Description of problem:
Finally I have a reproducer issue that was hunting me last months. I'm glad it is not that serious and looks like it has been in openssh-server package forever. I can reproduce it on both EL5 and EL6.

The reproducer is very simple:
# service sshd restart
Stopping sshd: [  OK  ]
Starting sshd: [  OK  ]
# netstat -putna | grep sshd
tcp        0      0 0.0.0.0:22                  0.0.0.0:*                   LISTEN      19565/sshd          
tcp        0      0 :::22                       :::*                        LISTEN      19565/sshd          
# /usr/sbin/sshd -p 133
# netstat -putna | grep sshd
tcp        0      0 0.0.0.0:133                 0.0.0.0:*                   LISTEN      19571/sshd          
tcp        0      0 0.0.0.0:22                  0.0.0.0:*                   LISTEN      19565/sshd          
tcp        0      0 :::133                      :::*                        LISTEN      19571/sshd          
tcp        0      0 :::22                       :::*                        LISTEN      19565/sshd          
# service sshd restart
Stopping sshd: [  OK  ]
Starting sshd: [  OK  ]
# netstat -putna | grep sshd
tcp        0      0 0.0.0.0:22                  0.0.0.0:*                   LISTEN      19565/sshd          
tcp        0      0 :::22                       :::*                        LISTEN      19565/sshd          
# service sshd restart
Stopping sshd: [FAILED]
Starting sshd: [  OK  ]
# netstat -putna | grep sshd
tcp        0      0 0.0.0.0:22                  0.0.0.0:*                   LISTEN      19565/sshd          
tcp        0      0 :::22                       :::*                        LISTEN      19565/sshd        

As you can see the restarting the sshd service when also sshd on port 133 is running kills this instance and the previous instance is still running. The next restarts fail.

Version-Release number of selected component (if applicable):
openssh-5.3p1-52.el6

How reproducible:
100%

Steps to Reproduce:
see description
  
Actual results:
service sshd restart kills incorrect sshd instance and then begins to fail.

Expected results:
service sshd restart restarts correct sshd instance and doesn't fail if original instance still running. This makes impossible to restart ssh daemon with new configuration for example.