I've been working with Cassandra for some time and get this error intermittently:
ERROR [TCP Selector Manager] 2010-09-23 08:42:18,138 service.CassandraDaemon Fatal exception in thread Thread[TCP Selector Manager,5,main]
Unfortunately when this happens the socket is left open so clients connect, but are never serviced. When this happens I am forced to restart the node.
We're running version 0.5.1. I know there are much newer versions, but we have been frozen on this for various reasons.
I am wondering if this is/was a known issue with this version, or if I am doing something wrong. My searches thus far have yielded nothing on the matter.
Any insights would be appreciated.