Hi, I trying to find a way to deal with adding nodes in runtime and still keeps the solr shards evenly divided. let's say I have 2 nodes with 2 shard (each node contain 1 shard) and sometime later the shards I have are fulling their entire appropriate Solr nodes.
I will want to add a new node to help with the memory load, right? (say with the same CPU, memory.. as the 2 existing nodes). But then how can I split the shards? (reminding that I have 2 nodes, each node has 1 shard ==> 2 shards total). If I will split the 2 shards I will get 4 shards and that cannot be divided evenly withing 3 nodes. Other option is to start in the first place with a biggest shard number per node. but since I don't know how many nodes I eventually will need to add, I cannot know at the starting point, what will be this magic number..... The 2 nodes case above is just an example. I need some generic way on how to deal with this uncertainty issue regarding the re distribution of shards when adding additional node mid-way through. Any idea? Thanks -- Sent from: http://lucene.472066.n3.nabble.com/Solr-User-f472068.html
