Send less initial pings for healthcheck

Previously we would end up sending too many pings to a node to find out
if it's healthy and in the meantime consider it as reachable.

We would send 120 pings, waiting 60 seconds between each ping for the
initial healthcheck. In the meantime, raiden would consider the node
healthy and try to route transfers through it.

Now for that special initial healthcheck we make sure to use a lot less
pings and immediately consider it unhealthy if this does not work.

Author: LefterisJP

raiden/settings.py

@@ -22,6 +22,6 @@
 DEFAULT_INITIAL_CHANNEL_TARGET = 3
 DEFAULT_WAIT_FOR_SETTLE = True
 
-DEFAULT_NAT_KEEPALIVE_RETRIES = 5
-DEFAULT_NAT_KEEPALIVE_TIMEOUT = 30
+DEFAULT_NAT_KEEPALIVE_RETRIES = 2
+DEFAULT_NAT_KEEPALIVE_TIMEOUT = 10
 DEFAULT_NAT_INVITATION_TIMEOUT = 180

raiden/ui/cli.py

@@ -161,7 +161,7 @@ def convert(self, value, param, ctx):
             'Max time in seconds for which an address can send no packets and '
             'still be considered healthy.'
         ),
-        default=120,
+        default=30,
         type=int,
     ),
     click.option(
@@ -284,9 +284,9 @@ def app(address,
         config['external_ip'] = listen_host
         config['external_port'] = listen_port
 
-    retries = max_unresponsive_time / DEFAULT_NAT_KEEPALIVE_RETRIES
-    config['protocol']['nat_keepalive_retries'] = retries
-    config['protocol']['nat_keepalive_timeout'] = send_ping_time
+    config['protocol']['nat_keepalive_retries'] = DEFAULT_NAT_KEEPALIVE_RETRIES
+    timeout = max_unresponsive_time / DEFAULT_NAT_KEEPALIVE_RETRIES
+    config['protocol']['nat_keepalive_timeout'] = timeout
 
     address_hex = address_encoder(address) if address else None
     address_hex, privatekey_bin = prompt_account(address_hex, keystore_path, password_file)