During my tests with HCL Domino 12 Beta2 I ran into a couple of issues where the Active Directory Password Sync did not work as expected. The problems are not caused by the beta code. However, I am writing down my observations here, since the errors may also appear later in a productive environment. Maybe my lines will save one or the other a long search.
Request Creator replicates with multiple servers
In the console log of a request creator you see replcation events with multiple servers.
There are 2 databases that are replicated on the request creator, Domino Directory (names.nsf) and Directory Assistance (da.nsf). Both databases are replicated with their admin servers to start every config update cycle. Important to know is that the admin server for each database is determined seperately separately. In my test environment, the admin server for da.nsf was different than the one for names.nsf resulting in multiple replication events.
Wrong configuration document
If you have more than 1 Domain Controller that syncs users passwords with your Domino environment, you must install and configure a request creator on each Domain Controller. In addition, you need a configuration document for each of your request creators. Each request creator should have the same configuration settings. Best practice would therefore be a single configuration document that applies to a group of servers.
During my tests I could observe that instead of the assigned configuration document the default configuration document was used.
There was no Statistics output in the console log and I could also not see any DEBUG log output. Also the configuration was not refreshed in the configured interval. I discussed this with HCL Development and it turnd out that the configuration document was not used due to a problem with the view index in names.nsf on the request creator.
I needed to stop the sync process on the request creator, open names.nsf in the Notes Client. After restarting the AD Domain Controller machine, the correct configuration document was used.
Most likely, this issue occurs when you use a configuration document for a single request creator and later switch to a group based configuration document. I would suggest to use a group based configuration document from the start on.
Is Active Directory Password Sync cluster ready?
You can designate as many request processors as you want, and it’s a good way to provide failover.
The decision as to where the AD Domain Controller sends a request is admittedly not overly sophisticated. It works sequentially through an in-memory list it holds of all servers which are designated as request processors, until it finds an available one. Have a look at the ($PWSProcessors) view in the directory. (Mark De Lellis, HCL Development)
This is where the list is built from, using all servers with Role “2” (Request Processor), in the order they appear in the view.