Hi, Can you provide more information : Solr version, how are you indexing (DIH, threading, ...), more details in Solr logs ?
Did you analyse JVM Gc logs ? Regards Dominique Le ven. 28 août 2020 à 22:53, amit3281 <amit3...@gmail.com> a écrit : > Hi, > > > > I am using Solr on EXT4 partition and have TLOG replicas in my collection. > I > > am using 2 Solr nodes to utilize 2 disk (for getting IOPS) for same > > collection. My collection has 150 shards. Each shard size is ~9GB and > > 48Million docs per shard. > > > > My shards frequently goes into recovery with error > > *o.a.s.h.RequestHandlerBase java.io.IOException: > > java.util.concurrent.TimeoutException: Idle timeout expired*. > > > > I am doing 90% times overwrites as getting few statistics update for same > > data. > > > > 1. Is this because of frequent updates? > > 2. Is this because of Huge number of shards? > > 3. Is this because of TLOG and big segments are getting transferred? > > > > Commit time is 15sec and I am trying to add 10k docs per second. > > > > > > > > > > > > -- > > Sent from: https://lucene.472066.n3.nabble.com/Solr-User-f472068.html > >
