Until recently my backup was running swimmingly, but when the tape drive started to act up I figured that it was time to replace it. I've got a pile of older DDS-2 tapes and use an HP SureStore 12000e with a cartridge, so I simply replaced the drive. Now I'm on another drive and the 4th replacement cartridge and am having the same problem, and am thinking that it's not the hardware.
The symptom is that the drive seems starved for data. If you can't keep the drive streaming efficiency of the storage goes down, and the drive has to stop, reverse and recover. This uses more tape for the same amount of data as well as taking a lot more time. I used to see the drive activity in gkrellm, now I don't see the activity at all. If I modify the output to /dev/null instead of /dev/st0 the disk activity is extreme, as expected. What I've done: changed the driver to use 128K as a buffer instead of the default 32K. Changed the block size in cpio() to 32K. Tried erasing tapes (don't know if it's an issue in DDS-2 tapes, was in old days). I even removed the other devices on the SCSI chain, making the tape drive the olny device on it. Finally I tried using dd() to write to the tape to see if I had the same issues on the drive, and it appeared to have the same problems - very slow data transfer to tape. It's hard to believe that I have 3 bum tape drives and/or 5 bad cartridges, but perhaps one of the drives has contamination that's gotten to all of them in my efforts to figure out the problem. Ideas? Time to go to something a bit newer school? (It's not like I'm using QIC or something...)
____________________________________________________ Want to buy your Pack or Services from MandrakeSoft? Go to http://www.mandrakestore.com Join the Club : http://www.mandrakeclub.com ____________________________________________________
