Also if you are using Block Join indexes, delete by query is a very bad idea unless the query is very carefully crafted to either ensure all children for a parent are deleted, or to avoid deleting any document that has children (it can be done, I had to do it a couple years ago, but it's really easy to get wrong). Block Join documents are better deleted by id so that the children can be (automatically) deleted at the same time.
https://solr.apache.org/guide/solr/latest/indexing-guide/indexing-nested-documents.html#maintaining-integrity-with-updates-and-deletes On Mon, Aug 26, 2024 at 7:05 PM Carlos Ugarte <cuga...@gmail.com> wrote: > If you are adding new documents or modifying existing documents while > issuing delete-by-query requests, you may see performance degradation on > non-leader replicas. Leaders apply updates in a specific order. In order > to guarantee they use the same order, non-leaders (certainly when using NRT > replication, unsure about TLOG replicas) might have to replay the > delete-by-query operation multiple times. You should see something along > the lines of "Reordered DBQs detected" in your log file. > > > > On Mon, Aug 26, 2024 at 12:41 PM mtn search <search...@gmail.com> wrote: > > > Hello, > > > > I believe I have heard there are some issues with Delete by Query for > > SolrCloud. I did not pick up on what the symptoms were. If this is > true, > > what are the issues? > > > > Thanks, > > Matt > > > -- http://www.needhamsoftware.com (work) https://a.co/d/b2sZLD9 (my fantasy fiction book)
