for power contrl commands, every cec fork one process. So if rpower multiple nodes on several cec, say 20 cec, there are more than 20 connection request to hmc at the same time. This make sshd on hmc failed to respond and refuse the connections request. Add some sleep between every fork per cec. This will not reduce the performance too much, since only sleeps between cecs.

git-svn-id: https://svn.code.sf.net/p/xcat/code/xcat-core/trunk@4153 8638fb3e-16cb-4fca-ae20-7b5d299a9bcd

xCAT-server/lib/perl/xCAT/PPC.pm

@@ -391,6 +391,7 @@ sub process_command {
             }
             $hw = @$_[0];
 
+            Time::HiRes::sleep(0.2);
             my ($pipe) = fork_cmd( @$_[0], @$_[1], $request );
             if ( $pipe ) {
                 $fds->add( $pipe );