Showing posts with label history. Show all posts
Showing posts with label history. Show all posts

Thursday, March 8, 2012

Blocking in distribution database...

Hello...
I am noticing that everytime the the Agent history clean
up job runs... it causes contention and blocking in the
distribution database.
Has anyone come across this? If so, how did you resolve
this. I am using SQL Server 2000 and every server in my
topology uses SQL Server 2000 as well.
Please advise...
Thanks,
Brett
are you running datamirror? Datamirror does not clean up efficiently.
Blocking in the distribution database is caused by 2 factors
1) too many push agents running simultaneously. Try to stagger and schedule
them. If you can migrate to pull subscriptions.
2) too much replication metadata pooling in your distribution database due
to subscribers who are frequently off line. If you are using anonymous
subscribers this exacerbates the problem. Consider using named subscribers.
"Brett" <anonymous@.discussions.microsoft.com> wrote in message
news:1ae2001c41e7a$fc3ac590$a101280a@.phx.gbl...
> Hello...
> I am noticing that everytime the the Agent history clean
> up job runs... it causes contention and blocking in the
> distribution database.
> Has anyone come across this? If so, how did you resolve
> this. I am using SQL Server 2000 and every server in my
> topology uses SQL Server 2000 as well.
> Please advise...
> Thanks,
> Brett
>

Friday, February 10, 2012

Bit more specific, Can one pass an NVARCHAR to IF UPDATED()?

Question is pretty much in the thread title.

I'm still plunking away on a history trigger that requires the evaluation of whether or not each column has been updated. If it has,t hen I'll be tossing a record into a history table.

Basically I need something to function along the same logic as this:

IF UPDATED(@.s)
BEGIN
--some code
END

Any thoughts you folks may have are greatly appreciated.About the closest I can get is the Columns_Updated() and maybe some fancy dancing with syscolumns.

-PatP|||About the closest I can get is the Columns_Updated() and maybe some fancy dancing with syscolumns.

-PatP

At least it's something to go on. I'll take a look at it tomorrow.

Much obliged good sir.