Problem:
I have a python node A running on a Ubuntu16 machine which every 10 seconds pings another node B (a service server) to see if it's still alive. B is running on an Android tablet using rosjava. Whenever the tablet loses network connection, A obviously fails to ping B
Then, I programmed A to reinitialize itself and wait for the service server to come online again. Once that happens, A successfully connects to B again but then pinging B out of that python script always fails ("Unknown network error contacting node: timed out").
This makes A to reinitialize itself again and the procedure above restarts. B is available though because otherwise A would not be able to successfully connect to the service server.
The only work-around to avoid this weird behaviour right now is to restart A.

If I make use of rosnode ping B it also works perfectly.
I had a look at this issue but it only refers to a situation where a node is stopped properly by the user, not because of network errors.