Yes, this is not ideal. It means that the worker collection needs to have N shards rather than N replicas. Changing this should not be difficult if you'd like to provide a patch.
Joel Bernstein http://joelsolr.blogspot.com/ On Thu, Jan 21, 2021 at 8:00 AM ufuk yılmaz <uyil...@vivaldi.net.invalid> wrote: > Looked at the source code of the parallel stream and it seems I need equal > number of SHARDS and workers count parameter. I thought I needed as many > replicas, it was shards. > Maybe helps someone. > > Sent from Mail for Windows 10 > > From: ufuk yılmaz > Sent: 21 January 2021 11:16 > To: solr-user@lucene.apache.org > Subject: Parallel streaming expression java.lang.IndexOutOfBoundsException > > Hello all, > > > https://lucene.apache.org/solr/guide/8_4/stream-decorator-reference.html#parallel > > I’m sending the same query in the docs, (just collection names changed) to > my Solr but always getting the exception: > > { > "result-set":{ > "docs":[{ > "EXCEPTION":"java.lang.IndexOutOfBoundsException: Index 1 out of > bounds for length 1", > "EOF":true, > "RESPONSE_TIME":93}]}} > > My query: > > null( > parallel( > WorkerCollection, > rollup( > search( > colA, > q="username: c*", > qt="/export", > fl="timestamp,user.id", > sort="user.id asc", > partitionKeys="user.id" > ), > over="user.id",count(*) > ), > workers="4", > sort="timestamp asc" > ) > ) > > WorkerCollection has 1 shard and 4 replicas on 4 different machines. I > double-triple checked for a silly syntax mistake but there’s none that I > can see. I tried doing this a few months ago with no success, getting the > same exception again. colA is on different machines from WorkerCollection. > user.id is a string field, timestamp is long. > > What am I missing? > > > Sent from Mail for Windows 10 > > >
