OpenSUSE 11.4 hangs on boot when installed as XenServer guest

English Install/Boot/Login
#1

I created an OpenSUSE 11.4 PV guest on XenServer 5.6, and about half the time, it fails to finish booting. it starts all the services, and fails immediately before it would display the login prompt. I’ve spent hours googling for this problem, and can’t find any reference to it anywhere, meaning that I may be the first to experience it.

has anyone else done OpenSUSE 11.4 on XenServer 5.6 and had this problem? were you able to resolve it, and how?

I don’t even know where to start looking at logs to try to find the problem, so any help in that regard would be very much appreciated.

thanks in advance.

#2

Hi,

we use OpenSuSE 11.4 x86_64 for Dom0 and we see a similar problem with OpenSuSE 11.4 guests.

So far we haven’t found a fix - apart from trying to boot the DomU several times.

D.

#3

I think we wound a solution / workaround…

We have always been using the ‘xencons=tty’ parameter for our DomU’s.
It seems that this parameter is causing problems with the newer Xen kernel, such as the one bundled with OpenSuSE 11.4…

If we omit that parameter, it seems to work fine (done over 50 reboots today).

Regards, D.

#4

I had this problem. What I did to get around it was, during the install, do NOT install updates. If you do, it will reboot and you’re stuck. Complete the install and you should be able to login without a reboot. The first thing you need to do is install the Xen kernel using zypper:


zypper install kernel-xen kernel-xen-base

Edit your grub config to use it by default. Now your VM will work fine.

What I still can’t figure out is how to get XenCenter to show me memory usage. I tried installing the XenTools RPMs from the CD, as well as installing various Xen tools in the package tree, but in Xen center it always says the VM doesn’t have the tools installed.

#5

On Wed, 07 Sep 2011 14:06:03 GMT, djsumdog
<djsumdog@no-mx.forums.opensuse.org> wrote:

>
>I had this problem. What I did to get around it was, during the install,
>do NOT install updates. If you do, it will reboot and you’re stuck.
>Complete the install and you should be able to login without a reboot.
>The first thing you need to do is install the Xen kernel using zypper:
>
>
>
>Code:
>--------------------
>
> zypper install kernel-xen kernel-xen-base
>
>--------------------
>
>
>Edit your grub config to use it by default. Now your VM will work fine.
>
>
>What I still can’t figure out is how to get XenCenter to show me memory
>usage. I tried installing the XenTools RPMs from the CD, as well as
>installing various Xen tools in the package tree, but in Xen center it
>always says the VM doesn’t have the tools installed.

As i understand it you have to install tools in the host and again in each
VM OS to get the data. It seems to be the same for VirtualBox and VMware
as well.

?-)

#6

I have also hit this problem. I have been doing a lot of research, and here are a few of my findings:

  1. this problem seems to only occur on VM’s with more than one cpu
  2. turning off/on irqbalanced doesn’t have any affect
  3. editing /etc/sysconfig/boot and setting RUN_PARALLEL=“no” has no effect
  4. the problem is most defiantly with the tty / console provided by XenServer and how Suse Handles this

I have devised a really great way to exercise this bug and bring it to the surface:
As root:

cd
cd .ssh
ssh-keygen -b 2048 -t rsa
(press enter 3 times for defaults)
cat id_rsa.pub >> authorized_keys2
ssh root@localhost ls

This allows root to ssh into the box without needing a password. Now edit /etc/init.d/rc . Put this at the end of the file, before the final line of “exit 0”:


test "3" = "$RUNLEVEL" && ssh root@localhost "sleep 10 && shutdown -r now" &

This assumes that your final runlevel is 3. Both these modifications together reboot the box indefinitely, until the bug occurs. When the bug occurs ssh is broken, and the box will hang without restarting. This provides an excellent way to automatically reboot the box and check for the bug.

I would advise making these modifications on a clone of your affected machine, and discarding it when you are done. I do not know the security implications of adding the key to authorized_keys2.

#7

Ok great! I think I have a solution. For me this reduces the bug from 1 in 5 times, to around 1 in 1000. I still see the bug but this helps a bunch:

======== My Almost 100% Solution!! ========
Edit /etc/inittab. Comment out all of the normal tty’s (they are unused in PV mode). For example:


#1:2345:respawn:/sbin/mingetty --noclear tty1
#2:2345:respawn:/sbin/mingetty tty2
#3:2345:respawn:/sbin/mingetty tty3
#4:2345:respawn:/sbin/mingetty tty4
#5:2345:respawn:/sbin/mingetty tty5
#6:2345:respawn:/sbin/mingetty tty6

Near the end of the file, there (may) should be a cons entry. Comment it out as well


#cons:12345:respawn:/sbin/smart_agetty -L 42 console

add this line to the end of inittab:


co:2345:respawn:/sbin/agetty xvc0 9600 vt102

From XenCenter: Right click on the VM, click Startup Options. Replace all the text in OS Boot Parameters with this:


xencons=xvc console=xvc0

If you have any troubles getting the vm to boot (blank screen) or only the very first boot messages and then no output, you can still get to your machine by adding the word " single" to the OS Boot Parameters. These problems indiciate console issues or a mismatch between inittab and the boot params

Please try this and let me know if it works for you. (danci1973’s solution didn’t work at all for me).

Resources:
More information but not the solution:
http://www.xen.org/files/xensummit_4/xensummit_linux_console_slides.pdf

agetty option, didn’t work:
http://e-books.bassq.nl/Citrix/Xen%20Virtualization%202008.pdf

Other people with similar problems, no solution:
Re: [opensuse-virtual] OpenSuse 11.4 domU hang on boot

Very helpful:
Fly By Unix: XenServer PV guest won’t launch login prompt

Regards,
Port Forward Podcast
Port Forward Podcast | Linux Programming Startups Bitcoin XenServer.

#8

After spending several hours today looking at this - this seems to dramatically improved my system. I was able to boot in to init 1, then sometimes switch to init 3 (and the commented out terminals don’t apply to run level 1 - so that ties in). I’ve rebooted a couple of times in to runlevel 3 without any issue!

Think I’ll wait for a while before trying 12.1!

Thanks - you saved me a bunch more hours I’m sure.

#9

Great! I’m glad that I could save anybody else lots of work. How many machines do you have running if you dont mind me asking? We are always looking to talk to cool people with tech backgrounds!

Regards,
Port Forward Podcast
Port Forward Podcast | Linux Programming Startups Bitcoin XenServer