I have a three-server Solr setup. Server 1 adds new data to the solr index every few minutes. Servers 2 & 3 use the index on server 1 using a file sharing configuration. To get the index updates, servers 2 & 3 run the simple command below every few minutes.
java -jar /path/to/solr/post/file/post.jar
The whole system has been running fine for a few months, but now that our index is growing in size, servers 2 & 3 become unresponsive during or after running the post.jar.
When architecting my solution, it made sense to have one index shared across servers. This allowed for one master data acquirer server and multiple query processing servers. Am I looking at this wrong? Do I need a unique (yet identical) index on each of my query processing Solr servers?