Windows 2003 sp1 File Replication problem

[sbradshaw]

I have 2 Windows 2003 sp1 Servers, Server1 and Server2; they are being used as web servers. The servers are at a Data Center 250 Miles away. We have file replication enabled between the 2 servers. We are replicating folders that contain pictures in them. There is a total of 319 Folders with total number of files at around 900,000 files totaling close to 12 Gb. Has anyone ever heard or tried to replicate this many files and folders?

Server 1 has 854,958 files in 326 Folders at 11.3GB

Server 2 has 311,671 files in 116 Folders at 3.04GB

Server 1 has close to all the files we want to replicate (I would say 98%) I think there are some on Server2 that have not replicated to Server1

Server 2 right now is rebuilding the files. It usually has around 243 folders

On Server 2 we deleted the Jet DB and let it rebuild starting last week on the 3/7/06. The Jet DB is still being updated as of 3/13/06 and now is over 1 GB

Server 1 Jet DB is 1,362,120 Kb and has not changed size since 3/7/06

I read somewhere that I should be able to ping the computers GUID. Can someone tell me how this is done? Also how can you find out the GUID of a computer? I looked in the logs of FRS and I saw 2 GUID that I thought was the servers but I have no way of knowing for sure nor do I know if I pinged them correctly.

I have seen a lot of event messages that mention DNS this is why I am interesting in pinging the GUID.

Also, Is there a proper way to restart replication on both servers? All I know to do is restart the FRS services on both servers.

Errors in the event log:

13552

13544

13555

13504

13559

13568

13506

13505

Warnings in the event log:

13508

13520

13522

13509

[cluberti]

That kind of replication is not unheard of, but at those data sizes you would need to have a relatively fast network (1GB or better), good chunks of memory (the file cache on those things will be huge, and you'll need at least 4GB of RAM to keep it fed), and at least a cadre of 10K SCSI disks (15K if you can swing it, and RAID10 if you are able, to really keep I/O bottlenecks from occurring). I would also recommend NOT running antivirus software in realtime scanning mode on either node as well, as this adds at least double the processing time per I/O request.

You would also do well to upgrade to R2 on those machines to get the DFS and FRS improvements in large file replication (copying only updated files and updated portions of files, faster replication, etc). FRS and DFS in R2 see huge improvements over their SP1 counterparts in heavy use scenarios.

Edited March 14, 2006 by cluberti