A bunch of stuff is now written to Tracing, and there are no logging trace calls any more. If a node is having issues on the read/write path and tracing can't actually save data, there is no way to see through logging what is going on. Would be nice if we made all Tracing messages also go out at logger.trace so that you could enable that to debug stuff.
Being able to change the RF of the system_traces KS might also help here, but there would still be classes of problems that it would be good to have the logging there for. Added
CASSANDRA-6016 for that.