Mark Whidby wrote:
> Hi,
> I'm trying to configure a RAID box with 12 x 1 Tb disks on an SL 5.0 system
> running kernel 2.6.18-8.1.8.el5 through this controller:
> 
> 04:06.0 SCSI storage controller: LSI Logic / Symbios Logic 53c1030 PCI-X 
> Fusion-MPT Dual Ultra320 SCSI (rev 08)
> 
> However, whenever I configure more than 3 of the disks in a LUN I get 
> SCSI errors.
> 
> For 4 disks in a RAID5 configuration:-
> 
> Nov  4 10:36:29 terra kernel:   Vendor: transtec  Model: 
> PV610S12R1A       Rev: 347G
> Nov  4 10:36:29 terra kernel:   Type:   
> Direct-Access                      ANSI SCSI revision: 05
> Nov  4 10:36:29 terra kernel: sdq : very big device. try to use READ 
> CAPACITY(16).
> Nov  4 10:36:29 terra kernel: SCSI device sdq: 5858973696 512-byte hdwr 
> sectors (2999795 MB)
> Nov  4 10:36:29 terra kernel: sdq: Write Protect is off
> Nov  4 10:36:29 terra kernel: SCSI device sdq: drive cache: write back
> Nov  4 10:36:29 terra kernel: sdq : very big device. try to use READ 
> CAPACITY(16).
> Nov  4 10:36:29 terra kernel: SCSI device sdq: 5858973696 512-byte hdwr 
> sectors (2999795 MB)
> Nov  4 10:36:29 terra kernel: sdq: Write Protect is off
> Nov  4 10:36:29 terra kernel: SCSI device sdq: drive cache: write back
> Nov  4 10:36:29 terra kernel:  sdq: unknown partition table
> Nov  4 10:36:29 terra kernel: sd 5:0:0:1: Attached scsi disk sdq
> Nov  4 10:36:29 terra kernel: sd 5:0:0:1: Attached scsi generic sg16 type 0
> Nov  4 10:36:31 terra kernel: sd 5:0:0:1: SCSI error: return code = 
> 0x000b0000
> Nov  4 10:36:31 terra kernel: end_request: I/O error, dev sdq, sector 
> 5858973568
> Nov  4 10:36:31 terra kernel: printk: 16 messages suppressed.
> Nov  4 10:36:31 terra kernel: Buffer I/O error on device sdq, logical 
> block 732371696
> Nov  4 10:36:31 terra kernel: sd 5:0:0:1: SCSI error: return code = 
> 0x000b0000
> Nov  4 10:36:31 terra kernel: end_request: I/O error, dev sdq, sector 
> 5858973568
> Nov  4 10:36:31 terra kernel: Buffer I/O error on device sdq, logical 
> block 732371696
> Nov  4 10:36:32 terra kernel: sd 5:0:0:1: SCSI error: return code = 
> 0x000b0000
> Nov  4 10:36:32 terra kernel: end_request: I/O error, dev sdq, sector 
> 5858973688
> ...
> Nov  4 10:36:33 terra kernel: sd 5:0:0:1: SCSI error: return code = 
> 0x000b0000
> Nov  4 10:36:33 terra kernel: end_request: I/O error, dev sdq, sector 
> 5858973688
> Nov  4 10:36:33 terra kernel: sd 5:0:0:1: SCSI error: return code = 
> 0x000b0000
> Nov  4 10:36:33 terra kernel: end_request: I/O error, dev sdq, sector 
> 5858973688
> Nov  4 10:36:33 terra kernel: sd 5:0:0:1: SCSI error: return code = 
> 0x000b0000
> Nov  4 10:36:33 terra kernel: end_request: I/O error, dev sdq, sector 
> 5858973568
> ...

Hi,
I'm just posting the solution to this in case anybody else hits a similar
problem. The errors were caused by a hardware configuration error.
The "Default Transfer Clock" on the raid was set to 80 MHz yet filesystems
bigger than 2 Tb require this to be 160 MHz. Once this was set correctly
everything was fine.

-- 
Mark Whidby
Infrastructure Coordinator (Unix) - Physics/Chemistry/EAES/Mathematics Team
Information Systems
Faculty of Engineering and Physical Sciences