Weird behavior with replicating with new server

HCL Domino Domino Forum
1

While I am waiting on Lotus to suggest something useful, I am curious if anyone else has noticed this behavior. Here is the outline:

  1. We bring up two new mail servers because of space issue and we will be moving user mail files to these new server.

  2. I create a new replica of User A’s mail file over to these new servers.

  3. Upon setting User A’s client setting to now replicate with a new server, their first replication does the initial one time thing and then pulls in 4000-20000 updates taking a long time for the user.

I have seen this in the past but now care more as we have to move alot of users. I have made sure that the replicate unread marks is set to all servers. I also performed an updall -r on the new server replica before the first replication as lotus has suggested so far. No luck. Any idea what would cause a first local to new server replication to take so long??

2

Subject: Weird behavior with replicating with new server

OK I am confused on what is being asked but here goes… if you are asking why the first replication take so long, ask your self this.

Is the DB already on the machine (local or server): guessing answer now

How much data is to be moved or copied- how fast is the network / Disk / blah blah blah

Is there anti virus scanning the NSF tranfer as it is being replicated… this can cause slowing.

Basically the first replication anywhere takes time. But after the DB is moved across then and only then will it only send the updates to the DB.

I hope this helps.

3

Subject: RE: Weird behavior with replicating with new server

I thought I was more clear than I was. The users have a local replica already that gets automatically replicated with our old two server mail cluster. This replication takes place every five minutes. I then create a server replica from one of the old mail servers to a brand new server. I then configure the local client for the user to replicate their local mail file with the new server and even though that server replica was created the day before or even the same day, their initial replication with that new server yields a bunch of updates even though their local file and that new server replica should be exact. As I mentioned, the setting fro replicate unread marks is set to all servers so that shouldn’t cause anything extra.

4

Subject: RE: Weird behavior with replicating with new server

I know it sounds obvious but just in case - you are creating a full replica immediately aren’t you, not setting it create the replica stub and get the docs at the next replication?

5

Subject: Weird behavior with replicating with new server

How big are these mail files? Since you’re moving to a new server because of space issues and with the cited 4000-20000 updates, I’m guessing pretty big. The longish initial replication with a ‘new’ replica (i.e. a replica on an unknown server) is going to be longish (directly proportional to the size/doc count of the db) because notes has to do basicly a full scan for differences. This is the same effect you’re warned against when you clear Replication History. Once a complete/full replication is done with a given replica, subsequent replications are faster because the replicator can take some shortcuts based on the history information.