Building a web application distributed over several nodes in a cluster, I want to explore wether we can do content search (not full-text search but data queries) using Lucene.net.
From what I can see, constructing an Akka.net-based cluster of Actors for indexing/searching may not be so difficult ... But achieving some of the functionality used in Elasticsearch would also be nice, particularly moving shards between nodes, replicating shards depending on topology ... etc.
If we post an "index this content" message to one of the index/search nodes, and that node goes down, then that cache is lost. On the other hand, if one of the nodes gets the message while a node already having indexed the content comes back, it will be duplicated.
So the Lucene.net indices need to be curated continously, I think. But how?