Saturday, February 25, 2012

Replication Logreader query

We had replication set up and working successfully for 2+ years between an intranet server and a webserver (offsite).

The web server was moved and the replication went out of sync. :mad:

Now when we try the replication it is trying to start from scratch and as you can imagine there is quite a lot of data after two years!

I have scheduled the logreader to run but after 8 hours it is still running is there any way to find out how long it should take?

We have removed any unnecessary data to try to help speed up the process.

Help appreciatedYou probably need the snapshot agent instead of the logreader agent. It takes the starting "snapshot" of the database to send to the subscriber, to restart replication.

-PatP|||We ran the snaphot first thing this morning - job outcome was successful.

Getting this message when we look at the snapshot step details for the run agent step:

"A snapshot was not generated because no subscriptions needed initialization. The step succeeded."

Any ideas on how to find out how long the logreader will take?|||Did you reinitialize the publications? That will flag the publication for a new snapshot. Just FYI, you probably want to script the indicies and the permissions on the subscriber, since the replicated tables will be destroyed and recreated as part of the process.

-PatP|||We have reinitialised the publications.

Thanks for the tip on indexes etc.

Any tips on finding out how long to complete the logreader and then distribution tasks? :confused:|||The logreader is eternal, if replication is going on, the logreader is running. The snapshot agent usually doesn't run too long, it simply copies the data from the published tables into flat files, but its performance is VERY dependant on your hardware (fast hardware/short runtime, slow hardware/well, you know that drill).

The distribution task depends on way too much stuff for me to take a guess at its performance... The network connection, the subscriber performance, the distributors disk, and about a half a gazillion other factors come into play.

Once the tasks start running, you can sometimes get better ideas about the performance, but knowing nothing about your configuration I can't even hazard a guess!

-PatP|||Pat

Distribution is executing - looks like it will be a long night :)

Thanks for your help - it is greatly appreciated

Cheers|||If you wander down to the distribution agents in your replication monitor (on the distributor), you can at least watch the "paint dry" via the status messages. It ain't much, but it is better than nothing!

-PatP

No comments:

Post a Comment