Hello 
I am trying to reindex about 3mil documents (~50gb) trough an ingestion pipeline.
POST _reindex?&wait_for_completion=false
{
  "source": {
    "size": 1, 
    "index": "src"
  },
  "dest": {
    "index": "dst"
    "pipeline": "reindex_pipeline"
  }
}
As you can see, I already tried lower batch sizes (even down to one)
but as the pipeline is doing some heavy computations, it seems at least one document is causing a scroll context to become invalid.
Is there any other way (besides lowering batch size) to fight this?
org.elasticsearch.action.search.SearchPhaseExecutionException: all shards failed
...
Caused by: org.elasticsearch.transport.RemoteTransportException: [node name][127.0.0.1:9300][indices:data/read/search[phase/query+fetch/scroll]]
...
Caused by: org.elasticsearch.search.SearchContextMissingException: No search context found for id [...]
(Single Node running ES6.7)