Posts made by ahachenberg

Thanks Chad,

I found this - [url]http://www.fogproject.org/wiki/index.php/FOGUserGuide#Full_Listing_of_Ports_used_by_FOG_server_and_client[/url]

Curious which of those would be required for snapin management/deployment.

Yeah, you read that correctly, I’m asking about modem connected clients in the year 2013. This requirement is upon me.

I know that my options are as follows to get a remotely connected client (no VPNs or WANs) are as follows:

1. Assign the fog server a public IP and domain name and point clients to it. But i dont want to do that just yet because i’m not read yfor all the security hardening.
2. Use something like SSH and create a tunnel from the clients to the server. but then again i guess i’d need a public IP or some port forwarding.

With that said, what ports would I need to forward to the Fog server? I’m not expecting to deploy giant packages or images to modem connected clients. Just the tiny updates and scripts plus an occasional screen share.

I’ll likely figure this out on my own but figure it’s worth posting here for any other poor schmoe that finds himself in the same requirement.

Try the wiki here:
[url]http://www.fogproject.org/wiki/index.php/Wake_On_LAN_Bug_workaround[/url]
[url]http://www.fogproject.org/wiki/index.php/How_to_use_the_wakeonlan_perl_script_instead_of_etherwake.exe[/url]

I’d also check to see that etherwake resides where fog is looking for it. Perhaps it’s somewhere else in Ubuntu 12.04. But, I can add that my Fog server is Ubuntu 11 and is a VM ontop of esxi 5 and WoL is working. Some of my Dell laptops would refuse to WoL and only do so after I applied default settings to BIOS and then reconfigure BIOS to enable WoL (and PXE).

And, beyond all of that, check you Ubuntu server’s syslog/messages after attempting to WoL. Maybe it’ll tell you something.

[IMG]http://www.thesharkguys.com/wp-content/uploads/2012/10/Borat_Great_Success.jpg[/IMG]

GREAT SUCCESS!!!

My two problems were (well, one really, RTFM):

1. [URL=‘http://www.fogproject.org/wiki/index.php/Managing_FOG#Active_Directory_Integration’]Follow this article[/URL] to configure your hostname changer service and then edit the config.ini where you will generate encrypted passwords
2. Follow [URL=‘http://support.microsoft.com/kb/818091’]M$ KB818091[/URL] to setup the account that will join nodes to AD

The renaming bit is working now after clearing up my encrypted password problem as noted in my last post. but still the deployed machines fail to join AD.

So now i turn to AD to ensure that I’ve crafted the service account correctly that is used to join hosts to AD. I found this article and followed it. I had previously not configured my service account correctly.

[url]http://support.microsoft.com/kb/818091[/url]

Trying the job again…

I went back to the wiki where i learned howto setup the passkey on the hostnamechanger ([url]http://www.fogproject.org/wiki/index.php/Managing_FOG#Active_Directory_Integration[/url])

I forgot that i needed to update the fogcrypt/etc/config.ini to include the new passkey. I just added that in and reproduced the encryped password for my service account which is used to join hosts to AD. Trying it out now…

in further testing this morning it seems that my test laptop is only renamed to the registered fog name when i chose not to join it to the domain. If i provide the AD settings the renaming task seemingly does not run nor does the machine join the domain.

[edit] i wonder if fog is trying to change the hostname and join AD at the same time. when we do this manually windows tells us to do so one at a time with reboots between.

I’ve made a good start with Fog and have run into a snag. The first set of images were all non-sysprepped but had fogservice included and would execute the renaming task successfully. They were not set to join a domain and no snap-ins selected.

I built my first Win7 sysprepped image and went through a series of edits and tweaks, mostly to my unattend.xml. Somewhere along the way my test laptop stopped taking the registered name in fog and retained the unattend.xml assigned name.

I did some searching and found this forum post [url]http://fogproject.org/forum/threads/windows-7-not-renaming-not-joining-to-ad.263/[/url]

So, I’m not a lone but I’m hoping for another solution. Are there any log files on the server that would provide hints?

My sequence of events for the image creation is as follows:
Install windows
enter audit mode with ctrl+shft+f3
apply customizations, driver packs, and windows updates
install fogservice
place setupcomplete.cmd and unattend.xml in respective locations
run chkdsk /f
upload image to fog server with label ‘presysprep’
boot that image again
run fogprep
run sysprep from cli
upload image to fog server without letting it boot into windows

Yep, this ([url]http://sourceforge.net/projects/freeghost/files/Kernels/[/url]) was my answer. I simply ran the update from within the fog server web admin page and my Dell Latitude e6530 was able to boot and register.

I’m still curious about building kernels with new drivers but at least i can learn that without pressure of getting this laptop going.

I did some searching in the forum here prior to posting. I saw that some other [URL=‘http://sourceforge.net/projects/freeghost/files/Kernels/’]kernels can be found here[/URL] and I’m gonna try that approach as a work around to my problem. For curiosity, here’s my problem;

I’m trying to add the latest Intel driver for the e1000e into my kernel and it’s throwing this error, but still continues and produces a bzImage. In testing, the laptop fails to bring up the NIC. I’m expecting the driver should work because booting a live Ubuntu 12.10 ISO, the Dell latitude e6530 is up and running with the NIC using the e1000e driver. I’m guessing that the latest Intel driver (on their site) is the same as what’s included on the Ubuntu 12.10 ISO. The error message below seems painfully obvious that i need to switch my builds machine from SMP to UP, but I’m not quite sure how to do that.

[CODE]/usr/src/linux-source-2.6.32/drivers/net/e1000e/Makefile:200: ***
/usr/src/linux-source-2.6.32/drivers/net/e1000e/Makefile:205: *** Warning: kernel source configuration (UP)
/usr/src/linux-source-2.6.32/drivers/net/e1000e/Makefile:206: *** does not match running kernel (SMP)
/usr/src/linux-source-2.6.32/drivers/net/e1000e/Makefile:208: *** Continuing with build,
/usr/src/linux-source-2.6.32/drivers/net/e1000e/Makefile:209: *** resulting driver may not be what you want
/usr/src/linux-source-2.6.32/drivers/net/e1000e/Makefile:210: ***[/CODE]

I attempted changing [CODE]CONFIG_LOCALVERSION=“SMP”[/CODE] but that didnt seem to change anything.

Current builds server:
Ubuntu
Linux AusFog01 2.6.32-45-server #102-Ubuntu SMP Wed Jan 2 22:53:00 UTC 2013 x86_64 GNU/Linux
Fog .32

Sources I’ve been using:
[url]http://www.cs.csubak.edu/~steve/cs216/Kernel-Build-HOWTO.html[/url]
[url]http://alien.slackbook.org/dokuwiki/doku.php?id=linux:kernelbuilding[/url]

Heck yes. I’m able to boot the fog image and blast images to the VM. Be it paravirtual or LSI Logic SAS. Havent tested the other modes yet tho.

RTFM, I tell myself…

This was the reason for my last two failures with init and CPU arch:

copied from the WIKI,
[SIZE=5][B][FONT=sans-serif][SIZE=19px][COLOR=#000000]Note[/COLOR][/SIZE][/FONT][/B][/SIZE]

[FONT=sans-serif][COLOR=#000000]When building on x86_64 systems you may need to specify that you are building for an x86 target architecture.[/COLOR][/FONT]
[FONT=sans-serif][COLOR=#000000][FONT=monospace]make ARCH=i386 menuconfig[/FONT][/COLOR][/FONT]
[FONT=sans-serif][COLOR=#000000][FONT=monospace]make ARCH=i386 bzImage[/FONT][/COLOR][/FONT]

Thanks for the tip, i’ll ultimately go that route if needed but I have the itch to get through this at present.

The kernel compiled and the bzImage was produced but it wont boot. i get this on the VM:
“kernel panic - not syncing: No init found. Try passing init= option to kernel”

And a thin client also fails with “This kernel requires x86_64 CPU, but only detected an i686 CPU”

So, it looks like I have more things to do in xconfig. Back to the console!

mkay, took out the ubuntu third party drivers but left the swap thing in place. looks like make bzImage completed. time to test!

Looks like that was yet another superfluous driver for some kind of remote. removed it. got further and yet another error:

LD vmlinux.o
ubuntu/built-in.o:(.bss+0xa40): multiple definition of debug' arch/x86/built-in.o:(.kprobes.text+0x70): first defined here ld: Warning: size of symbol debug’ changed from 49 in arch/x86/built-in.o to 4 in ubuntu/built-in.o
make: *** [vmlinux.o] Error 1
root@AusFog01:/usr/src/linux-source-2.6.32#

more info on my server:
minimalist/base/headless install of Ubuntu Server 11.10. Then installed fog .32 per instruction. then added other tidbits. and ultimately, through this process, added GDM and X.

So, this isn’t a fix to the problem, but I’ve foudn that driver is the D-Link 2000 network driver. I dont need that for my environment so i simply removed it and i think i got a bit further with the make bzImage job.

however, on to the next failure:
CC ubuntu/lirc/lirc_atiusb/lirc_atiusb.o
In file included from ubuntu/lirc/lirc_atiusb/lirc_atiusb.c:67:
ubuntu/lirc/lirc_atiusb/…/kcompat.h:245:2: error: #error “LIRC modules currently require”
ubuntu/lirc/lirc_atiusb/…/kcompat.h:246:2: error: #error " ‘Loadable module support —> Module unloading’"
ubuntu/lirc/lirc_atiusb/…/kcompat.h:247:2: error: #error “to be enabled in the kernel”
make[3]: *** [ubuntu/lirc/lirc_atiusb/lirc_atiusb.o] Error 1
make[2]: *** [ubuntu/lirc/lirc_atiusb] Error 2
make[1]: *** [ubuntu/lirc] Error 2
make: *** [ubuntu] Error 2
root@AusFog01:/usr/src/linux-source-2.6.32#

Still tracking this. Figured I’d post my findings in case I’m not alone.

more of the error, still googling…

root@AusFog01:/usr/src/linux-source-2.6.32# make bzImage
scripts/kconfig/conf -s arch/x86/Kconfig
CHK include/linux/version.h
CHK include/linux/utsrelease.h
SYMLINK include/asm -> include/asm-x86
CALL scripts/checksyscalls.sh
CHK include/linux/compile.h
VDSOSYM arch/x86/vdso/vdso-syms.lds
LD arch/x86/vdso/built-in.o
LD arch/x86/built-in.o
CC drivers/net/dl2k.o
drivers/net/dl2k.c: In function ‘rio_ioctl’:
drivers/net/dl2k.c:1294: error: implicit declaration of function ‘capable’
drivers/net/dl2k.c:1294: error: ‘CAP_NET_ADMIN’ undeclared (first use in this function)
drivers/net/dl2k.c:1294: error: (Each undeclared identifier is reported only once
drivers/net/dl2k.c:1294: error: for each function it appears in.)
make[2]: *** [drivers/net/dl2k.o] Error 1
make[1]: *** [drivers/net] Error 2
make: *** [drivers] Error 2
root@AusFog01:/usr/src/linux-source-2.6.32#

So, I’d say to some degree that i know enough Linux to be dangerous, but i might be selling myself short. Anyway, I’ve got my first Fog env up and running and have completed a number of imaging tasks. The need has come up for me to become able to add drivers to the kernel image that’s passed onto clients booting PXE.

I’ve followed instructions here:
[url]http://fogproject.org/wiki/index.php?title=Building_a_Custom_Kernel[/url]
and [url]http://dinomite.net/blog/2007/setting-up-ubuntu-for-building-kernel-modules/#.T48antUcuo5[/url]

The only item I added in the GUI (from make xconfig) was “VMware PVSCSI driver suport” I then saved and closed that guy then ran the make bzImage. the product of that was, after some time:

drivers/net/dl2k.c: In function ‘rio_ioctl’:
drivers/net/dl2k.c:1294: error: implicit declaration of function ‘capable’
drivers/net/dl2k.c:1294: error: ‘CAP_NET_ADMIN’ undeclared (first use in this function)
drivers/net/dl2k.c:1294: error: (Each undeclared identifier is reported only once
drivers/net/dl2k.c:1294: error: for each function it appears in.)
make[2]: *** [drivers/net/dl2k.o] Error 1
make[1]: *** [drivers/net] Error 2
make: *** [drivers] Error 2
root@AusFog01:/usr/src/linux-source-2.6.32#

This is what i have for my fog server and where this failure occured:
root@AusFog01:/usr/src/linux-source-2.6.32# uname -a
Linux AusFog01 2.6.32-45-server #101-Ubuntu SMP Mon Dec 3 15:54:27 UTC 2012 x86_64 GNU/Linux

And I’m running fog .32