The general reason for not connecting tape drives and disks on the same controller is that the different blocksize causes hell for all involved, and performance drops significantly - especially if doing local disk-tape backup. I'm not sure about the card in use here. If this is a dual controller card, then you will probably be okay. If it's a single SCSI bus, then you're mixing them.
There's another good reason to avoid these in cluster configurations. SCSI bus resets are sent to the disks as part of the cluster control. Sending this to a tape drive is bad news - if it's a single spindle drive it will be rewound. If it's in the middle of writing, it will get an I/O error and probably get marked suspect/poor/ugly, kicked out of the drive, and you lose its remaining capacity.
The other reason I've seen for CRC errors in the past is simply that the last block was not written fully to tape. NetWorker used to throw out these errors a few years back most times it hit the end of tape, since the data was verifiably incorrect for that block. It wasn't a problem, since the block got written correctly at the start of the next tape, but check that this isn't at the physical end of the media - might be something similar coming up in your environment. Just go to end of tape on a tape that gave this error and write an MB or so - if the write is successful, you hit this before the end. Being at the end of tape when you do this shouldn't cause any problem for recoveries through Backup Exec - its data is still there.
A sysadmin should never cross his fingers in the hope commands will work. Makes for a lot of mistakes while typing.
