I have also looked at this and would like to gain more perspective on it. Another related wrinkle to watch out for: I'm pretty sure that Solr replication doesn't re-use core readers over existing segments. I noticed this (along with my colleague Patson) when looking at unexpectedly high CPU usage on PULL replicas, iirc.
On Fri, Aug 23, 2024 at 3:25 PM David Smiley <dsmi...@apache.org> wrote: > > Can anyone explain how the ReplicationHandler/IndexFetcher manipulates > index files (adds files/segments, removes some too) without > interfering with the live index? I see IndexFetcher can call > SolrCoreState.closeIndexWriter. I've also seen evidence in code > nearby that suggests another data dir may get written to, so maybe > there's some kind of swap. I'd like a big picture explanation of > what's going on. It's confusing stuff to stare at from a code detail > point of view, and an absence of documentation. > > ~ David Smiley > Apache Lucene/Solr Search Developer > http://www.linkedin.com/in/davidwsmiley > > --------------------------------------------------------------------- > To unsubscribe, e-mail: dev-unsubscr...@solr.apache.org > For additional commands, e-mail: dev-h...@solr.apache.org > --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@solr.apache.org For additional commands, e-mail: dev-h...@solr.apache.org
