windows 10: deployment in different parts of the network fails

[fogusnew]

Hi everyone,
I manage a small network with ~ 80 win PCs. Fog has been a real challenge and tremendous help (before and after it worked). However: I’m still learning.
So I came across a very weird behavious, and I’d like to have some inputs on this issue.

The schools network is basically devided in two big rooms. One (locally) very close to the physical fog server, connected with 100GBit, another - connected through an other single 1GBit / 100Mbit network switch - a bit further away.

I was able to successfully create multicast sessions in the second room, deploying a win 10 1909 image to 25 PCs, without any problems - until … Windows started to configure itself with several reboots. About 15 of he 25 machines got stuck with the message:

Windows setup could not configure to run on this computer’s hardware

The other 10 machines worked well. All the PCs have the same hardware (HP), and the testing of the imageing process worked well.
So I was desperate and tried to image some machines one by one: Unfortunately with the same result. After several hours of research for this kind of error message I decided to give up.
Well, one last try I was willing to make, to find the error. I took one machine to the other room and started the imaging process.
Guess what: It worked without any error. So the soultion was to move each single computer to the other room and do the imaging. Every single machine worked flawlessly.

So how crazy is that?

Any comments? I’d really like to hear your oppinions.

I’m sure it is a windows issue, but it may well be caused by the different network connection. So I don’t know where to put this question…

However: any ideas how to track this down and solve it? I’d like to do imaging without moving machines…

Thanks
Chris

[george1421]

Multicasting in general is a beast of its own, not specifically related to FOG. So you have 2 rooms one images correctly and the second doesn’t via multicast. If you were to take one of the computers in the room that fails to image and send a unicast image to that computer does it image correctly?

Also when you send the multicast image out do you send the image to both rooms or computers in both rooms at the same time or only a single room at a time?

[george1421]

Also you might want to add some clarity to your original post. I think the 100Gb link is a type-o. Also which room is which in that is it the room with the 100Mb/s switch the failing room or the one with the… maybe 10Gb uplink?

[fogusnew]

@george1421
First: Thanks for pickung up the topic!

To your questions:

• The machines of the “failing room” that failed to come to life via multicast also failed to come alive via single cast. Only if I moved them to the other room the windows config came to and end.
• I multicast only one room (the failing one). But 10 machines worked very well.

In fact, the deployment by fog seemed to work all the time. But the Windows config after the imaging process got stuck on several PCs.

[fogusnew]

@george1421 said in windows 10: deployment in different parts of the network fails:

Also you might want to add some clarity to your original post. I think the 100Gb link is a type-o. Also which room is which in that is it the room with the 100Mb/s switch the failing room or the one with the… maybe 10Gb uplink?

Alright, sorry that was not clear.
Room 1 (the working one): Clients -> 100Mb/s -> Switch (Room1) -> 100Mb/S -> FOG Server
Room 2 (the failing one): Clients -> 100Mb/s -> Switch (Room2) -> 1GB/s Uplink -> Switch (Room1) -> 100Mb/s -> FOG Server

I know that is not an optimal setup… but funds are limited and the network is old

[george1421]

@fogusnew Ok so room 1 has a 100Mb/s switch and then you have room 2 daisy chained off of room 1. Now you mentioned that room 2 is connected by a 1Gb link, how did you do this? Did you connect the uplink ports on both switches together and that is how you connected them. For multicasting that 100Mb/s switch may not be fast enough depending on the number of multicast clients. (just thinking about the network layout at the moment).

Are either of those 100Mb/s switches managed switches where they have a management interface you can look at?

[fogusnew]

@george1421
yes, they are manageable, but I did not create any special setting. All standard, they just run as out of the box. As far as I remember (I’m not on site at the moment) - they have different ports, most 100Mb/s, but at least one 1GB/s port.each. So that is why the different speeds exist. But at the end the clients of both rooms are connected via 100Mb/s. Does it actually matter which speed is possible between the two switches?

My suspicion is different: After deploying, windows needs to restart at least two times (hardware, get into the domain, etc…) . Cloud it be that the FOG client forces Windows to restart too early, maybe before Windows got all the hardware drivers from online?

[Daniel Miller]

@fogusnew If you have the fog client installed in a sysprepped image you are deploying, the fogservice should be disabled prior to the sysprep.
https://wiki.fogproject.org/wiki/index.php/FOG_Client#FOG_Client_with_Sysprep
I wouldn’t think this would present intermittently in one room but not another, but I have seen weirder things happen. You might want to recapture with the fog service disabled and the service start configured in SetupComplete.cmd and reattempt the deployment and see how things behave then.

[george1421]

@fogusnew said in windows 10: deployment in different parts of the network fails:

Cloud it be that the FOG client forces Windows to restart too early, maybe before Windows got all the hardware drivers from online?

As in the link that Daniel mentioned, when you installed the fog client did you set it to be disabled in the golden image, and then use the setupcomplete.cmd batch file to turn it back on? If not this IS the issue. If the fog client does start too early in the OOBE process it will reboot the computer before OOBE completes making a botched install. This is beyond when fog imaging is involved.

[fogusnew]

Hi folks!
Thanks for taking time to look into this.
And yes, you’re right about the FOG client. It’s inside the image and the service is activated from the beginnen. While researching I came across the mentioned link after I experienced those problems.
It was just such a weird behavior I was unsure and did not try out the delayed strting of the service.
I’m going to modify my image in the next weeks. If the issue continues I’m going to open a new thread.
Thanks a lot!
Great help from a great community!

Chris
P.S. How to mark this issue as “solved”?