Open liveforeverx opened 9 years ago
I think for a moment why we don't use TCP keep-alive?
no objections to tcp-keep-alive, but it is out-of-scope here. the watchdog has to do a higher level functional check and (more or less brutally) restart the connection (or even the full application).
using tcp-keep-alive to prevent that the watchdog will ever bark is okay (but out of scope here).
should we consider ongoing communication as watchdog reset. I.e. CAPWAP Echorequests are not used if there is ongoing communication between the nodes. Or do we this as optimisation.
What are the defaults for watch dog? How often send? How many PONG are allowed to fail, is the good and failure sate known to the application? how ist the sate reported? ist there a defined (log) message if watchdog fails?
If no PONG answer to PING, the error should be written and connection should reconnect.