We have a new server built for a client who require tape backup/archive that has proved nothing but a pain in the butt with no resolution
Symantec say it's hardware, Quantum say it's the HBA or cable - any ideas?
We get this error when running Quantum xtalk diags:
HBA detected a problem during the write buffer operation.
HBA Error [000000E0] : SRB_STATUS_INVALID_SRB
LSI Adapter, Ultra320 SCSI 2000 series, w/1020/1030 - drivers 1.28.3.67 whish Windows says are current
Quantum LTO3 HH drive
xTalk Version 6.4.1.1
___________________________________________________________
Start: 03/01/2011 17:03:09
Device: QUANTUM ULTRIUM 3 PW1018AME50528 0:3:7
Device_Health_Check
____________________________________________________________
Ready Test
No Problem Detected
____________________________________________________________
Record Drive Settings
Current Compression Status = Enabled
Current Compression Mode 0 == Full SCSI control of compression
No Problem Detected
____________________________________________________________
Drive Detection Test
Drive Type:Ultrium 3 HH
No Problem Detected
____________________________________________________________
SCSI Interconnect Test
Sense: CC: CC ASC: CC ASCQ: CC = "No Sense Available for this condition"
VS Sense: 0
Catagory: N/A
Details: No error
HBA detected a problem during the write buffer operation.
HBA Error [000000E0] : SRB_STATUS_INVALID_SRB
Problem Detected
____________________________________________________________
Drive Information Report
The drive family is unknown.
Report Complete
____________________________________________________________
Quick Write Read Test
Setting compression on
Setting variable block mode
The root problem, specifically is that the diagnostic software you are running sent a CDB (The SCSI WRITE BUFFER) command which is used to initiate a certain type of self test. The tape drive does not support that particular self-test (The SRB_STATUS_INVALID_SRB).
This is normal behavior. There are a half-dozen types of self-tests, plus vendor-unique ones. The SCSI controller did what it is supposed to do. So did the tape. The problem, if any, is that the test isn't supported on that tape .. but since the tape said it isn't supported (properly), then no harm, no foul.
Now all that worked, but here is the problem. The Sense information is bad. It should be sense 05, the ASC should be 20 or 21, and ASCQ could be just about anything. Sense data should be returned. 3 bytes worth, not that CC: CC ASC: stuff
No way did the tape send hex codes CC CC CC. This isn't valid sense data.
So 100% sure that you have at least a software bug. It returns invalid sense data.