[MacTUG] Open Directory replication and failover?

[Ian Turner]

of course we are running the full golden triangle

we had an incident the other day where our "master" OD got accidentally 
unplugged by other work in the server room

Suddenly, none of the clients were getting their OD Managed settings

There was an indication that replication was having an issue, so I was 
chasing that
But then it appeared replication WAS working

so I retested failover - still "no go"

I have ended up doing the moderately drastic step of "demoting" the 
replica to standalone, then "re-replica-ing" it

Now replication is demonstrably working

but failover is NOT

Lots of googling made me try an experiment:
- on one client, add both the Master and the Replica in the ldap plugin 
under Directory Service Utility
** That worked for failover; and in fact, I had found one user article, 
supported by what I found in the .plist - that said "always link to the 
replica" because failover "up" works, but "down" doesn't in Snow Leopard!

Cannot find anywhere in the Apple Documentation or KB that supports this.
Hard for me to believe the Master doesn't publish a list of replicas to 
the clients as a much more maintainable/scalable setup.  The 
documentation does say the clients will "find" the "nearest" replica 
when they start up - ie the one that responds fastest on their network

*********************

anyone have any ideas?
any similar experience?

Has anyone gotten failover to work on SL?

(silly me, based on how good it was in Tiger and Leopard, once I had 
verified replication was working in the April rush to get the SL 
migration done, "failover" was not properly tested!)

I'm wondering if "Apple" assumes we are running Apple DNS, and providing 
OD servers through that? - of course we use DNS from our AD; being 
"golden triangle-ized"