Windows Server 2003 R2 Standard 32bit master server, each of the SAN media servers is Windows 2003 R2 Standard 64bit.
We recently upgraded from 5.1mp3 to 6.5.6. To say we are experiencing problems would be a gross understatement.
We have 2 fibre connected media servers connected to a Dell ML6000 library.
Each media server can 'see' 2 LTO4 drives.
I do Full backups on Friday at 1700 and Differential Incrementals Monday through Thursday also at 1700. Done it ever since 5.1mp3 came out, never once had a problem.
The media server I am currently working on has a 6TB LUN which contains 1TB of replicated data (Microsoft DFS).
The replicated data is from 54 replication groups, full mesh between the media server and remote server(s).
The ALL_LOCAL_DRIVES was getting the 'C' and 'D' and 'E' drives, where E is the LUN. All that would be on the 'E' drive would be the parent file structure. The actual files were being (not) backed up in the Shadow Copy Components path for lack of a better word. When an error would occur (can not read file, remote rewind of drive, etc) the backup would fail and the image would be invalidated. I thought that with create checkpoint every 15 minutes meant that if a backup failed I would only loose 15 minutes worth of writes.
Questions:
Would I be better off to split this out into 3 policies each with 18 DFS groups specifically named in the Backup Selections?
If I did as stated above and set Multiplexing to 6 would it multiplex 6 of the DFS groups at a time?
Why didn't the 15 minute checkpoints keep the entire job from failing?
I do have an Enterprise support case for this, I am soliciting more opinions.
Thanks for any assistance,
Ed
