Question : Replica problem - Master replica two weeks off RW replicas

I have a four server Netware 5.1 server setup.  All have the latest SP, and are all at DS8.85.  Put simply, all the servers are the same, with different roles for each (i.e. all patched the same)

There is :
MAIN
BM
OFFICE

Main functions as the Master Replica, with BM and OFFICE having RW replicas
Main is a Single, and the others are Secondary.  Time is in sync, with a +/- of 0

"Report synchronization status"

Start:  Friday, June 3, 2005  12:51:40 pm Local Time
Retrieve replica status

Partition: .[Root].
  Replica: .OFFICE.wic               6-03-2005 12:47:37
  Replica: .MAIN.wic                 5-19-2005 17:27:39
    Server: CN=OFFICE.O=wic          6-03-2005 12:25:48  -128 Local
      Object: CN=usrcat.O=wic
    Server: CN=BM.O=wic         6-03-2005 12:25:48  -625 Remote
      Object: [Root]
    Server: CN=OFFICE.O=wic          6-03-2005 12:33:13  -128 Remote
      Object: CN=usrcat.O=wic
  Replica: .BM.wic              6-03-2005 12:47:29
All servers synchronized up to time:          5-19-2005 17:27:39  Warning

Finish:  Friday, June 3, 2005  12:51:41 pm Local Time

*** END ***

(That day (5-19-2005) is of note, as we had a switch failure that caused the servers to stop talking for about 20 minutes).

We rebooted the server, no effect.

We have done:
- Unattended full repair
- Repair Local DS Database
- Synchonize the replica on all servers (Done on both Main and BM)
- Repair selected replicas

We are getting -698 and -654 error codes

Any suggestions?

Thanks,

TN

Answer : Replica problem - Master replica two weeks off RW replicas

The symbol that looks like an angled yellow hourglass is what the Master Replica's symbol is.

What we have to worry about is the red bar.  Can you right-click that icon and select "replica information" to see what NDS manager thinks the main server's replica status is?  Do you have a single-partition tree, or are there other partitions' replicas we have to worry about?

It seems from what I see in your posts that the contextless login catalog isn't syncing to the master replica, and that's the object that keeps kicking out the lock error.  Can't tell if anything else is having problems or if the other errors stem from this.  It seems, further, that your BM replica and Office replica are in sync with each other and it's the Main replica that's out of sync.

Have you added users or made major changes since the switch hiccup?  If not, you might be best off making the Office server the Master, then forcing the main server's replica to take the objects from the Office server by running dsrepair on the main server (now a r/w replica), advanced options,  replica and partition options, replicas stored on this server (select the partition), view replica ring, select main server (should say replica type read/write), receive all objects for this partition.  

That will force the main server's replica to receive all of the objects, fresh, from the Office server (master replica).

Think about it to see if that makes sense to you.  Don't just do it - I'm not certain about this - it's just what occurs to me at the moment. It kinda depends on what NDS manager says the master replica's status is.