll create a new collection with the changed schema and reindex with the contents of the old one. > To answer your question from another thread, SolrCloud is the preferred > way to run Solr. It would not surprise me if an upcoming major version > of Solr will ONLY run in cloud mode. Some of Solr's more advanced > features will only work in cloud mode. Got it. -- Mark Raynsford | https://www.io7m.com
