Hey there,
sorry for letting you wait this long. Today I found some time to test a little bit and things look a lot better meanwhile.
I did another fresh install from the current svn and now the iPXE boot works without any flaws! I will now test to Up- and Download some images. I will keep you up to date!
Okay, tried the new build. After not being able to add a host via pxe register or manually via the web interface I checked the database and the hostMAC field is missing on table hosts. As it is a fresh installation there still seems to be some faults in your database setup script.
Can you tell me where the script is located to check it for errors?
I will check the new version tomorrow morning, otherwise I will provide you the error log.
Also: PM with email. But maybe this might be interesting for other people, too.
Yes, it is. The only thing I reused is my datastore I mount as /images folder.
After installation again i can log in and everything looks quite fine. But again newly added hosts are not shown in the host list.
Yes, I will send you an address today.
Meanwhile I tried a fresh installation and during the db schema installation I got the following error:
[CODE]Update/Install Successful!
Click here to login.
The following errors occured
Update ID: 130 - 5
Database Error:
No database selected
Database SQL:
INSERT INTO fog.hostMAC (hmMAC,hmHostID,hmPending) (SELECT pmAddress,pmHostID,‘1’ FROM pendingMACS)[/CODE]
I think there is an error in the db setup script. Can you hint me where I can fidn it?
Okay, so I just tested build 2389 (current svn checkout).
The good news is that it now boots to the fog menu without any problems.
The bad news is that this is almost the only thing working.
Problem 1: On the web interface I can only see hosts created BEFORE the upgrade to the current build. Newer hosts exist in the database and when I try to add them again I get told that they already exist, but I can’t see them.
Problem 2: Quick Registration got stuck. The database entry still got created (at least the host entry, not sure about the rest).
Problem 3: Although the host entry exists, it still says Host in NOT registered. Which is probably caused by…
Problem 4: No network connectivity in FOG menu. In the IP settings there is no address attached to the network interface, therefore ping to the server fails of course.
Compatibility check tells me I should be fine.
Any more ideas at this stage? If i have some spare time I will try and do a fresh install instead of an upgrade.
I will do and try again on monday. I’m already on my weekend.
Hey there,
as I really like the fog software I am currently trying to use is for more and more types of different machine. And currently I am stuck with trying to get one machine to boot via iPXE. The Network Adapter is a Broadcom netXtreme II. I run fog 1.2.0 on CentOS 6.5.
I tried several versions of the kpxe file but unfortunately without any success until now.
With the original [B].kpxe[/B] file it seems to start but then I get this and it’s stuck:
[CODE]Could not configure console: Input/output error (http://ipxe.org/1d8c2139)
Could not boot: Input/output error (http://ipxe.org/1d8c2139)
Could not boot: Input/output error (http://ipxe.org/1d8c2139)
Broadcom UNDI PXE-2.1 v2.8.5
Copyright 
2000-2006 Braodcom Corporation
Copyright 1997-2000 Intel Corporation
All rights reserved.[/CODE]
With the original [B].kkpxe[/B] file I get this and the server continues with the next boot devices:
[CODE]Could not configure console: Input/output error (http://ipxe.org/1d8c2139)
Could not boot: Input/output error (http://ipxe.org/1d8c2139)
Could not boot: Input/output error (http://ipxe.org/1d8c2139)
PXE-M0F: Exiting Broadcom PXE ROM.[/CODE]
With the original .kpxe.INTEL i get this (the last line keeps repeating):
[CODE]CLIENT MAC ADDR: XX XX XX XX XX XX GUID: XXXXXXXX XXXX XXXX XXXX XXXXXXXXXXXX
CLIENT IP: 192.168.1.22 MASK: 255.255.255.0 DHCP IP: 192.168.1.1
PXE->EB: !PXE at 9FA.0040, entry point at 95FA.00D6
UNDI code segment 95FA::4964, data segment 982A.5D00 (576-619kB)
UNDI device is PCI 06:00.0, type DIX+802.3
576kB free base memory after PXE unload
iPXE initialising devices…Ethernet addr: XX:XX:XX:XX:XX:XX
Broadcom NetXtreme II (B2) PCI-X 64-bit 133MHz
Valid link not established
Ethernet addr: XX:XX:XX:XX:XX:XX
Broadcom NetXtreme II (B2) PCI-X 64-bit 133MHz
NIC Link is Up, 1000 Mbps full duplex, receive & transmit flow control ON
WARNING: Using legacy NIC wrapper on XX:XX:XX:XX:XX:XX
ok
iPXE 1.0.0+ (e5878) – Open Source Network Boot Firmware – http://ipxe.org
Features: NFS HTTP HTTPS iSCSI DNS TFTP AoE SRP bzImage ELF MBOOT PXE PXEXT Menu
Configuring (net0 XX:XX:XX:XX:XX:XX)… ok
/default.ipxe… ok
http://192.168.1.1/fog/service/ipxe/boot.php… ok
http://192.168.1.1/fog/service/ipxe/bg.png… transmit timed out
transmit timed out
transmit timed out
…[/CODE]
With the undionly.kpxe.INTEL from [URL=‘http://svn.code.sf.net/p/freeghost/code/trunk/packages/tftp/undionly.kpxe.INTEL’]this[/URL] link it’s just the same as the file included in the installed package.
At last I built my own undionly.kpxe from the iPXE git repository, then I got this messages looping again and again:
[CODE]CLIENT MAC ADDR: XX XX XX XX XX XX GUID: XXXXXXXX XXXX XXXX XXXX XXXXXXXXXXXX
CLIENT IP: 192.168.1.22 MASK: 255.255.255.0 DHCP IP: 192.168.1.1
PXE->EB: !PXE at 9FA.0040, entry point at 95FA.00D6
UNDI code segment 95FA::4964, data segment 982A.5D00 (576-619kB)
UNDI device is PCI 06:00.0, type DIX+802.3
576kB free base memory after PXE unload
iPXE initialising devices…ok
iPXE 1.0.0+ (a48a) – Open Source Network Boot Firmware – http://ipxe.org
Features: HTTP iSCSI DNS TFTP AoE bzImage ELF MBOOT PXE PXEXT Menu
net0: XX:XX:XX:XX:XX:XX using undionly on UNDI-PCI06:00.0 (open)
[Link:up, TX:0 TXE:0 RX:0 RXE:0]
Configuring (net0 XX:XX:XX:XX:XX:XX)… ok
net0: 192.168.1.21/255.255.255.0
Next server: 192.168.1.1
Filename: undionly.kpxe
tftp://192.168.1.1/undionly.kpxe… ok
…[/CODE]
So this is the point I ran out of ideas. So can you help me out and point me to where I should look next?
Thank you, that really helps a lot!
Hey Tom,
I didn’t want to offend you at all and I’m well aware how much work you put into this project and that it’s not comparable to a huge company’s commercial product. My point of view is just taht I have some very strict requirements and I try to find a way to manage this task with an open solution, especially as I think the design of fog is way better than acronis or ghost.
The aim of my question was just whether there is such a freature or if someone has an idea how to make it work with currently implemented possibilites. So thank you for your answer.
Again about the list:
Yes, I’m aware that my use case is not the most common.
I’m just still comparing to commercial solutions and Ghost / Acronis are able to change the IP during deployment process.
I also found out that since Win2k8 / Win7 you are abe to join an AD offline by providing a prepared file and the djoin.exe tool. Is there a way to trigger the execution of a remote batch file after deployment without using snapins?
What I am currently trying with Ghost and would like to to is:
What I would need is just a way to trigger a command execution without any preparation on the image. But I guess there is no way to do this with fog yet 
Hey there,
basically it’s about the question in the title? Is it possible to change a host’s IP address and join an AD without the use of Sysprep or Fog Services (like it is possible to change the hostname without these tools)?
I can not find an definite answer and was not able to find a solution during my current tests. So is there really no way to perform these tasks or am I just missing something?
Okay, I found a solution that works for me, although it is not in a state to use is as a patch. My solution is ONLY for the download task, it does NOT fix the upload task and you have to do some manual steps everytime you create a new image (and only for mps images). I did the following:
204 if [ -f $tmpMBR ]; then
205 if [ “$mbrsize” != “32256” -a “$has_GRUB” != “1” ] && [ “$mbrsize” != “512” ] && [ “$mbrsize” != “2048” ]; then
206 dots “Restoring Partition Tables”;
207 sgdisk -gel $tmpMBR $hd 2>&1 >/dev/null;
208 gptcheck=“yes”;
209 elif [ “$mbrsize” == “32256” -o “$has_GRUB” == “1” ] && [ “$osid” == “50” ]; then
210 dots “Restoring MBR and GRUB”;
211 restoreGRUB “${hd}” “1” “${imagePath}”;
212 if [ -e “${imagePath}/d1.partitions” ]; then
213 echo “Done”;
214 dots “Extended partitions”;
215 sfdisk $hd < ${imagePath}/d1.partitions &>/dev/null;
216 else
217 echo “Done”;
218 dots “No extended partitions”;
219 fi
220 elif [ “$mbrsize” == “2048” ]; then
221 dots “Restoring Lenovo Specific MBR”;
222 dd if=$tmpMBR of=$hd bs=512 count=4 &>/dev/null;
223 echo “Done”;
224 else
225 dots “Restoring MBR”;
226 dd if=$tmpMBR of=$hd bs=512 count=1 &>/dev/null;
227 if [ -e “${imagePath}/d1.partitions” ]; then
228 echo “Done”;
229 dots “Extended partitions”;
230 sfdisk $hd < ${imagePath}/d1.partitions &>/dev/null;
231 else
232 echo “Done”;
233 dots “No extended partitions”;
234 fi
235 fi
After these steps your workstations should boot to windows 7 again after running a deployment task - well, at least mine do.
Hey there,
once again I encountered a problem where I might need your help.
We are currenty getting new machines for our location. The new machines are all lenovo ThinkCentre E73 with Windows 7 on them. The first time i tried to deploy an image the computer just showed me a blinking curser.
Meanwhile I found out that lenovo machines use an mbr which does not have the usual size of 512, but 2048 byte. I tried to manually copy the mbr from a working machine to a freshly deployed machine an it booted again.
As I don’t want to manually do this with every single machine after a deployment task I’m currently looking for a solution to automate this.
I found out that the *.mbr file in the image folder is nothing more than a dump of the first 512 bytes of the disk, well the mbr. So i replaced this file with the whole lenovo mbr, but the tool refuses to use this file (probably because of the unusual size; i set all permissions to be the same).
Is there any known way to automate the process, or even tell fog to use the first 2048 byte as mbr? I know that symantec ghost supports such a flag (ghost -ib), is there such an option / workaround in fog?
Hey there,
when using multicast download there should be only one connection to the storage node. Unfortunately, the node uses slots for each multicast client. So you have to massively turn op the number of allowed connections when using a lot of multicast, which can decrease the perfomance when a lot of unicast connections get established.
It would be great if multicasts get somehow grouped together als single tasks. This would also make the task view look way more clear.
Okay, I got the problem fixed. The issue really was the script only applying the hostname change to the last partition restored. These are the steps I’ve done:
root@e21a101:~# cd /var/www/fog/service/ipxe/
root@e21a101:/var/www/fog/service/ipxe#
root@e21a101:/var/www/fog/service/ipxe# xz --decompress init.xz
root@e21a101:/var/www/fog/service/ipxe# mkdir initmountdir
root@e21a101:/var/www/fog/service/ipxe# mount -o loop init initmountdir
root@e21a101:/var/www/fog/service/ipxe# vim initmountdir/bin/fog.download
Then I added the changeHostname call to line 281 like the following:
268 elif [ “$imgType” == “mps” ]; then
269 parts=fogpartinfo --list-parts $hd 2>/dev/null;
270 diskLength=expr length $hd;
271 for part in $parts; do
272 partNum=${part:$diskLength};
273 echo " * Processing Partition: $part ($partNum)“;
274 sleep 2;
275 imgpart=”$imagePath/d1p$partNum.img";
276 if [ ! -f $imgpart ]; then
277 echo " * Partition File Missing: $imgpart";
278 else
279 writeImageMultiCast $part
280 fi
281 changeHostname;
282 done
root@e21a101:/var/www/fog/service/ipxe# umount initmountdir/
root@e21a101:/var/www/fog/service/ipxe# rmdir initmountdir
root@e21a101:/var/www/fog/service/ipxe# xz -z -9 -C crc32 init
That’s it. This way it works perfectly for me. But keep in mind this is a rather quick and dirty fix and only applies to distribution by multicast with a mps type image (Multiple Partition Image - Single Disk (Not Resizable) - (2))! So let’s hope that there will be a clean change to the repos that targets all type of distributions and images.
Could it possibly be that i have to change some setting to run the hostname change task like mentioned in this thread? [url]http://fogproject.org/forum/threads/change-windows-hostname-early-just-after-imaging-before-booting.4348/[/url]
I don’t think the last partition deployed is the windows partition, so if the default settings didn’t change this could be our problem. Can you tell me where to look at to solve this problem? I wasn’t able to find these settings as I don’t have the folder /bin/fog
[INDENT=1]Note that also with ‘changeHostname’ in it’s current position in /bin/fog, it will only attempt to change the name in the LAST partition - i copied it to after every partition restore[/INDENT]
[INDENT=1] [/INDENT]
Can you tell me where I can find this in fog 1.1.2? Unfortunately I cannot find this folder in my installation.
I just checked the registry and it still tells me the old hostname. A search for the new hostname inside the registry didn’t find any results, too.
I will check it tomorrow morning, I am currently not at the place anymore. Thanks already for all the help.