On Wed, Jun 21, 2006 at 02:40:34AM +0200, Arjan van de Ven wrote: > On Tue, 2006-06-20 at 16:59 -0700, Arild Jensen wrote: > > I have an interesting problem on my Linux box: > > > > Setup: > > IBM x445 server > > 2 x QLA2340 fiber HBAs > > Brocade Director 12000 switch > > Hitachi HDS 9960 RAID array > > SLES 9 OS, kernel-bigsmp-2.6.5-7.201 > > > > Problem: > > The qla2xxx driver *always* reports 36 SCSI2 LUNs, regardless of what I > > configure on the RAID array. It should report 17 SCSI3 LUNs. > > why should it report 17? Because I configured our RAID array to assign that box 17 LUNs. > if you have set up 17 luns, showing up as 36 devices is not that > strange; first you have the management lun, so the kernel should see18 > if you configure 17 real disk ones, and then you have 2 paths to the > device, so that is times 2 ==> 36. Now the multipathing in the linux > kernel then should bundle those into multipathing devices, after the > kernel sees them. If that's not happening you should configure the dm > multipathing component.... >From each HBA I am seeing 36 LUNs. Since they are SCSI2, the OS only sees the first 8, though when I dig deeper, such as looking at /proc/scsi/qla2xxx/*, I see that each HBA reports 35 LUNs. However, you might be right about something regarding the multipathing. Your math certainly adds up. I'll try to boot the box with only one HBA and see what the OS+HBA are reporting. > was the machine that is working also using multipathing? Yes. Thanks for the reply, Arjan! -- Arild Jensen - : send the line "unsubscribe linux-scsi" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
