Merge replication problem

  • Hi everybody,

    I run merge replication with a central publisher and 10 subscribers. The Publisher is SQL2000 SP3 while most subscribers are still SQL 7.0 SP4. Everything was working fine for months but lately I get strange errors on two subscribers. The exact message is:

    The merge process could not update the list of subscriptions.

    The merge process could not update the list of subscriptions.

    (Source: Merge Replication Provider (Agent); Error number: -2147201010)

    ----------------------------------------------

    Cannot insert duplicate key row in object 'sysxlogins' with unique index 'sysxlogins'.

    (Source: SQLINT88(Data source); Error number: 2601)

    It happened now twice during the last 2 weeks and both times the same two subscribers. When I manually restart the merge agent it works normal. Does anybody has any idea what is going wrong ???

    Thanks a lot Markus

    [font="Verdana"]Markus Bohse[/font]

  • Did you ever delete those subscriptions and then re create them again? Maybe they wasn't cleaned right.

    Check resubscribing the servers to your publication.

  • Yes, I deleted the subscriptions before, but the last time was months ago when we upgraded the Distributor/Publisher and everything was clean. Anyway why should the system create new logins once the whole thing is running. I'm lost

    [font="Verdana"]Markus Bohse[/font]

  • I would recommend to run the merge agent within the command line using replmerg.exe.

    You can copy the parameters usend in the job wich runs the agent, and used them with the .exe. Also recommend to change the verbose level to 2 so that you can see all the output and check what is the agent doing.

    For me using the command line to run the agents is the best way to trouble shoot, because you can know exactly what the agent is doing.

    Search in the output for the error and if you need help, also post it here.

  • I agree that running the agent from the command line is the best way for troubleshooting. The disadvantage especially with errors like this one, which happens very irregularly and only once every couple of weeks, is tha fact that you get huge outputfiles. Is there a way to autocreate a new file everyday, or maybe a file rollover like in Profiler ?

    [font="Verdana"]Markus Bohse[/font]

  • I don't think so. I would let the agents running from the jobs of SQL while Things are ok, And when I have a problem, I would run them from the command line.

    I woudn't use the command line always because the files coud get too big.

Viewing 6 posts - 1 through 5 (of 5 total)

You must be logged in to reply to this topic. Login to reply