Recently I've been using cqrlog on several systems; one right next to the system hosting the mysqlDB, and a couple others that are far remote and accessing over an ssh tunnel.
 
I've noticed that if I have the cluster active, very frequently I'll recieve the mysqlDB error message where the remote is unreachable (though, it still is, as there is a local mysql monotor window running over the same path..).
 
This will happen with either local/remote instances of cqrlog and it's really much much more frequent if I've got the cluster on.  I suppose there could be enough cluster activity as to cause the msql-connection verifier to miss an attempt; how many times must the mysql command fail before it declares the remote gone and tosses up an error (I have not trolled the source..)--would it be possible to configure both the frequency and number of failures at some point?
I guess I should also ask: what additional info is helpful?  Running debug with the cluster active produces a lot of output, IIRC, the cluster stays connected, so if I'm not right there when it loses the sql db, I haven't captured the other messages.  I'll redir to a file and share that once I'm in the shack. 
Otherwise, no real issues at all; it works wonderfully and I enjoy using it.