This is solved. The problem was with the IO blade in the library. These were the only 2 drives on that IO blade, when we replaced the IO blade everything started working as expected and I am not getting fragment errors.
Can you double-check tape drivers used in Windows Device Manager for all LTO3 tape drives?
Can you check output of 'tpautoconf -t' to see if all LTO3 drives are on the same firmware?
Are all LTO3 drives connected to same or different HBA?
Can you confirm that all settings and drivers for hba's are the same?
Can you double-check tape drivers used in Windows Device Manager for all LTO3 tape drives?
Yes it shows it fine in the device manager. Shows fine in the scan output as well.
Can you check output of 'tpautoconf -t' to see if all LTO3 drives are on the same firmware?
All the IBM drives are on the same firmware level 93GM
Are all LTO3 drives connected to same or different HBA? These are 10 drives shared amoung 1 master and 4 media servers
Can you confirm that all settings and drivers for hba's are the same? yes they should but let me cross verify.
Yes setting are same.
The 2 "failing" tape drives, are they connected to the same media server as the others ?
The values are very off the regular buffer settings, buffer setting must be the power of 2 (64K, 127K, 256K etc etc).
mount of data read (610432) does not match the fragment kbytes (1056128) for backup id XXX_1221131992, copy 1, fragment 37
Yes these 2 drives are connected with others. That is the data read it shows in that error, it is not the fragment size.
I checked that these were written with buffer size of 131072 in the past, 128K.
I logged a call with Quantum but they are not finding anything on the library side.
