codec exception not caught causes inputworker to stop running

[frapex]

If a codec exception is not caught within the codec itself, it is caught by the inputworker.
This causes the inputworker to raise an exception and stop running completely.

It can be triggered by using logstash-codec-collectd under high load. Exceptions are raised, but not caught.

I have a patch that solves the problem. The exception is caught and handled just like when an TCP listener dies.

[frapex]

PATCH:

--- a/udp.rb      2015-08-01 17:49:33.678261866 +0200
+++ b/udp.rb      2015-08-01 19:06:07.369765789 +0200
@@ -99,7 +99,9 @@
         end
       end
     rescue => e
-      @logger.error("Exception in inputworker", "exception" => e, "backtrace" => e.backtrace)
+      @logger.warn("Exception in inputworker", "exception" => e, "backtrace" => e.backtrace)
+      sleep(5)
+      retry
     end
   end # def inputworker