proliant ml370g2 and infortrend raid: device seen twice! Lun 33!

[Imanuel von Cube]

we have a 2node proliant cluster with dualchannel adaptec160 and an infortrend dualchannel scsi2ata external raid.

The Raidsystem has 2 logical drives with the follwing drive/channel/lun configuration

Channel 0 ID 0 Lun 0 ->logical Drive 1
Channel 0 ID 0 Lun 1 ->logical Drive 2
Channel 1 ID 0 Lun 0 ->logical Drive 1
Channel 1 ID 0 Lun 1 ->logical Drive 2

I've connected channel1 of server1 to channel1 of infortrend and terminated the channel
I've connected channel1 of server2 to channel2 of infortrend and terminated the channel

In Netware6.5 (and in 6.0) I see FOUR devices on Server 1:
0x0013: [V321-A3-D0:0] Compaq 64-Bit Slot 0 Port 1 ID 0 IFT IFT-7250F [HDD]
0x001E: [V321-A3-D0:1] Compaq 64-Bit Slot 0 Port 1 ID 0 LUN 1 IFT IFT-7 [HDD]
0x003D: [V321-A3-D0:20] Compaq 64-Bit Slot 0 Port 1 ID 0 LUN 32 IFT IFT [HDD]
0x003E: [V321-A3-D0:21] Compaq 64-Bit Slot 0 Port 1 ID 0 LUN 33 IFT IFT [HDD]

and also on Server 2
0x0011: [V321-A3-D0:0] Compaq 64-Bit Slot 0 Port 1 ID 0 IFT IFT-7250F [HDD]
0x0019: [V321-A3-D0:1] Compaq 64-Bit Slot 0 Port 1 ID 0 LUN 1 IFT IFT-7 [HDD]
0x0038: [V321-A3-D0:20] Compaq 64-Bit Slot 0 Port 1 ID 0 LUN 32 IFT IFT [HDD]
0x0039: [V321-A3-D0:21] Compaq 64-Bit Slot 0 Port 1 ID 0 LUN 33 IFT IFT [HDD]

So I guess, the raid somehow mirrors channel 2 to high LUNS on channel 1 and vice versa?

Except for looking messy that would not be a problem, BUT:

if I create a partition on V321-A3-D0:1 and a shared pool USER on that partition it appears
as a SECOND POOL WITH THE SAME NAME on V321-A3-D0:21

Those pools are not editable independently, if you delete/rename one, you'll delete/rename both,
althoug one USER can appear as ACTIVE and the other as INACTIVE.

THAT is a problem for failover scripts, which fail and the resource goes comatose.
The sympton only happens on partitions on logical drive 2 (with the LUN 1)
I have several other partitions/pools on logical drive 1 (LUN 0), which behave normally.

I think my problem will be fixed, If I learn how to teach either the Raidcontroller or the
adaptec inside the servers not to "see" the other channel.

starting adpt160m.ham without /LUNS returns only the first logical drive (useless for me) and only with Lun 0

adpt160m.ham lun_enable=03 /Luns returns four devices
Lun 0,1,32,33

adpt160m.ham lun_enable=03 returns only the first logical drive (useless for me) and only with Lun 0

It seems (and docu says that also) that lun_enable is not supported anymore on netware 6.5

-I have recreated the pool/partition from the scratch
-I did nss /poolverify and nss /poolrebuild
-I have NW6.5SP5 and the latest adpt160m.ham
-I have the latest firmware on the infortrend and on the proliants

Can someone help?
Imanuel von Cube
->Xpost to cluster-services.support.novell