fix(remoteshell): wait for the ssh daemon to end before a new one starts

A kill only asks the kernel to end a process, so it returns before the process
is gone. The fallback started a new daemon on the next line, which could find
the port still held by the old one, fail to start, and leave the node with no
daemon at all.

Wait for the processes to end, for up to ten seconds, before the new daemon
starts. The test that a process is still there sends no signal.

xCAT/postscripts/remoteshell

@@ -648,7 +648,19 @@ fi
 #try to kill the process and start
 if [ "$?" != "0" ];then
    PIDLIST=`ps aux | grep -v grep | grep "/usr/sbin/sshd"|awk -F" " '{print $2}'|xargs`
-   [ -n "$PIDLIST" ] && kill -9 $PIDLIST
+   if [ -n "$PIDLIST" ]; then
+      kill -9 $PIDLIST
+      waited=0
+      while [ $waited -lt 10 ]; do
+         alive=0
+         for pid in $PIDLIST; do
+            if kill -0 $pid 2>/dev/null; then alive=1; fi
+         done
+         if [ $alive -eq 0 ]; then break; fi
+         sleep 1
+         waited=`expr $waited + 1`
+      done
+   fi
    /usr/sbin/sshd
 fi
 kill -9 $CREDPID