Hi, I realize a weird behaviour I can't explain (so far we are still running in master/slave mode) Requesting the collection I see logs randomly against the two available shards "green_shard1_replica_n1" and "green_shard2_replica_n2" 2018-06-21 15:35:40.970 INFO (qtp1873653341-17) [c:green s:shard2 r:core_node4 x:green_shard2_replica_n2] o.a.s.c.S.Request [green_shard2_replica_n2] webapp=/solr path=/select params={q=id:"508364329"&qt=morelikethis&json.nl=arrarr&fl=*,score&rows=40&rows=5&wt=json} status=0 QTime=1 2018-06-21 15:36:05.679 INFO (qtp1873653341-70) [c:green s:shard1 r:core_node3 x:green_shard1_replica_n1] o.a.s.c.S.Request [green_shard1_replica_n1] webapp=/solr path=/select params={q=id:"508364329"&qt=morelikethis&json.nl=arrarr&fl=*,score&rows=40&rows=5&wt=json} status=0 QTime=5 2018-06-21 15:36:11.022 INFO (qtp1873653341-17) [c:green s:shard1 r:core_node3 x:green_shard1_replica_n1] o.a.s.c.S.Request [green_shard1_replica_n1] webapp=/solr path=/select params={q=id:"508364329"&qt=morelikethis&json.nl=arrarr&fl=*,score&rows=40&rows=5&wt=json} status=0 QTime=2 2018-06-21 15:36:17.031 INFO (qtp1873653341-70) [c:green s:shard1 r:core_node3 x:green_shard1_replica_n1] o.a.s.c.S.Request [green_shard1_replica_n1] webapp=/solr path=/select params={q=id:"508364329"&qt=morelikethis&json.nl=arrarr&fl=*,score&rows=40&rows=5&wt=json} status=0 QTime=2 2018-06-21 15:36:21.800 INFO (qtp1873653341-17) [c:green s:shard1 r:core_node3 x:green_shard1_replica_n1] o.a.s.c.S.Request [green_shard1_replica_n1] webapp=/solr path=/select params={q=id:"508364329"&qt=morelikethis&json.nl=arrarr&fl=*,score&rows=40&rows=5&wt=json} status=0 QTime=2 2018-06-21 15:36:26.668 INFO (qtp1873653341-70) [c:green s:shard2 r:core_node4 x:green_shard2_replica_n2] o.a.s.c.S.Request [green_shard2_replica_n2] webapp=/solr path=/select params={q=id:"508364329"&qt=morelikethis&json.nl=arrarr&fl=*,score&rows=40&rows=5&wt=json} status=0 QTime=0 In case the running node selects "green_shard1_replica_n1" I'm getting results but I will always get no morelikethis suggestions in case the other shard is selected. The number of shards for the test collection is 2, replication factor 1 (it's only a test index for a small number of docs) but the behavior is the same for a huge collection (30 Mio docs) What might be the reason for this behavior? - thanks for any hints Günter