[quote=“Todd Norcross, post: 2204, member: 629”][URL=‘http:///’]http://fogproject.org/forum/threads/compiling-a-custom-kernel-for-acer-iconia-tab-copied-from-old-forums.216/[/URL]

I have been trying these steps all afternoon, and have not been able to “make bzImage” I only get
[B]make: *** No rule to make target `bzImage’. Stop. o_O [/B]

I was also getting that message when I was trying to run “make xconfig.” Once it finally worked I went in, changed the following:

[B]Because the video on these devices is ATI Radeon HD 6250 (PCI Express based) I also deselected the “Device Drivers - Graphics Support - AGPGART (AGP Support)” and selected “Device Drivers - Graphics Support - Direct Rendering Manage (XFree86 4.1.0 and Higher DRI Support) - ATI Radeon”.[/B]

saved and exited my file tried to compile it and got the above stop error. I was working in the linux-3.2.11 kernel.[/quote]

Did you go through the instructions at [url]http://dinomite.net/2007/setting-up-ubuntu-for-building-kernel-modules/#.T48antUcuo5[/url] for getting the kernel source (not just headers) and updating the MAKEFILE with the EXTRAVERSION information?

I know it’s a lot of jumping back and forth between procedures, maybe some day it can be consolidated and made easier to follow.

All these warnings and error messages while booting the kernel can pretty much be ignored. The kernels that are available through the Fog install, and through the web UI upgrade process are all meant to fit as much hardware as possible. They include drivers for things like acer laptops with faulty fan controls (acerhdf), and trusted platform systems (tps65010) and the carillo mch whatever thingy. It tries to load every driver it knows about, and fails with a warning or error if the right hardware is not found, so you see all those messages.

It’s not really anything to worry about. The problem is that usually when the kernel fails, it fails silently without telling you what it was doing when it died. You have to run the debug mode to see how far it gets and what kills it. To run the debug mode, you have to set a master password in the Fog web UI.

was this error from the VM you are uploading, or the real machine you are trying to image?

Can you try the debug mode to gather info on where it’s failing?

also, is the real machine to which you are deploying in IDE or AHCI mode in the BIOS?

Can you go to fog and check out the /images/ directory for this image? Does it have the d1.mbr file? Does that file have size > 0? Do you have .img files for this image?

Well, it’s time to establish some baselines. Can you create a new VM, install Windows 7, reboot and take the image, and deploy that to a real machine, doing nothing special and only allowing the 2 windows partitions to be created ( 100mb boot and main OS partition)? Use single partition-re-sizable for the image type when uploading. If that doesn’t work, try multiple-partition, single disk. If that works, we can move through the steps you are following and see where it dies.

There is the ping/hostname resolve that happens in the web UI, which might be useful if it was modified to update that information in the database. Maybe in Fog 0.33, when we can create custom pages easier, something like this can be setup.

There might be something available if you use the Auto Log-Out feature. I do not, so I can’t see what kind of data is being stored in the table.

In vmWare, did you set the HDD type to IDE instead of SCSI/SATA? I’m not that familiar with vmWare, so I could be completely off track.

I’m using Ubuntu Server 10.04 LTS, updated, with Fog 0.32. I have a raid array of 6 drives, raid 5, in a logical disk presented to the system. I use LVM for all my partitions (/boot, /, swap, and /images). I have no issues with images such as this (49.8GB is largest image I have). This setup is about 3 months old now (new hardware).

clean install of Win7, without oem partitions? Did you choose the right image type single-part (resize) or multi-part, single disk? Did you sysprep?

Does the 2nd machine have multiple hard drives attached, or attached to different sata ports?

I only dump and restore the fog database, not the entire mysql instance. Minor changes to the commands used, documented well in the mysqldump section of the mysql website.

You can also tar+gzip your /opt/fog/, /var/www/fog/, and /tftpboot/ folders to preserve your fog installation. I host my images on a raid 5 array, so hopefully I do not lose them because of a disk failure. I have almost a 400GB of image files, so backing them up to another location takes a while.

Well,…

dhcp is easy to install and configure on Fog and you can have fog server IP addresses from an unused subnet for use when imaging, and then assign static IP addresses to the clients later on from a different subnet or block of addresses.

Say you use 192.168.1.0 for servers, 192.168.2.0 for workstations, and 192.168.3.0 for all other devices (printers, etc). Your network is basically flat, just using different subnets for different devices or buildings. As long as everyone is setup with a matching subnet mask and the same default gateway, you should be fine.

Install dhcp on Fog, configure it to serve addresses for 192.168.100.0/24. While PXE booting and imaging, the machines will not interfere with your static addressing scheme since they are on different subnets.

Another way to do this is to use MAC filtering, where the DHCP server only hands out addresses to known MACs, but this is a lot of administrative overhead. See: [url]http://www.fogproject.org/wiki/index.php?title=Integrating_FOG_into_an_Existing_Network_in_non_intrusive_mode[/url]

Another way is to use ProxyDHCP to only answer DHCP requests that are requesting PXE Boot info, and use a different subnet than your static IP’s. See: [url]http://www.fogproject.org/wiki/index.php?title=Using_FOG_with_an_unmodifiable_DHCP_server/_Using_FOG_with_no_DHCP_server[/url]

[quote=“Robin Commander, post: 2232, member: 64”]I’ve had exactly the same experience as described here when imaging with Windows 7, and solved it in exactly the same way (after much wailing and gnashing of teeth !)

I used the Sysprep from the WAIK kit on the Microsoft site, left the 100Mb partition in place and used Single disk, single partition function within FOG.

One thing that I did discover was that if I used an OEM (Panasonic) restore DVD to get the base build to CF-19 Toughbooks, whatever I did I couldn’t get the resulting image to work correctly when used as a basis for imaging to further computers. Still don’t know why, but I did find something on the web to the effect that Microsoft frown on this anyway. Once I swapped to using a volume license key version everything was sweet as a nut.

FOG sure is a godsend and is well worth the effort to keep persevering on with. Great job guys

Cheers
Robin[/quote]

Most OEM recovery discs create a reserved partition or partitions for doing a live recovery from the hard drive. This confuses Fog 0.32 somehow. I have fixed this booted to a GParted Live and removing the OEM recovery partition and resizing/rearranging the main OS partition, leaving alone the small extra win7 partition.

Also, if you install from the Win 7 disc, when it reboots and asks for your computer name/user name, hit the ctrl+shift+F10 to reboot into audit mode/sysprep mode. This lets you do all your install/configure stuff quite nicely.

Good to know Sk00t3r. I severely dislike the host name resolve/ping part of the web UI in Fog 0.32. I found there is an option in the fog settings to disable that completely. My hosts lists now finishes so quickly.

Please clarify. It works when you unicate upload/download, but not when you multicast?

Are the any requirements for keeping your deployment LAN separated from your Production LAN, like multicast filtering, security, etc?

Are the machines in your production LAN already able to interact with the Fog server for client updates, snap-ins, etc?

Joshua, try looking for the article about installing Linux on the PowerEdge 2650 server, which has a PERC4 raid controller. It may lead you to the answer eventually, I hope.

Sid, the custom kernel options are per HOST, or per GROUP and ONLY apply to tasks created for the host or group through the Web UI. If you boot to the fog menu and choose any of those options, it ignores your kernel settings for the host because the kernel to use is hard-coded into the menu file.

To clarify: If you can register and inventory with the default kernel (bzImage), you are good to specify a different kernel or kernel parameters for upload/download tasks. If you can’t register or inventory some machines with the default kernel, you’ll need to kernel swap or compile a custom kernel.

Being a long time Linux user but new to kernel compiling, I took this challenge a few months ago and documented it well on the old forums and copied it to the new forums.

[url]http://fogproject.org/forum/threads/compiling-a-custom-kernel-for-acer-iconia-tab-copied-from-old-forums.216/[/url]

does F10 bring up a boot menu.

Did you login as root, or sudo -i ?