[Pacemaker] Master/Slave Confusion

[Erich Weiler]

Hi All,

Forgive a probably elementary question, but I'm new to Pacemaker and am 
not clear on exactly how a Master/Slave relationships exist.  Here's my 
confusion:

My initial though was that with a master/slave service, the service is 
started on both nodes (assuming I have 2 nodes).  But, I want my virtual 
IP to be tied to the master node only.  In my case, I have 2 nodes that 
have LDAP servers on them, one is the master and the other is the 
replicated slave.  Most of the time, I'd like the virtual IP to be tied 
to the master LDAP node because that one has write access for folks to 
change LDAP attributes like passwords and what not, but the slave is 
read only.  So I want the master to have the floating IP if it is alive 
at all times.  If the master one dies, I'd like the virtual IP to float 
over to the slave so folks could still access LDAP that way.  Then if 
the master comes back, it immediately grabs the IP back.

However, it seems that when LDAP starts, the IP needs to be live on each 
node for the LDAP server to bind on that IP.  Is that how the 
master/slave setup works in pacemaker?  Does it use iptables or 
something to block the IP on the non-active node when the slave doesn't 
have the floating IP?

Or maybe I'm completely misunderstanding how that works...?

Thanks for any insight...

-erich