undistributed commands problem

  • I have a transactional replication between 2k5 and several 2k servers

    one subscription died saying that is was waiting for the snapshot after it had been running for weeks

    so i redid the replication to that server, now it says that there are 330000 undistributed commands

    yet if i check the record counts on the tables all are matching

    so i am not sure how to fix the problem

    the server in issue has a lot of locks on it that the developers are trying to get fixed

     

  • It appears the replication isnt clearing the command history

    after about 2 weeks the number suddenly went from 500,000 to 10

     

  • Did you get anywhere with this?

    I have a SQL2005 - SQL2008 replication giving me similar grief.

    [font="Arial"]-----------------------------------------------------------------
    Rich N - Development Manager @ Delta Technology Solutions Ltd - M.O.T.H.E.R. your virtual Sys Admin - Remote Server Health Monitoring - 24 x 7 www.deltatsl.com
    [/font]

  • Here is something that I have seen happen that has raised the undistributed commands.

    1. When ever there is a network hiccup this raises its ugly head. And after that the best thing to do is to restart the Distribution agent.

    2. Whenever there are long duration locks/blocks in the subscriber this happens. For that we set up Row level version in the subscriber. That helped reduce it.

    3. IO bottlenecks on the subscriber. Usually you use the subscriber for reporting. That means you have reports that run on historical data. This causes IO bottlenecks. At that time there will be IO related waits on the subscriber. Usually what can be done is to optimize the queries.

    -Roy

Viewing 4 posts - 1 through 3 (of 3 total)

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