[gPXE] Error at booting VM from iSCSI: Connection reset (0x0f0d6039)

[Gustav Brock]

Hi all

Michael came with the magic clue: A bridge.

All firewalls are shut down, no antivirus is installed - still from the packet capture Michael could read that the connection happened but was rejected by the target.
Setting up a machine with two NICs in bridge could reveal why:

http://ipxe.org/howto/pcap (which he just created)

But I didn't reach to arrange for this.

Combined with an entry in the event log about a "duplicate address found" (which should not be the case) that reminded me, that I have had a bridge configured on the VM host between two physical NICs. It was disabled but have had the same address as the physical NIC currently in use by the VM to boot.
Disabling, however, wasn't enough; the mini driver had to be uninstalled (removed) from the device list.

That done, the VM booted at once! Complete with 64 bit, two CPUs and one SCSI drive. Stupid me - that took me two days and a clue from Michael to find out. So now I look forward to a relaxing weekend.

Let me at this moment express my gratitude to the excellent gPXE project and the great wiki and the people behind. I have studied many pages over and over. It is a major work of great value for people like me trying to find the ins and outs and what to do and when.

Have a nice weekend!

/gustav


>>> "Gustav Brock" <gustav at cactus.dk> 22-10-2010 00:00 >>>
Hi all

I created a gPXE iso to boot a Win2003 64-bit virtual machine in a VMware 2.0.2 host running on Win2008 64-bit.
The target is created with StarWind running on a physical Win2003 32-bit machine. No password, chap, or iSec. Windows Firewalls are off.

On the Win2008 host the target can be connected with the MS initiator.
If the virtual Win2003 machine is booted from a normal vmd harddisk, the target can be connected with the MS initiator.

However, if the virtual Win2003 machine in booted with the gPXE iso, the target cannot be connected. The command and error is:

sanboot iscsi:192.168.1.151::::iqn.2008-08.com.starwindsoftware:test
Could not initialise iSCSI device: Connection reset (0x0f0d6039)
Could not boot from iscsi:192.168.1.151::::iqn.2008-08.com.starwindsoftware:test: Connection reset (0x0f0d6039)

How to overcome this? My feeling is that somehow the target cannot "see" or "call back" the virtual NIC in the VM but why?

The virtual NIC is e1000 which is configured via the VMware network setup to bridge to one physical Intel PRO/1000 GT adapter with address 192.168.1.211. Internet Connection Sharing (ICS) is off.
At first I tried with the default boot script using DHCP. But as I'm in an Active Directory, I have and will have to use my MS DHCP servers. I noticed the setting for Root Path that has to be done but that didn't help. Finally, and as this is supposed to be a server, I changed the script to fixed IP and removed the DHCP lookup:

#!gpxe
echo "Greetings! Hit Ctrl-C to bail out."
sleep 5
ifopen net0
set net0/ip 192.168.1.223
set net0/netmask 255.255.255.0
set net0/gateway 192.168.1.1
set net0/dns 192.168.1.8
set keep-san 1
sanboot iscsi:192.168.1.151::::iqn.2008-08.com.starwindsoftware:test
boot

but that didn't change anything. Same error.
If I stop the script with Ctrl+B, I can from my workstation ping both 151 and 223 and the host NIC at 211.

Any clues or what to debug, please? Googling doesn't reveal anything.
I should add that my previous experience with gPXE is zero so every hint is appreciated.

/gustav