Thanks to you both! Due to the nature of our system, a very large legacy Solr 6 M/S (Leader/Follower) , our plan is to update schema on the Follower, and then the Leader. Once complete we will start indexing docs to the Leader with data in new schema fields. I have a question about replication.
Any concerns about replicating the index from a Leader with the old schema (no new fields) to a Follower with the new schema (2 new fields added)? Under this condition the index will not contain any docs with the new fields as the Leader schema does not. Also, when applying the schema change to the follower, any concerns about copying the new schema (classic) to the config set on the VM, restarting Solr (without stopping polling on the follower)? This seems to be the most efficient approach. It seems like the risk is small if replication occurred during this event. It would fail and try again on the next poll. We are testing these scenarios, but helpful to get the community's input! Thanks, Matt On Thu, Aug 14, 2025 at 7:58 PM ufuk yılmaz <[email protected]> wrote: > If I remember correctly the newly added fields shouldn’t be > “required=true” right? Haven’t checked it myself though > > — > > > On Aug 15, 2025, at 7:13, Chris Hostetter <[email protected]> > wrote: > > > > > > : new fields to our schema and not immediately re-index documents. So > early > > : on only new docs will be populated with the new fields. > > : > > : Any concerns with this approach? In some documentation I seen > direction > > : that a re-index must happen on any schema change. > > > > As a good rule of thumb, and schema *change* should be handled by > > re-indexing, but you are correct that in most cases *adding* new fields > > (and then gradually populating them overtime) is generally fine > > > > > > -Hoss > > http://www.lucidworks.com/ > >
