I had a single-node instance (0.5) that has some data (1 SCF has 1 key and about 500k columns). I upgraded the node to 0.6 release yesterday and added another 0.6 node to try out clustering. The two nodes seem to work fine (discovered each other), but reading the data on either node only returns a tiny fraction (using pycassa). I had set the replication factor to be 2. I shut down the two nodes and reverted the first node back to 0.5, and can read the data out fine. Then I tried to read data from that node with 0.6, and it is also fine, so it is not pycassa's problem.
I wonder what is the right way to do clustering in my case and how long "eventual" consistency will take. I want to use one node to collect data and the other node for analysis. Thanks for your time.
