Backup Exec 2010 R3 corruption errors when changing buffer size on LTO 4

[JamesSAU]

Hi,

I have a Dell ML6000 tape library with two IBM LTO 4 drives installed, and fiber attached to our BE 2010 R3 server (Windows 2008 R2, 16GB RAM, 8 XEONs). To try and improve the throughput of various jobs (400-1000MB/min) backing up Windows 2003/2008/2008R2 servers with agents. I have tried increasing the buffer size on the LTO4s from the default 64K to various sizes up to 1MB. Each time I get various errors stating that a communication failure has occurred. Errors are:

Final error: 0xe000ff14 - A communications failure has occurred with a Sharepoint resource.

Final error: 0xe000fe30 - A communications failure has occurred.

e000ff12 - A communications failure has occurred with an Exchange Store resource.

The thing is I get substantially better throughput when using a larger buffer size - up to 3500MB/min, therefore I am loathe to go back to the default buffer size!

In looking at restoring the data from these failed jobs, it seems that its always the last file backed up per server that is being reported as being corrupt...when going back to the default settings to error disappears.

Any help would be greatly appreciated! Surely the LTO4 drives, which are rated at a much higher speed that what we get, can handle a larger buffer size?!

[JamesSAU]

I'm running BE 2010 R3 SP1 and 167567

[CraigV]

Hi,

You didn't perhaps increase the block size did you? You'd get an error message when increasing this.

Alternatively you can decrease the buffer size down to 512K, and remove the options listed below the Preferred Configuration settings on the drive.

Run the backup again and try a restore...

Thanks!

[JamesSAU]

I didn't increase the block size for exactly that reason, I have tried ticking and unticking the read/write scsi pass-through options as the library is SAN attached but it doesn't make any difference. Both single block mode boxes are unticked also.

[pkh]

I found that I cannot just increase the buffersize without increasing the blocksize.  See my article on this subject.

https://www-secure.symantec.com/connect/articles/tuning-my-lto4-tape-drive

[CraigV]

OK...so my suggestion would be as follows:

1. Default your settings again, and run a successful backup. Make sure you can restore some data thereafter.

2. Change the buffer settings 1 level up, and repeat 1.

3. when you hit a failure, I'd suggest you use the setting before that...