'out of memory' when spinning up a 1.2 TB disk

[lemons_terry at emc.com]

Hi

 

I have a system using a modified form of Red Hat Linux 7.3 with a 2.4.21
kernel.  The system uses QLogic Fibre Channel adapters to connect to a
CLARiiON CX300.  On the CX300, I have 8 LUNS.  The first LUN (LUN 0) is 8
Gbytes in size; the other seven LUNs (LUN 1 - LUN 7) are 1.2 TB in size.  In
/etc/modules.conf, I have, in addition to other entries:

 

options sd_mod max_scsi_disks_fs={n}

 

If I set 'n' to 1, mkinitrd a ram disk, and boot that ram disk, the system
boots up fine.  In application that runs on the system accesses the large,
1.2 TB LUNs with no problem.

 

If I set 'n' to 8, ram disk, and boot that ram disk, the system does not
boot correctly.  For all of the large LUNS, I see the following error
message:

 

Kernel: sdb: Spinning up disk ... <3> Out of memory: Killed process 15
(insmod)

 

Since the system works normally if 'max_scsi_disks_fs' is set to 1, I'm
assuming that the FC adapter driver is functioning correctly, and is not
causing this problem.  And I've read elsewhere that file systems (e.g.,
ext3) that run on Red Hat Linux 7.3 support 1.2 TB (and larger) disks.  So,
I'm not sure what is causing this problem.

 

Thoughts?

 

Thanks!

tl