Fixing when mesos-agent is unreachable for log fetching.

If we couldn't get the logs from the agent, it would block inside the
case <-timer select making the main thread not be able to handle
messages sent to StatusUpdate nor ResourceOffers resulting in consuming
the entire cluster :(. This is a stop gap until I figure out how to
properly fix this.

main.go

@@ -232,16 +232,19 @@ func printLogs() {
 			}
 		case <-timer:
 			if readableStatus != nil {
-				if finished {
-					time.Sleep(3 * time.Second)
-				}
-				x := printLog(readableStatus, oout, os.Stdout)
-				y := printLog(readableStatus, oerr, os.Stderr)
-				if finished && x == 0 && y == 0 {
-					return
-				}
-				oout += x
-				oerr += y
+				go func() {
+					if finished {
+						time.Sleep(3 * time.Second)
+					}
+					x := printLog(readableStatus, oout, os.Stdout)
+					y := printLog(readableStatus, oerr, os.Stderr)
+					if finished && x == 0 && y == 0 {
+						log.V(1).Infof("framework terminating")
+						os.Exit(exitStatus)
+					}
+					oout += x
+					oerr += y
+				}()
 			}
 		}
 	}
@@ -378,6 +381,4 @@ func main() {
 	}()
 
 	printLogs()
-	log.V(1).Infof("framework terminating")
-	os.Exit(exitStatus)
 }