I am not sure if this is the best approach to this problem but I was curious if a single solr server could be both a master and a slave without causing index corruption? It seems that you could setup multiple replication handlers in the SOLR config, /replication /replication2 and have one be master and another be a slave syncing from another server. Here's why: 1) I want to build an index using data stored in our own local datacenter generated using M/R and our MySQL DB 2) This index would be synced with a Solr Master sitting in EC2 3) Series of EC2 solr slaves replicate from EC2 Solr master for scaling purposes.
I figure this would save costs (both time and money) over having all EC2 slaves slave from our datacenter. The index isn't that big but I figure transferring it once would be best. I was going to setup my local datacenter process hourly and let it sync accordingly. Any pitfalls to this? Thanks Amit
