11.2 was a failure but 11.1 is stable (at least for me). As far as stale shares go, you can reduce them by checking for work faster than what is set by default. Add the switch -etht which by default is set to 200ms and reduce the time your rigs wait to check for shares. I have mine set to check every 20ms with this switch -etht 20

In my experience this keeps stale shares under 5% (Mine is currently at 2% using 11.1). But you can do some pings to make sure that your problem isn't latency between your router and the pool or even latency on your own network as well.