Time Machine stalls when backing up server data

Question

I have an Xserve set up as a Workgroup Server at one of my locations. I've decided to simplify the backup scheme at this location, so I'm doing away with my shell scripts and just turning on Time Machine and pointing it at a large external array. This way I can enable a few selected users on site to easily retrieve files from the backup. I've done this a few times, to much success.
This time, I'm experiencing a new problem which I've not been able to solve. The users in this workgroup have quite a bit of data, about 1TB, much more data than I've dealt with using this setup.
*The Problem:*
Time Machine begins copying data during the initial backup and achieves a transfer rate of about 10MB per second, not spectacular, but fine. When the backup reaches about 80%-90% complete, it stalls or slows to a crawl. It gets to about "900GB of 1.0TB" and the transfer rate slows to about 1MB per minute. Time Machine does not deliver any errors. It continues to back-up, just very slowly. So slowly sometimes it appears to be stopped.
I am backing up to a sparsebundle disk image, just like Time Machine will do over the network.
*I Have:*
• reformatted the destination volume
• disabled Spotlight indexing on the destination volume
• disabled Spotlight indexing on the source volume
• copied the data directly to the destination volume using Finder (with success and fast performance)
• reset the Time Machine preferences
• used the default disk image Time Machine creates
• used a disk image with a 4TB size limit
• used a disk image with no partition layout
• used a disk image with a GUID (GPTSPUD) layout
• used a disk image with a Apple Partition Map (SPUD) layout
• used a disk image with increased band size (128MB)
• connected to the backup array using FW800
• connected to the backup array using USB 2.0
• unlocked all files on the source volume
• set all permissions on the source volume (chmod)
The system is an Intel Xserve, 10.5.7, 2.8Ghz, 2GB RAM, 80GB boot drive, 2 x 1TB data drives. I am only backing up the data being shared to the workgroup, not the server boot volume. The destination volume has plenty of available space.
Checking the logs (Console filtering for 'backupd') reveals several of these errors:
+8/27/09 12:27:51 AM com.apple.backupd[45548] SystemFlippers: Consumed too much data for FREF ID 131 (pBase = 0x4f839c0, p = 0x4f83ac6, pEnd = 0x4f839c7)+
+8/27/09 12:27:51 AM com.apple.backupd[45548] CoreEndianFlipData: error -4940 returned for rsrc type FREF (id 131, length 7, native = no)+
I don't have enough foo to make sense of those errors or even tell if they're related. I also don't know if 12:27 AM is when the backup began to slow down.
I anticipate some answers will be "Time Machine was built for end users not servers", "Time Machine wasn't intended for this purpose." Which I don't necessarily agree with in this case. For the purposes of discussion, this machine could be considered just a desktop with file sharing enabled, backing up the same amount of data.
I've tried to provide as much information as I can think of, and appreciate any feedback.
Message was edited by: Arty Admin - minor correction.

Answer 1

It's a bug and should be reported to Apple's feedback.
Even though TM is failing in the manner described, is it backing up up your data correctly ?
Message was edited by: BarryXSharp