pxe network boot - segfault at 0 ip

Discussions about using Linux guests in VirtualBox.
Post Reply
SSMinnow
Posts: 4
Joined: 25. Feb 2012, 01:19

pxe network boot - segfault at 0 ip

Post by SSMinnow »

I am stumped and hoping that someone can give me some guidance here. I am trying to pxe boot a couple of vms off a rhel 6.2 kickstart server and no matter what I do I keep getting a segfault error followed by a kernel panic. Outside of using a completely different kickstart server, which does work for normal non vm kickstarts, I have tried two different boxes running virtualbox to see if this was server related and they both give me this same error. So I suppose I am not eliminating the fact that the kickstart server could still be the problem, but I am skeptical. The two servers running virtual box are as follows:

HP Proliant 380GL G6
Sunfire x4100 m2

the kickstart server is also a HP Proliant 380GL G6

All three of these boxes are on the same network, all have iptables turned off, and there is no firewall between them. The vms get an ip from the kickstart dhcp server and give me the menu.default option screen just fine. I select the kickstart menu option and the vms load the vmlinuz and initrd.img just fine I believe since they make it past that point. However, sometime after that they both end up seg faulting. I have played with hardware acceleration on/off, io apic on/off, and motherboard virtualization on/off but still get the exact same results. I am attaching the log files of the two vms and also a screen capture of the error. I have been digging through the forum posts and haven't seen anyone else with this similar problem that I could find. Hopefully someone has an idea.

Thanks!
Screen capture
Screen capture
virtualbox_kernel_panic_screen.tiff (41.56 KiB) Viewed 2911 times
Attachments
Proliant-VBox.log
proliant log
(75.44 KiB) Downloaded 19 times
Sunfire-VBox.log
sunfire log
(62.44 KiB) Downloaded 14 times
SSMinnow
Posts: 4
Joined: 25. Feb 2012, 01:19

Re: pxe network boot - segfault at 0 ip

Post by SSMinnow »

Apparently my screen capture didn't work like I wanted, trying again here.
Attachments
Screen capture
Screen capture
virtualbox_kernel_panic_screen.tiff (41.56 KiB) Viewed 2910 times
SSMinnow
Posts: 4
Joined: 25. Feb 2012, 01:19

Re: pxe network boot - segfault at 0 ip

Post by SSMinnow »

Just wanted to give an update here on what I have found as I have continued to troubleshoot this. I came to the conclusion that the initrd.img kernel was causing this ip segfault, however I am not sure where. I have tried various parameters at the grub/anaconda menu to see if setting the ip or dhcp parameters make any difference but they don't.

Finally I ended up grabbing the initramfs off a rhel 6.2 vm that I installed on the same host server off a .iso cd image and that has gotten me further. I continue to get dracut error messages saying I need to specify a root= parameter. I have tried specifying the UUID of the vm harddrive, but there is nothing on the drive yet so that obviously fails. I have tried changing this to point to the rhel 6.2 kickstart server and an install.img file I pulled off the cd, however that still gives me an error saying 'dracut: fatal: don't know how to handle 'root=live:http://<kickstart ip>/tftpboot/RHEL-6.2/install.img'. It would appear that it does recognize the 'root=live:' parameter because right before this error I get the message "liveroot is now 'live:http://<kickstart ip>/tftpboot/RHEL-6.2/install.img'". Anyone have any suggestions as to what liveroot should point to and how to get this to work? I am confident that I do not want a root= pattern that points to the vm because the install and hence kernel boot needs to come from the kickstart server.
SSMinnow
Posts: 4
Joined: 25. Feb 2012, 01:19

Re: pxe network boot - segfault at 0 ip

Post by SSMinnow »

Another update here. I made headway on this today. I figured out how to get past the error that I was getting and to have the vm kickstart successfully. The problem ended up being caused by boot options in the grub menu. So for example our grub menu would come up, I could either select enter to boot or tab to see the boot command. Our boot command looked like this:

RHEL-6.2/vmlinuz initrd=RHEL-6.2/initrd.img ks=http://<ip of ks server>/profile/kick1.ks ksdevice=bootif noipv6 console=tty0 console=ttyS0,9600 BOOTIF=<mac address of guest vm interface>

However, when the vm guest would boot it would get all the way to the point that I have described above and give me the error 'segfault at 0 ip'. The solution that I found today, was to remove any reference to an interface in the boot command. So at the grub menu I would change the command to this:

RHEL-6.2/vmlinuz initrd=RHEL-6.2/initrd.img ks=http://<ip of ks server>/profile/kick1.ks

The vm guest booted right up, completed the kickstart and is up and running. There were a couple of hiccups that I still need to iron out, like the install still asked for the language setting. I also have to make sure the disk creation in the .ks file matched with how much disk space I was giving the vm in the software. For instance we were creating 4 partitions of 10gb and more each and the vm harddrive settings was set as 8gb which caused some errors. Fixing that resolved the problem though.

Anyway, just wanted to give everyone an update and document this in case anyone else runs into it. I am still unsure why I cannot specify the interface in the boot command. I also don't know if I needed to remove all the options that I did or just the ones pertaining to the interface. I'll have to do some more testing to find out. BTW, I did end up just using the initrd.img and vmlinuz files off the RHEL iso/dvd. No changes at all to those files like I was previously messing with.
Post Reply