Ghost Solution Suite

I been trying for a couple of weeks to resolve this without success.

I am running GSS2 on a Win2k3 SP1 server.

I have built a couple of PCs with different SOEs. Hardware is identical on both. M/B = Intel D945Gxx and single SATA HDD

I have successfully pushed the Console Client to both PCs.

When I try to run a create image task, evrything works fine until it gets to the "Clone" step. At this point, it will fail at any point - it has failed at anywhere between 1% and 99% on many many occasions. Sometimes when the clone fails, the PC reboots itself to PC-DOS and then again to WinXP without intervention. Sometimes - usually when it fails at 99% - it will just sit there and do nothing (I left it alone for 16 hours with no change). Sometimes when it locks at 99%, the Console will report the failure and either provide the error.txt file or not - there does not appear to be any consistency with the error file creation. Other times the Console reports that the task is still running until I cancel it - again sometimes I get an error file and other times I don't.

I eventually gave up on the Create Image tasks and tried a Push Image task only to have similar difficulties.

However, I did receive an error message that may explain at least some of the issues I am experiencing. It appears that at some random point during an image push or pull, the Client is disconnecting the Console: "Task reports error from the GhostCast Server - Connection Rest by Client." This is not always reported i.e it is not reported with the Create task but does come up with the Push tasks

Any assistance would be appreciated.

Hi Rob,

I am just jumping to a conclusion here, but this looks like something to do with SATA drive and NIC card sharing the same IRQ.

What network driver do you use? If it is universal network driver, could you select the 'Universal Packet driver V 2.0 (irqcfg)' ? This will try to resolve any conflicts it finds while in DOS.

If this does not work, please let us know what network template is used, and what kind of BIOS configuration options you have in the mother board for SATA controller.

Krish

Hi Krish

apologies for slow response.... really bad week.

> I am just jumping to a conclusion here, but this
> looks like something to do with SATA drive and NIC
> card sharing the same IRQ.

I had a look at this when it apppeared in one of the rare error logs with a recommendation to change the IRQ:

>> Found Mass storage & network controller sharing IRQ
>> Reckon you should move the irq to 9

However, I was not sure which to move. Does it matter?

> What network driver do you use? If it is universal
> network driver, could you select the 'Universal
> Packet driver V 2.0 (irqcfg)' ? This will try to
> resolve any conflicts it finds while in DOS.

I tried the V2.0 (irqcfg) but it didn't help. I am now using an Intel driver see below.

> If this does not work, please let us know what
> network template is used,

I have created a new Template for this PC model using the Intel Pro/1000 PM

This stopped the imaging process issues but highlighted another issue - During the PC-DOS boot process, when the DOS Mode Client tries to initialise the network, I get the following error message:

"Network initialisation failed: The DOS Mode Client cannot proceed. Unable to obtain IP address via DHCP."

I believe that it is due to this problem that I have been unable to push Client PCs into the AD using Configuration. I have worked around this by adding IP, DNS, WINS and Gateway addresses into the TCP/IP Settings in the Configuration for the PC.

Now I have found yet another problem. As I resolve each issue, more become evident. After the Task completes and reports 100% success and the PC reboots, when I log in for the first time, an error box appears with the message: "Ghostwalker cannot run on Windows NT. Please reboot to DOS." I am running GHSTWALK.exe /fni /sure - In Target OS in Execute Command


>and what kind of BIOS
> configuration options you have in the mother board
> for SATA controller.

Drive Configuration:
Use Auto Mode - Enable
ATA/IDE Mode - Enhanced
Configure SATA as - IDE
S.M.A.R.T. - Enable
HDD Pre-delay - 0
SATA Port 0 - WDC WD800JD-00LS - 80.0 GB

There are no other configuration options for SATA.

Regards
Rob

Hi Rob,

Yes, some weeks are bad - that's the life :-)

It does not matter what the value is usually, but the problem is some BIOSs would not give an interface to change the IRQ settings. The irqcfg.exe in the driver I suggested could do it automatically. As a matter of fact, it could be used with any driver by adding to config.sys, just like in the provided driver.

What happened when you use V2 (irqcfg) dirver? I forgot to mention that your clients PXE bios should be enabled for this to work. It does not need to boot to PXE, but should be enabled. Usually it is something like 'Enable LAN boot'

The issue of not been able to initialize the network driver is usually due to incompatible drivers. I am surprised if it worked after providing IP data(basically you use a static IP instead of DHCP - I presume that you have a DHCP server in the network).

The last error regarding Ghstwalk.exe is not an issue, the error message is correct. Ghstwalk.exe is an DOS application, and if you want to run it you need to select 'in Ghost partition' instead of 'target OS'. If select target OS, it will try to run it in Windows and generate the error message you got.

Anyway, why do you want to execute Ghstwalker? If you want to change the SID when cloning, just select the check box 'use ghost walker to change the SID' in clone tab of the task. Then it will automatically run Ghost walker with correct parameters and change the SID before boot to Windows.

Hope this helps, let us know the progress.

Krish

Hi Krish

I tested the way I have the Task set up and confirmed that my configuration of the entire task was successful repeatedly.

I then made the single change to the Client PC Template - from Intel to V2.0(IRQCFG)

When I use the V2.0(irqcfg) template, it reports the following:

... Resolve: (0x01, 0x00, 0x00): IRQ: 0x0B, Int#: 0x0A, Link: 0x60
Class: Network, , SubClass: Ethernet
Vendor: 0x8086, Device: 0x108c RoutePINS: 0x60, 0x61, 0x62, 0x63
Failed to assign new interrupt to network device.

It then continues with the task anyway until it gets to the end of the Ghost imaging session - completes to 100%.

The screen goes black and an abort message appears:

Abort (20016): An exception has occurred
Hit any key to continue... ST9Bad_alloc
hit <enter>

NOTE: as soon as I hit <enter> to acknowledge the ST9Bad_alloc message, the Console task fails. If I don't hit <enter> the task continues for extended but finite period and then dies.

After hitting <enter> a new message:appears:

Abort details will be dumped to .\GWALKERERR.txt - ok (y/N)? Y
hit <enter>

Drops back to the Client screen for about a second (client screen shows sending info to server and receiving acknowledgements - about 2 lines of each)

PC reboots through PC-DOS and sits at the client screen (as described above but sending and receiving continuosly until I turn the PC off.

Rob

Hi Krish

Sorry, forgot to confirm that Network Boot is set.to Enable.

Rob

Hi Rob,

Thanks for the update,

I think we can leave the network issue aside for the moment, since the imaging process completes. If it is working with a good speed, network setup should be OK.

The real issue is the bad_alloc message after completing the task. Once it fails, it just does not try to boot to Windows, since it is not sure if there is a bootable OS after task fails. Assuming clone task was OK, but it failed at Ghost Walker for some reason, if you press Ctrl+C at client, it should boot to Windows.

Do you execute any commands as part of this task? Specially Ghost Walker ?

If you have any command executions, could you remove them and try again?

If you don't execute Ghost Walker, could you clear the 'use ghost walker to change SID' check box and run the task again? Ghost Wlaker is used to change the SID, but it is OK to test without changing the SID to confirm if the issue is Ghost Walker related.

Krish

Hi Krish

Just a little info to help explain the use of the Intel Template:

When I use the Intel template, it reports the following:

Network initialisation failed: The DOS Mode Client cannot proceed.
Unable to obtain an IP address via DHCP.

When I set a static IP and Subnet Mask for the PC in the Configuration set, everything works using this Template. I have found that that is all I need - I don't need static DNS, WINS or Default Gateway entries.

Doing this has no effect on the V2.0(IRQCFG) template - it still fails as described.

Can you give me step by step instructions that will allow me to access the GWALKERR.txt file which I believe is written to the Virtual Partition? This file is supposed to be helpful to Tech Support in diagnosing this sort of problem - I am told.

Thanks
Rob

Hi Rob,

When client starts sending/receiving details to server, press Ctrl+X. Then it will go to DOS prompt. (Like C:\Ghost>)

If you give the DIR command, you can see if the file is there. You can copy it to a floppy or something.

After that, if you want to get in to the same screen, run

C:\Ghost> ngctdos

if you want to go to Windows again (assuming cloning was successful), just run the command

C:\Ghost> ghreboot


Rob, still I am really confused about your situation. Could you let me know the following? Sorry about asking this over and over, but I really need to know this to understand what's going on.

1. Do you have a DHCP server in the network? Do you use any Static IP addresses in machines?
2. Do you execute ghstwalk.exe as the part of the task? If yes, what happens if you remove the command execution?
3. If not, could you try clearing the checkbox 'use Ghost walker to change the SID' in task properties, clone tab?

Krish

Hi Krish

First, thanks very much for the instructions - I'll grab the file on the next failure.

1. Yes we have DHCP servers throughout our networ which spans several domains.

2. Yes, I use GHSTWAK.exe /fni /sure from the Execute Command tool. I have found it to be more reliable than the check box due to the st9bad_alloc errors I have been getting - I wasn't sure what was causing them so once I had a task that worked - I stuck with it.

I have just tried a Task without any commands and no check box. I used the V2.0(IRQCFG) template and the Task completed successfully.

3. I am now going to try the same Task but this time using the check box.

Stand by :-)

Rob

Hi Krish

Well.....

Tried and failed.

Used the check box and the st9bad_alloc error returned.

I have captured the gwalkerr.txt file - not as informative as I thought...

=====================================
Ghost SID Update Utility - Abort dump
-------------------------------------

Date : Thu Feb 1 18:01:06 2007
Error No. : 20016
Message : An exception occurred.
St9bad_alloc

Version : 11.0.0.1502 Build 1502

-------------------------------------

Memory Manager Details
----------------------

Heap Available at start : 0
Current Heap Available : 0

----------------------

Program Call Stack
Globals::dumpDetails
Globals::abort

Rob

Hi Krish

Just tried the same Task again - this time using Execute Command "GHSTWALK.exe /fni /sure" and without the check box.

The Task completed successfully (except for the Client Inventory - but who cares?)

This points directly at the Check box implementation of Ghostwalker.

Any ideas on where to go from here? Is there any more testing you would like me to do?

Rob

Hi Krish

Now I am really intrigued.

I ran a second Task same as the last one and the Client Inventory failed again.

I ran a straight Refresh Congiguration and Refresh Inventory after applying the Intel Template to the Client PC. The Refresh Configuration step worked but the Refresh Inventory failed again with the same error message:

"The client inventory failed - check that the client has the correct service packs and updates installed to enable WMI. See the chapter entitled 'Using Client Inventory' in the manual for more information."

This error has never occurred before - well, not until the last couple of task runs.

I just ran the same clone task but after applying the Intel Template to the client PC. The Inventory step failed again.

Weird.

Rob

Hi Rob,

Thanks for all the messages with lot of details.

I think the main issue you had was Ghost Walker failing in DOS when try to change the SID. It seems that �fni switch solves that problem. Currently, it is not possible to pass parameters for Ghost Walker when you run it by ticking the check box. But since it is working satisfactorily when used as a command execution, it is OK. Please confirm that actually it runs and SID is changed in the client machine. This should be a hardware related issue, could you please give us the hardware details of your clients?

The second problem is inventory not getting collected. What OSs are your clients? By any chance, are they NT? If it is so, you need to install the WMI support apps since NT does not support them by default. They are available in the installation CD.


Krish

Hi Krish

I have done a lot of testing of GSS over the last 8 months. I came to the conclusion back when I was testing GSS1.1 that the Ghostwalker checkbox option does not work with clients that have PCI-Express. I definitely suggested this back then to the Support people and (I think) in another thread but I guess no-one really understood what I was getting at. I have done the same testing with GSS2.0 over the last couple of months and have come up with the same results. The issue has finally been escalated so that it can be looked at from the coding viewpoint and hopefully it will be resolved fairly soon.

The Execute Command function using the command line and switches:

"GHSTWALK.exe /fni /sure"

Works fine with any client I have tried and definitely changes the SIDs.

The second issue: failing on Refresh Inventory - All my OSs are either Win2k SP4 or XP Pro SP2. I have just run another Push Image task with the Refresh Inventory added in and it worked. Not sure why the previous attempts failed but I will continue to test over the next couple of days - I'll let you know.

Rob.

Hi Rob,

Thanks a lot for the information, specially for the relationship with PCI Express and Ghost Walker.

We tried to re-produce this issue using machines with PCI Express, but it did not happen.

In your machines, is there a card in PCE Express bus? If so, what card is it?

What brand/model of machines do you use?

Krish

Hi Krish

Well it appears that the fault does not lie with GW completely. Part of the problem was the Uuniversal Packet Driver v2.0 (irqcfg) (UPD) and the other part appears to have been my installation of GSS2.

The UPD appears to cause the st9bad_alloc error message when the system tries to run GW either from the checkbox or from the Execute Command.

Even though I had changed the Client properties of the Machines to use the Intel Pro 1000 PM template that I have created, for some reason, it wasn't sticking and would swap back to the UPD if I switched between the Execute Command to run GW and the GW check box - can't work out why it did this. Although, this explains why I had success with the Execute Command but failures when using the check box.

I removed and re-installed the GSS2 on the server and manually installed the OS on the 6 test PCs and pushed the client to each from the Console. I started again from scratch, basically.

I set up and ran a Push Image Task using the check box to 6 PCs on a schedule every 30 minutes for 7.5 hours. GW did not fail once. I have not used the Execute Command GW since I re-installed GSS2.

The schedule ran the task 15 times: 6 PCs done as a group:

1 Task failed - "the client PCs (all 6 of them) rebooted to the Target OS during the task." This occurred during the Clone step. No idea why.

1 Task returned 6 warnings - 5 of the PCs were unable to join the Domain: "The specified domain either does not exist or could not be contacted."

9 Tasks returned 5 warnings - 5 of the PCs were unable to join the Domain: "The specified domain either does not exist or could not be contacted." Each time a different PC joined the Domain: PC 1,6,4,5,4,3,5,4,3.

1 Tasks returned 4 warnings - 4 of the PCs were unable to join the Domain: "The specified domain either does not exist or could not be contacted." PCs 4 & 5 joined.

2 Tasks returned 3 warnings - 3 of the PCs were unable to join the Domain: "The specified domain either does not exist or could not be contacted." PCs 4,5 & 6 joined in both tasks.

1 Tasks returned 2 warnings - 2 of the PCs were unable to join the Domain: "The specified domain either does not exist or could not be contacted." PCs 1,2,3 & 4 Joined the Domain.

Rob

Hi Rob,

Thanks for the post. The first failure "client PCs rebooted to the Target OS during the task." usually happens when the machine boot to Windows after Clone session, without connecting to Console using DOS client. Basically, Console see that the clone is over, but some how DOS client does not contact Console, then Windows Client contact console. If you interrupt the process after cloning and run ghreboot, you can generate this message. And it seems that the clients continue to work after that. Was this the first task?

Domain joining issue seems to be very similar to the situation when DC is not registered in DNS. I have seen this some times, it looks like the entry is not created some times during installaiton.

Could you run dcdiag from the resource kit and see if it complains about anything? If it is what I suspect, dcdiag should find it.

Krish

Hi Krish

The reboot to target OS error occurred in the second task.

The second issue - PCs not joining the domain:

I'm not sure that a missing DNS entry would cause this issue as it was successful for at least 1 PC and up to 4 PCs in each of the 15 tasks. If the DNS entry was missing, would this still be possible? Also, would the AD account stubs be able to be created successfully in the "Create Account" step of the Tasks? These are 100% successful in all of the tasks.

I keep coming back to PCI Express. When the PCs boot up to DOS at the beginning of the task, when it gets to the point of acquiring an IP address from the DHCP server, it fails every time. (please note that this occurs with both the Intel Pro 1000 PM Template and the Universal Packet Driver v2 irqcfg Template)

However, the task continues to completeion and success - apart from the "Warnings" about not joining the Domain.

I'll run the multiple instances of the task again over a period of time with static IP addresses and let you know the result.

Rob

Hi Rob,

Thanks for the response.

I cannot think of any reason for the error in second task, but will keep an eye on it. I will see if there are any other scenarios that could produce the same result.

Regarding the domain join situation, create account and actually joining the machine take different paths. The first one is an already authenticated connection but second one is creating a new connection. If it is not too much of a problem, it would be really useful to see what dcdiag reports. It is included in the resource kit, and you can run it from command line from any member of the domain.

Krish

Hi Krish

What version of Windows server resource kit has dcdiag? I have tried win Server 2003 but its not in the resource kit.

Sorry Rob, it is not the resource kit, but the support tools. You can download it from Microsoft

http://www.microsoft.com/downloads/details.aspx?familyid=6EC50B78-8BE1-4E81-B3BE-4E7AC4F0912D&displaylang=en

Krish

Hi Krish

Below are the results of running dcdiag and dcdiagdns:

Domain Controller Diagnosis

Performing initial setup:
Done gathering initial info.

Doing initial required tests

Testing server: Bunbury\WSBY000DC
Starting test: Connectivity
......................... WSBY000DC passed test Connectivity

Doing primary tests

Testing server: Bunbury\WSBY000DC
Starting test: Replications
......................... WSBY000DC passed test Replications
Starting test: NCSecDesc
......................... WSBY000DC passed test NCSecDesc
Starting test: NetLogons
......................... WSBY000DC passed test NetLogons
Starting test: Advertising
......................... WSBY000DC passed test Advertising
Starting test: KnowsOfRoleHolders
......................... WSBY000DC passed test KnowsOfRoleHolders
Starting test: RidManager
......................... WSBY000DC passed test RidManager
Starting test: MachineAccount
......................... WSBY000DC passed test MachineAccount
Starting test: Services
......................... WSBY000DC passed test Services
Starting test: ObjectsReplicated
......................... WSBY000DC passed test ObjectsReplicated
Starting test: frssysvol
......................... WSBY000DC passed test frssysvol
Starting test: frsevent
......................... WSBY000DC passed test frsevent
Starting test: kccevent
......................... WSBY000DC passed test kccevent
Starting test: systemlog
......................... WSBY000DC passed test systemlog
Starting test: VerifyReferences
......................... WSBY000DC passed test VerifyReferences

Running partition tests on : ForestDnsZones
Starting test: CrossRefValidation
......................... ForestDnsZones passed test CrossRefValidation
Starting test: CheckSDRefDom
......................... ForestDnsZones passed test CheckSDRefDom

Running partition tests on : DomainDnsZones
Starting test: CrossRefValidation
......................... DomainDnsZones passed test CrossRefValidation
Starting test: CheckSDRefDom
......................... DomainDnsZones passed test CheckSDRefDom

Running partition tests on : Schema
Starting test: CrossRefValidation
......................... Schema passed test CrossRefValidation
Starting test: CheckSDRefDom
......................... Schema passed test CheckSDRefDom

Running partition tests on : Configuration
Starting test: CrossRefValidation
......................... Configuration passed test CrossRefValidation
Starting test: CheckSDRefDom
......................... Configuration passed test CheckSDRefDom

Running partition tests on : hdwa
Starting test: CrossRefValidation
......................... hdwa passed test CrossRefValidation
Starting test: CheckSDRefDom
......................... hdwa passed test CheckSDRefDom

Running enterprise tests on : hdwa.health.wa.gov.au
Starting test: Intersite
......................... hdwa.health.wa.gov.au passed test Intersite
Starting test: FsmoCheck
......................... hdwa.health.wa.gov.au passed test FsmoCheck



Domain Controller Diagnosis

Performing initial setup:
Done gathering initial info.

Doing initial required tests

Testing server: Bunbury\WSBY000DC
Starting test: Connectivity
......................... WSBY000DC passed test Connectivity

Doing primary tests

Testing server: Bunbury\WSBY000DC

Running partition tests on : ForestDnsZones

Running partition tests on : DomainDnsZones

Running partition tests on : Schema

Running partition tests on : Configuration

Running partition tests on : hdwa

Running enterprise tests on : hdwa.health.wa.gov.au
Starting test: DNS
Test results for domain controllers:

DC: WSBY000DC.hdwa.health.wa.gov.au
Domain: hdwa.health.wa.gov.au


TEST: Records registration (RReg)
Network Adapter HP Network Teaming Virtual Miniport Driver:
Warning: Missing GC SRV record at DNS server 165.118.171.37 :
_ldap._tcp.gc._msdcs.hdwa.health.wa.gov.au
(Ignore the error if DNSAvoidRegisterRecord registry key or its Group Policy
has been configured to prevent registration of this Record.)

Error: Record registrations cannot be found for all the network adapters

Summary of DNS test results:

Auth Basc Forw Del Dyn RReg Ext
________________________________________________________________
Domain: hdwa.health.wa.gov.au
WSBY000DC PASS PASS PASS PASS PASS FAIL n/a

......................... hdwa.health.wa.gov.au failed test DNS

The single failure was due to GP and so was expected.

Rob

Hi Rob,

Thanks for running it, I would expect it to fail at test connectivity if it is the problem I was suspecting. Seems it is fine.

By looking at the log, seems you have a multiple domain controller environment, and the issue could be related to it.

There is a log file create at the clients \windows\debug folder, named 'netsetup.log'. Could you send the log files from a successful and an unsuccessful client?

Since this information could be sensitive, would you like to email them to me at the following address?
gssforummoderator1@symantec.com

Krish

I was experiencing the ghostwalker abort 20016 error as well. I then changed the universal network driver in the ghost boot partition to reflect my NIC (which is an intel Pro100) and everything worked fine. Just an FYI.