Compiling a new RHEL-4 kernel

[Nigel Wade]

Simon wrote:
> 
> On Dec 21, 2005, at 4:33 AM, Nigel Wade wrote:
> 
>>
>>
>> You ought to have a different sg device for each LUN if sg is  working 
>> correctly. At least that's what I get for multiple LUNs on  a SCSI 
>> RAID. Each LUN on the RAID has a different sg device, and is  mapped 
>> to a different sd device.
>>
>> What is the contents of /proc/scsi/sg/devices? This should show you  
>> the SCSI id and LUNs of all identified SCSI devices on the system.
> 
> 
> Well, it doesn't mean much to me, but here it is:
> 
> [root at www4 ~]$ cat /proc/scsi/sg/devices
> 1       0       0       0       0       1       4       0       1
> 1       0       1       0       0       1       4       0       1
> 1       0       8       0       3       1       2       0       1
> 2       0       0       0       0       1       63      0       1
> 2       0       15      0       3       1       3       0       1
> 2       1       5       0       1       1       3       0       1
> 2       2       15      0       3       1       3       0       1

It's only telling you the same info as sg_scan does, but in a much less 
understandable format. Column 1 is the controller number, column 2 is the 
channel, column 3 is the SCSI id and 4 is the LUN. The others are to do with 
device types.

> 
>>
>> Do you have sg3_utils installed? If not I'd suggest doing so, this  
>> package really helps with sorting out sg/st/sd assignments. For  
>> example, sg_scan on my system show this:
> 
> 
> Yep:
> 
> [root at www4 ~]$ sg_scan -i
> /dev/sg0: scsi1 channel=0 id=0 lun=0
>     IBM-ESXS  MAP3735NC     FN  C101 [rmb=0 cmdq=1 pqual=0 pdev=0x0]
> /dev/sg1: scsi1 channel=0 id=1 lun=0
>     IBM-ESXS  VPR073C3-ETS10FN  S370 [rmb=0 cmdq=1 pqual=0 pdev=0x0]
> /dev/sg2: scsi1 channel=0 id=8 lun=0
>     IBM       39M6750a S320  0  1    [rmb=0 cmdq=0 pqual=0 pdev=0x3]
> /dev/sg3: scsi2 channel=0 id=0 lun=0
>     IBM       SERVERAID         1.00 [rmb=0 cmdq=1 pqual=0 pdev=0x0]
> /dev/sg4: scsi2 channel=0 id=15 lun=0
>     IBM       SERVERAID         1.00 [rmb=0 cmdq=0 pqual=0 pdev=0x3]
> /dev/sg5: scsi2 channel=1 id=5 lun=0
>     HP        Ultrium 3-SCSI    G37B [rmb=1 cmdq=0 pqual=0 pdev=0x1]
> /dev/sg6: scsi2 channel=2 id=15 lun=0
>     IBM       EXP400   S320     D110 [rmb=0 cmdq=0 pqual=0 pdev=0x3]
> 
> (that's a useful tool - cheers :-)

Ok, so the only device on scsi2, channel 1 is the Ultrium tape drive. No sign of 
your tape changer. It looks like the system isn't set up to scan LUNs. I don't 
know of any way to fix that on a live system without shutting down all SCSI 
devices and unloading the SCSI drivers.

> 
>> Incidently, I have this entry in /etc/modprobe.conf:
>>
>> options scsi_mod max_luns=32
> 
> 
> ... and I think that's the difference - this was what Matt Brookover  
> suggested as well :-)
> 
> I'm waiting for a new machine to arrive before I go fiddling with  
> devices though, in case anything bad happens. The entire machine is  
> SCSI, internal and external RAIDs, and I don't want there to be any  
> surprises on a currently-active machine :-)

It's almost guaranteed that if something can go wrong, it will go wrong, and at 
a time which will cause the most embarassment and damage.

-- 
Nigel Wade, System Administrator, Space Plasma Physics Group,
             University of Leicester, Leicester, LE1 7RH, UK
E-mail :    nmw at ion.le.ac.uk
Phone :     +44 (0)116 2523548, Fax : +44 (0)116 2523555