Tom,
I don’t believe I am using a custom init.gz, but I am getting the issues you describe including the skipping and partial copies. It does however actually get to that point where the original bzImage would fail/lock at that point. If using a custom init.gz would help, it would be interesting to know how to do that. I’m going to try with your 11 MB image today and see what happens, didn’t get a chance yesterday.
Jaymes, and Tom,
Jaymes, from now on I will try that option and see if it recognizes the partitioning scheme any better in the future.
Tom, well the sector by sector copy I did with the dd option worked on the host upload, but when I went to deploy it to the client, it just skipped it as if it didn’t exist and went right to being “done”. There didn’t seem to be any error message either that I could see, but it scrolled by incredibly fast all times I tried to see it.
So I think I’ll try your 11 MB image you linked in the post above and see what happens when I try to upload/deploy with that. Here’s hoping!
Tom,
Ok, I will try that if need be, but for image type I’m using the: Multiple Partition Image - Single Disk (Not Resizeable) as the selection in image management. Do you think that might be giving it a problem? I think I might try: Raw Image (Sector By Sector, DD, Slow) instead. Who cares if it’s slow at least if I only have to do that for uploading and not for deploying it out.
If that doesn’t work, then I will try the kernel you posted above, and see what happens, I think I will do that anyways simply to test it for you on hardware that is relatively new.
I’ll let you know either way.
Tom,
At first it seemed to be able to image the XPS 8700 host correctly but there were a few possible issues…
I’m not certain if it’s relevant, but the disk on the host I wanted to image FOG seemed to think it was sdb when on the host itself it’s sda, but not certain if that makes a difference.
The host I was trying to take the image off of has 5 partitions, one of which is an extended, and another within the extended, the other 3 are primary. They are all ext4 partitions, and there is nothing with LVM going on. When FOG tried to image them, it seemed to image the first 3 very very quickly, almost as it if skipped them, and the 3rd (the swap partition) it seemed to skip altogether. The final partition with most of the data in it it seemed to image 69 GB of the 1 TB, and it went fairly quickly. Most of that 1 TB was emtpy space,and it seemed like the kernel skipped it for that reason.
However, when I tried to deploy the image to another machine of exactly the same configuration, it failed to deploy the image properly, and I am certain that was due to it’s having issue imaging the original host.
I’m not certain exactly what happened, the only thing I can think of is fdisk -l on the original host per partition shows something akin to:
partition x does not start on physical sector boundary. It says that for all 5 partitions, and I’m not sure if that is giving FOG any problems. I could reinstall CentOS 6.4 on the host if need be to fix it.
Let me know what you think. If necessary I could provide screenshots, but it’s quite a strange issue.
Tom,
Sounds good, and I am trying it now. Thus far, it’s a little more “strange” than the original kernel, but it does seem to get past getting hung up on whatever piece of hardware the original bzImage was. However, when it tries to run the kernel on the client, I get a odd pause for 45 seconds to a minute. However, after that so far it seems to behave normally.
I do like how it “appears” to ignore uploading the free space, and only uploads actual data, that’s a nice welcome change. I’m uploading a host now, and attempting to deploy it to another system of the same type. So far so good, but it’s not complete yet, I’ll let you know!
Also, you were probably right about it being a hardware issue… most likely because the video card is a newer ATI Radeon model. Now that it skips over that, things are actually proceeding.
Tom,
Thanks a bunch, I will try it, but how exactly should I install that kernel, is there a specific process? Also, is there a way to set FOG so that it works only as a cloner, and doesn’t necessarily try to adjust for hardware profile changes?
As the subject suggests, I can’t get the Quick/Full inventory to complete when trying to inventory a Dell XPS 8700 (I7 machine) with the FOG server. It worked perfectly on all of the Dell Optiplex 780 (Core 2 machine) machines, however it seems to be failing with this newer machine for some reason.
I’m running the FOG server on an I7 machine, running Fedora 13 kernel 2.6.34.9-69.fc13.x86_64 and it is on a private network with only the gigabit router, the fog server, and the machine I am trying to inventory.
What happens is, this:
Loading fog/kernel/bzImage…
Loading fog/images/init.gz… ready.
[ 1.022257] tps65010: no chip?
[ 1.232307] acpiphp_ibm: ibm+acpiphp_init: acpi_walk_namespace failed
[ 1.232509] Could not find Carillo Ranch MCH device.
[ 1.232689] uvesafb: failed to execute /sbin/v86d
[ 1.232734] uvesafb: make sure that v86d helper is installed and executable
[ 1.232780] uvesafb: Getting VBE info block failed (eax=0x4f00, err=-2)
[ 1.232825] uvesafb: vbe_init() failed with -22
[ 2.527112] drm/i810 does not support SMP
[ 3.722891] Error: Driver ‘mdio-dpio’ is already registered, aborting…
[ 3.730585] cs89x0: request_region(0x320, 0x10) failed
Then, at this point it just stops and seems to hang forever. Not even the cursor blinks as far as I can tell. It doesn’t matter whether I choose Quick Inventory, Full Inventory, choosing Client System Information to determine the IP/MAC. That is as far as I can get no matter which option is chosen essentially.
I’m curious if anyone knows what the problem is… so far usually it’s been something simple, and I’m hoping that this is the case here as well!
Thanks in advance
–CrazyHorse
[url]http://www.fogproject.org/wiki/index.php?title=Change_NFS_location[/url]
Between your suggestions Tom, and the link above, I was able to solve all of the issues I’ve had so far as I can tell. This is basically a solved thread. Thanks a ton.
Ah, I see, well, the issue is, the images that are being stored ON the FOG server are in /images right now, and that’s what I’m looking to move to /opt. Is there an easy way for me to change that within the config file, and then mv the /images directory? Or would moving the directory, and then setting up a symbolic link to it be a better idea? Either way, I have to keep the filesystem structure the way it is, and all of the large amount of free space is over in /opt on the linux side.
Tom,
This particular server has 12 GB of memory. All of the operations seem to function, but it’s like you said, it could be php. I’m not certain how I’m running php, it’s however the defaults for an updated fedora 13 / FOG 32 are set up. As far as the RAM state, I did notice that all but 70 MB of RAM was being used during a host image upload process, however, I didn’t check it at any other point. It’s just odd.
Tom,
Yes, you are correct, as the image was uploading, the image host showed it was being uploaded to something like /tmp/pigz1
Do you know of any way to specify to FOG to use a different temporary space, as well as redefine where the /images folder is? In my case for instance, moving it to /opt/images ?
As the topic subject suggests, I’m wondering if it is possible to get FOG to use another temporary directory during its image uploads? I am running FOG on Fedora 13, with some disks in a LVM configuration, but I set /opt to have the lion’s share of that space, / and specifically /tmp has much much less, yet FOG seems to want to upload the image to there instead of somewhere in /opt (where FOG is also installed). Hopefully I’m just missing it somewhere in the documentation, but you would think this would be something people would want (to keep / from filling up accidentally)
Secondarily, though I am pretty sure it’s true, does anyone know if FOG compresses its uploaded images by default? It seems that way, but wanted to make sure. Also, is there any way to make it so that those images are in an un-compressed state by default? (Likely no one would ever want to do this, but curious nonetheless)
Let me know, thanks!
–CrazyHorse
Tried it with an actual router, initial loads on the pages are still taking 30 seconds, but after it loads the pages once, it seems to be able to switch instantly without any further issues. Either way, FOG appears to be working properly now, I can handle the initial delay once on each major page. Thanks Tom!
–CrazyHorse
Tom,
Thanks a lot for responding! The looping issue you describe sounds very much like what I’d imagine the problem is. The reason I have it on a switch is that the switch is gigabit, and transferring the images would be much quicker that way! All of the routers I have are 100 m/b only so transferring 1 TB+ images for a test would be a little cumbersome. 
However, I might try it your way just to see if the issue clears up.
I would have liked to put it on the real network here, but I don’t have control over that, and FOG says it needs a static IP. Plus, I don’t want it interfering with their own PXE system they have for windows machines.
Either way, thanks Tom, I’ll try it and let you know.
–CrazyHorse
I thought I would ask the community on this one to see if there is any advice on my issue.
Running FOG server on Fedora 13 currently. I set it up on a private network to test, (which is essentially just a 5 port switch and the systems with private IP’s) I had FOG be the DHCP server, and the DNS server, and I set the Fedora 13 IP to be 13.0.0.1, the default gateway the same, and DNS to be 13.0.0.2. I did not configure a DHCP router address, I said no to that question in the install script. I filled in addresses for every question in the install script, and wasn’t sure if any should be left empty.
Anyways, all install operations seemed to complete successfully, however, now when I try to go to [url]http://13.0.0.1/fog/management[/url] it DOES connect, however, only after 30 seconds of “spinning” in the browser(Firefox). It’s that way for any operation I perform in the management page, it will work, but only after a 30 second wait time. (Keep in mind that when I put the server on the real network, I can get out to any internet webpage without any trouble.)
I’d rather not have to manage it this way, as it’s extremely annoying to even browse around. Also, I’m worried that with it in this state, it might have issues when I attempt to upload my first image, so I want to get this in hand first. Does anyone know what I might have mis-configured? It’s also possible I screwed up in Fedora somewhere as well.
Let me know if anyone has any ideas.
Thanks!