CirquedeSQLeil (12/21/2010)

---

That is an excellent...

If the default trace is enabled (default is on) then you can use the script HERE [/url]to read through that and find out possibly.

Once you have replication configured and your initial snapshot delivered there is no reason that you should HAVE to reinitialize again. Granted it may be necessary as a best option...

So, there is very little you can do to alleviate that aside from making sure that your IO subsystem can handle that activity; which at present it is not.

Ok - thanks for the update. Please let me know what you see with those two counters.

Yes, I see what you did. What I need to know though is Checkpoint pages/sec high when CPU is high and is there always a direct correlation. The wait type...

What do you see if you set up perfmon to collect CPU information and SQLServer:Buffer Manager Checkpoint pages/sec?

Brian - Have you reviewed this - http://blogs.technet.com/b/vipulshah/archive/2006/11/30/understanding-perfmon-counters-while-troubleshooting-sql-server-performance-issues.aspx

Your write measures seem high but based on what you stated your read measure looks ok. What is your primary type of activity...

http://technet.microsoft.com/en-us/library/ms187946.aspx - This procedure, sp_scriptpublicationcustomprocs, will create the procedures for you. A couple of notes here, first set your text output in SSMS to largest possible, second this is run...

That is a recommended practice and the large transaction log backup post index rebuilds should be expected. I will say though that you will still see some log bloat even...

Ok - sorry about that. You should definitely look at getting the latest service packs in place then to solve that issue.

Pei Zhu-415513 (12/20/2010)

---

David, Yes I meant out of sync (missing transaction). Also Tlog was bloated and continued to grow until I broke the replication or ran repldone.

I wonder if...

Pei Zhu-415513 (12/20/2010)

---

The issues (bloated tlog) I used to experienced were on sql2005 sp1 and sql2008 rtm.

I used structured dynamic codes something like 'use dbname; alter index .. rebuild online'...

Have you read this? Sounds like skipping commands during index operations might be an issue. I stumbled across this when looking for (vainly I might add) a document that shows...

As I stated those are NOT replicated commands. Here is an article[/url] that walks you through testing this.

Enjoy.

Edited: Previous wording sounded bad. 🙁