Hey all,
My first post to the forum, and of course like all noobs it's a problem. I'm hoping someone else has some experience with this strange error.
Site background:
NBU 6 MP2
Solaris 9 sparc master/media server
Linux RHEL AS 64-bit media server to handle the raw partition backups, since Solaris wants to write a label on the disk.
L180 tape library
All on a SAN connected to Sun 6920 storage
Problem description:
Carved a test 36 GB LUN out of the 6920 to test our brand spanking new (and not yet a proven concept) backup solution. Put test data on it and ran a backup of the three partitions raw. Raw is the only way the proprietary appliance system can be backed up.
Backup failed on the third partition, after first two were successful. Error:
09/08/2006 14:50:31 - Error bpbrm (pid=20876) from client mirabu2: ERR - Read error at block 67850524 reading 262144 bytes in file /dev/sdg3. Errno = 5: Input/output error
09/08/2006 14:50:45 - end writing; write time: 0:21:01 the backup failed to back up the requested files (6)
At first we figured this was a bad block on the disk that simply needed to be written to before the system would recognize that bad block. So, we made another 36 GB test LUN from different segment of the storage, loaded test data, and started a backup. Same error message. Exactly the same, even down to the same "bad" block number.
Meanwhile, wrote 0's and 1's to the first test LUN in order to write to every block to trigger the recognition of the bad disk. But no failure occurred.
So, now it seems like NetBackup is having a problem with the block size of the read, perhaps the size represented by the storage feed and the block size that NBU is trying to read are causing this error. My questions are:
1. Has anyone encountered this error while trying to read/backup raw filesystems?
2. Where does one tune the block size for NBU 6 MP2? I know about the buffer size tuning in SIZE_DATA_BUFFERS and NET_BUFFER_SZ, but is block size tuneable?
Thanks in advance for all responses.