iSCSI Configuration

[clahti]

Hello all:

I have a freenas box that was serving iSCSI to 12 ESXi 5.0 hosts. It was working perfectly, no issues. The previous working configuration was:

ISCSI-1 192.168.1.x/24, jumbo frames, mpio
ISCSI-2 192.168.2.x/24, jumbo frames, mpio

Here was the interface assignments on the freenas box

eth1 192.168.1.21
eth2 192.168.1.22
eth3 192.168.2.21
eth4 192.168.2.22

This was all working fine. For a different reason (an unnamed vendor storage product tipping over), I had to reconfigure my iSCSI networks so that I could have half the cluster talk on the original two networks via a single interface, and the other half on the other two new networks. I now have busted this out to four discreet iSCSI networks as follows:

ISCSI-1 192.168.1.x/24, jumbo frames, mpio
ISCSI-2 192.168.2.x/24, jumbo frames, mpio
ISCSI-3 192.168.3.x/24, jumbo frames, mpio
ISCSI-4 192.168.4.x/24, jumbo frames, mpio

Here are the new interface assignments on the freenas box:

eth1 192.168.1.21
eth2 192.168.2.21
eth3 192.168.3.21
eth4 192.168.4.21

So the first six ESXi hosts are fine, two less paths per ISCSI-1 and ISCSI-2 networks and are happy. The problem is that I have added the two new networks using the same configuration, just added to it. The other six ESXi hosts can ping the new network interfaces fine, but cannot discover the LUNs for some unknown reason. Here are the exact settings:

Portals
192.168.1.21:3260 192.168.2.21:3260 192.168.3.21:3260 192.168.4.21:3260

Initiators
192.168.1.0/24 192.168.2.0/24 192.168.3.0/24 192.168.4.0/24

Authorized Access
no users defined

Targets
vmdisk1 vmdisk1 Disk

Device Extents
extent0 ZVOL zvol/pool0/zvol0

File Extents
No extents defined

Associated Targets
vmdisk1 extent0


Again to reiterate the first two iSCSI networks still work fine, the new two do not. Everything is pingable where expected, jumbo frames turned on everywhere, etc. Am I missing something simple here?

Thanks!

 

[clahti]

also changing Initiators to ALL did not solve the problem either. Do I need to bounce the iSCSI service to effect this change? That would be bad as I have many production VMs running on the first six ESXi hosts that can see the LUN from the original iscsi networks.