Distribution Deadlocks with growing distribution database

  • We are replicating to a reporting server using distribution with an ETL server with Business Objects data services software. The distribution database has grown to 40 gigs and cleanup is taking over 15 hours to complete. I have stopped the cleanup job. There are only a couple hundred rows in msrepl_Commands.

    After rebooting the server, distribution went down to 500 mg. Turned cleanup back on and it ran for seconds. After two days, it grew back up to 30 gigs and cleanup is taking 10+ hours before I stop it.

    I'm seeing locks with Repl-logReader. It seems to be between the distribution database and the ETL server hitting distribution at the same time. I understand why that would happen and it makes sense, but nothing has changed recently and our setup has been working for about a year. Any ideas on why distribution would be blowing up and why cleanup is taking so long while very few records are in msrepl_commands would be Greatly appreciated!

  • I have an update. We've found the problem and corrected it. Turned out to be very simple.

    We had our distributor agent using the verbose history agent profile. That had a parameter, HistoryVerboseLevel, which was set to 2.

    I'm not exactly sure what that does, but I changed all existing distribution agents to use the default agent profile which has HistoryVerboseLevel set to 1.

    It's been a day now, and the version ghost records are being cleared out and database size is remaining small.

Viewing 2 posts - 1 through 1 (of 1 total)

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