This is an old revision of the document!
====== Debugging Windows iSCSI boot ====== {{ :nullmodem.jpeg?200x200|A null-modem cable}} To debug Windows iSCSI boot problems, you will need a second Windows machine running //windbg//, and a null-modem cable to connect the two machines together via their serial ports. The debugging machine does not need to be running the same version of Windows as the iSCSI-booting machine. This tutorial is designed to help you capture the information that will be needed to diagnose your iSCSI boot problems. ===== Preparing the machines ===== Connect the null-modem cable between the serial port of the debugging machine and the serial port of the iSCSI-booting machine. (If one or both of the machines do not have serial ports then you can use a FireWire cable instead, but these instructions assume that you are using a null-modem cable.) ===== Preparing the tools ===== Download the latest version of //windbg// as part of the Debugging Tools for Windows package from [[http://www.microsoft.com/whdc/devtools/debugging/installx86.mspx]] and install it onto the debugging machine. Locate the Microsoft iSCSI initiator (which you have already downloaded and installed on the iSCSI-booting machine), and install it onto the debugging machine. ===== Preparing the iSCSI-booting machine ===== On the debugging machine, start the Microsoft iSCSI Initiator utility (//Start// -> //All Programs// -> //Microsoft iSCSI Initiator// -> //Microsoft iSCSI Initiator//). Go to the //Discovery// tab and click on //Add// to add a new target portal. Enter the IP address or DNS name of your iSCSI target. {{ :screenshots:windbg_iscsi_add_target.png?400x300 |Adding an iSCSI target}} Go to the //Targets// tab. You should see a list of available iSCSI target IQNs. Select the one containing your iSCSI-bootable Windows disk, and click on //Log On// to connect to the target. {{ :screenshots:windbg_iscsi_connect_target.png?400x300 |Connecting to the iSCSI target}} The iSCSI-bootable Windows disk should now show up as an extra drive in //My Computer// on the debugging machine. Start up //Notepad// and open the file //X:\boot.ini// (where //X:// is the drive letter for the iSCSI-bootable Windows disk). You should see something like [boot loader] timeout=30 default=multi(0)disk(0)rdisk(0)partition(1)\WINDOWS [operating systems] multi(0)disk(0)rdisk(0)partition(1)\WINDOWS="Windows Server 2003" /fastdetect Duplicate the ''multi(0)disk(0)rdisk(0)partition(1)...'' line, and append the options /debug /debugport=com1 /baudrate=115200 /break (where //com1// is the serial port on the iSCSI-booting machine). The resulting file should look something like [boot loader] timeout=30 default=multi(0)disk(0)rdisk(0)partition(1)\WINDOWS [operating systems] multi(0)disk(0)rdisk(0)partition(1)\WINDOWS="Windows Server 2003" /fastdetect multi(0)disk(0)rdisk(0)partition(1)\WINDOWS="Windows Server 2003" /fastdetect /debug /debugport=com1 /baudrate=115200 /break Save this modified //boot.ini// file and close //Notepad//. Go back to the Microsoft iSCSI Initiator utility. On the //Targets// tab, select the one containing your iSCSI-bootable Windows disk, and click on //Details//. Tick the box next to the session identifier (which will be a long hex string such as //ffffffff810ca00c-4000013700000001//), and click on //Log off// to disconnect from the target. {{ :screenshots:windbg_iscsi_disconnect_target.png?400x300 |Disconnecting from the iSCSI target}} ===== Starting the iSCSI-booting machine ===== {{ :screenshots:windbg_select_debug.png?360x200|Windows boot menu}} Switch on the iSCSI-booting machine. You should reach the Windows boot menu offering you a choice such as Please select the operating system to start: Windows Server 2003 Windows Server 2003 [debugger enabled] Select the option which includes "[debugger enabled]", and press Enter to continue. You should see the screen go black, and the machine will appear to have frozen. This is because it is waiting to be controlled by the debugging machine. ===== Using the debugger ===== On the debugging machine, start the Windows debugger (//Start// -> //All Programs// -> //Debugging Tools for Windows// -> //WinDbg//). Start a kernel debugging session (//File// -> //Kernel Debug//). After a brief pause, you should see several messages appear, ending with Break instruction exception - code 80000003 (first chance) nt!DbgBreakPoint: 8081d971 cc int 3 {{ :screenshots:windbg_start.png?400x300 |Windows debugger attaching}} Allow the iSCSI-booting machine to continue using //Debug// -> //Go//. After several seconds, you should see iSCSI boot messages start to appear. For a successful boot, these will look something like IscsiBP-IscsiBPInitialize: enter IscsiBP iBF table found at FC564950 (FC545000) IscsiBP iBF table copied to E12B9A08 IscsiBP Found device \??\PCI#VEN_10EC&DEV_8139&SUBSYS_00000000&REV_20#3&13c0b0c5&0&18#{cac88484-7515-4c03-82e6-71a87abac361} IscsiBP Handle 800000B4 has IpAddress 10.254.254.15 Subnetmask Prefix - 24 is 00 ff ff ff ff ff 00 00 00 00 00 00 00 00 00 00 IscsiBP Handle 800000B4 has SubnetMask 255.255.255.0 IscsiBP Handle 800000B4 has DefaultGateway 10.254.254.2 IscsiBP nic 0 config -> 0 IscsiBP-IscsiBPInitialize: exit Found iSCSI Boot Parameters at 810CABA0, attempting iSCSI boot Allocating the Boot Connection Pool of size 30 Allocating Connection Pool 810C5F10 Allocating the Boot Connection Pool entry 0 SUCCESS Allocating the Boot Connection Pool entry 1 SUCCESS ... Waiting 60 seconds for boot disk iSpSystemWorkerThread: Entering iSpSystemWorkerThread: Coming out of wait iSpSystemWorkerThread: Start Logon Completion Thread 810C7008 Delay 10 seconds for Disk to arrive....... iSpSystemWorkerThread: After Logon Completion Thread Delay for Disk Complete, continue with boot NIC device \??\PCI#VEN_10EC&DEV_8139&SUBSYS_00000000&REV_20#3&13c0b0c5&0&18#{cac88484-7515-4c03-82e6-71a87abac361} has PCI bus type ISCSI Boot Status 0, (0, 0, 0, 0) If you are using [[:sanbootconf|sanbootconf]] then a successful boot will look something like iSCSI Boot Parameter Driver initialising Found iBFT at 9cfb0 OEM ID "FENSYS" OEM table ID "gPXE" Found iBFT Initiator 0: Flags = 0x3, valid, boot selected iSNS = 0.0.0.0 SLP = 0.0.0.0 Radius = 0.0.0.0, 0.0.0.0 Name = iqn.2000-01.org.etherboot:host_10_0_0_90.home Found iBFT NIC 0: Flags = 0x3, valid, boot selected IP = 10.0.0.90/24 Origin = 0 Gateway = 10.0.0.6 DNS = 10.0.0.6, 0.0.0.0 DHCP = 0.0.0.0 VLAN = 0000 MAC = 52:54:00:33:22:11 PCI = 00:03.0 Hostname = host_10_0_0_90.home Found NIC with MAC address 52:54:00:33:22:11 at "\??\PCI#VEN_10EC&DEV_8139&SUBSYS_00000000&REV_20#3&13c0b0c5&0&18#{ad498944-762f-11d0-8dcb-00c04fc3358c}\{7C0E3A80-E8E7-4B81-933E-C997BF973430}" iBFT NIC 0 is interface "\??\PCI#VEN_10EC&DEV_8139&SUBSYS_00000000&REV_20#3&13c0b0c5&0&18#{ad498944-762f-11d0-8dcb-00c04fc3358c}\{7C0E3A80-E8E7-4B81-933E-C997BF973430}" iBFT NIC 0 is PDO 8124DCF8 iBFT NIC 0 is NetCfgInstanceId "{7C0E3A80-E8E7-4B81-933E-C997BF973430}" Found iBFT target 0: Flags = 0x3, valid, boot selected IP = 10.0.0.6 Port = 3260 LUN = 0000-0000-0000-0000 CHAP type = 0 (None) NIC = 0 Name = iqn.2006-05.home.dolphin:winxp CHAP name = CHAP secret = Reverse CHAP name = Reverse CHAP secret = iSCSI Boot Parameter Driver initialisation complete iSCSI boot parameters requested iSCSI boot parameters requested Found iSCSI Boot Parameters at 81208008, attempting iSCSI boot Allocating the Boot Connection Pool of size 30 Allocating Connection Pool 81205850 Allocating the Boot Connection Pool entry 0 SUCCESS Allocating the Boot Connection Pool entry 1 SUCCESS ... Waiting 60 seconds for boot disk iSpSystemWorkerThread: Entering iSpSystemWorkerThread: Coming out of wait iSpSystemWorkerThread: Start Logon Completion Thread 81205008 Delay 10 seconds for Disk to arrive....... iSpSystemWorkerThread: After Logon Completion Thread Will notify BusChangeDetected Delay for Disk Complete, continue with boot ISCSI Boot Status 0, (0, 0, 0, 0) For an unsuccessful boot, the messages will end with a "Bugcheck Analysis"; this is the debugging equivalent of the Blue Screen of Death that occurs when the debugger is not attached. ===== Using the output from the debugger ===== Save the debugging transcript to a file using //Edit// -> //Write Window Text to File//. You can e-mail the saved debugging transcript to <etherboot-discuss@lists.sourceforge.net>, along with any other information that you think may be relevant. Please note that the debugging transcript is the most important information; if you provide a full description of your setup and your problem, but do not provide the debugging transcript, then we will probably not be able to solve your problem. You can also often find us in the #etherboot IRC channel on FreeNode. ====== Common problems ====== {{ :screenshots:win2k3_iscsi_install.png?400x300|Installing the Microsoft iSCSI initiator}} * Did you download and install the special boot-capable Microsoft iSCSI initiator as documented in [[:sanboot:winnt_iscsi|Adding iSCSI boot support to Microsoft Windows]]? If you installed the standard (non-boot-capable) Microsoft iSCSI initiator, then you will not be able to boot via iSCSI. * Did you install the checked build (the file ending in //-x86chk.exe//)? If you installed the free build (the file ending in //-x86fre.exe//, then you may not see any diagnostic messages. * When you installed the boot-capable Microsoft iSCSI initiator, did you tick the option to "Configure iSCSI Network Boot Support" and select the correct network card to be used for iSCSI boot? ((This option is not available on Windows XP; you will need to also download and install [[:sanboot:winnt_iscsi_sanbootconf|sanbootconf]] in order to boot via iSCSI.)) * If you are using Windows XP, did you download and install [[:sanboot:winnt_iscsi_sanbootconf|sanbootconf]] in addition to the boot-capable Microsoft iSCSI initiator? * If you are using iSCSI CHAP authentication, are all passwords at least 12 characters long? (The Microsoft iSCSI initiator will refuse to attempt authentication with shorter passwords.) ===== VMware-specific problems ===== VMware has a nasty habit of changing the PCI bus:dev.fn number assigned to the virtual NIC, which breaks the Windows driver bindings and so prevents a successful iSCSI boot. In most cases, you should ensure that the ''.vmx'' file for the VM that you are attempting to boot via iSCSI contains scsi0.present = "TRUE" even though there is no local hard disk attached.((VMware seems to assign PCI bus:dev.fn numbers dynamically. If you have a hard disk and NIC both present, then VMware will typically assign the hard disk controller as PCI device 00:10.0 and the NIC as PCI device 00:11.0. You perform the Windows installation within the VM, then detach the hard disk, transfer the contents of the virtual hard disk to your iSCSI target, and try to boot from it. Since you have detached the hard disk, VMware will now assign the NIC as PCI device 00:10.0, Windows will fail to enumerate the NIC, and the iSCSI boot will fail. By setting scsi0.present="TRUE" you can force VMware to create a hard disk controller (with no disks attached) as PCI device 00:10.0, and so return the NIC to being PCI device 00:11.0, as Windows expects.)) ====== Other useful techniques ====== ===== Displaying the boot driver list ===== The Windows Sysinternals [[http://live.sysinternals.com/LoadOrd.exe|LoadOrd.exe]] utility can be used to produce a list of boot drivers on your system, which should look something like: Start Group Tag Service Display Name ~~~~+~~~~~~~~~~~~~~~~~~~~~~~+~~~~+~~~~~~~~~+~~~~~~~~~~~~ Boot Boot Bus Extender 1 ACPI Microsoft ACPI Driver Boot Boot Bus Extender 2 PCI PCI Bus Driver Boot Boot Bus Extender 3 isapnp PnP ISA/EISA Bus Driver Boot Boot Bus Extender 5 ACPIEC Microsoft Embedded Controller Driver Boot System Bus Extender 6 Compbatt Microsoft Composite Battery Driver Boot System Bus Extender 4 IntelIde Boot System Bus Extender 1 Pcmcia Boot System Bus Extender 8 MountMgr Mount Point Manager Boot System Bus Extender 9 Ftdisk Volume Manager Driver Boot System Bus Extender 12 dmload Boot System Bus Extender 13 dmio Logical Disk Manager Driver Boot System Bus Extender 5* PartMgr Partition Manager Boot System Bus Extender n/a* VolSnap Boot NDIS Wrapper n/a* NDIS NDIS System Driver Boot NDIS 13 E1000 Intel(R) PRO/1000 Adapter Driver Boot Base 1 KSecDD Boot Base 20 iscsiboot Boot PNP_TDI 5 IPSec IPSEC driver Boot PNP_TDI 3 Gpc Generic Packet Classifier Boot PNP_TDI 4 Tcpip TCP/IP Protocol Driver Boot PNP_TDI 7 PSched QoS Packet Scheduler Boot SCSI miniport 25 atapi Standard IDE/ESDI Hard Disk Controller Boot SCSI Class 2 Disk Disk Driver Boot FSFilter Infrastructure 1 FltMgr FltMgr Boot Filter n/a* PxHelp20 PxHelp20 Boot n/a* n/a* BTKRNL Bluetooth Protocol Stack Boot n/a* n/a* iScsiPrt iScsiPort Driver Boot Network* 2* Mup Mup Boot n/a* n/a* ohci1394 Texas Instruments OHCI Compliant IEEE 1394 Host Controller You need to run //LoadOrd.exe// on the system that you will attempt to boot via iSCSI, **before** transferring the disk image to the iSCSI target.
