I installed Fedora 16 and then ran:
sudo yum install xen
Xen installed without errors and I rebooted. In the Grub2 list I selected Xen 4.1.2 and then selected the option:
Fedora Linux, with Xen 4.1.2 and Linux 3.3.0-4.fc16.x86_64
The boot process starts (i notice that loading ramdisk is very short) and then shortly thereafter the boot fails and the system reboots without any error.
I tried checking the boot log but I'm not sure I'm looking at the right log.
The only problem I see in the boot is
Started SYSV: sandbox, xguest and other apps that want to use pam_names$ Failed to start Xenstored - daemon managing xenstore file system
On Sat, 24 Mar 2012, Chris Jones wrote:
I installed Fedora 16 and then ran: sudo yum install xen
Xen installed without errors and I rebooted. In the Grub2 list I selected Xen 4.1.2 and then selected the option:
Fedora Linux, with Xen 4.1.2 and Linux 3.3.0-4.fc16.x86_64
The boot process starts (i notice that loading ramdisk is very short) and then shortly thereafter the boot fails and the system reboots without any error.
Recent F16 kernels are broken when running as dom0. 3.2.9-1 was the last kernel that worked. 3.2.9-2 was broken with Bug 802407 https://bugzilla.redhat.com/show_bug.cgi?id=802407 though this is now fixed. Unfortunately everything after 3.2.9-4 is broken with Bug 804347 https://bugzilla.redhat.com/show_bug.cgi?id=804347 though I did do a scratch build to try a proposed fix which is (temporarily) at http://koji.fedoraproject.org/koji/taskinfo?taskID=3923761 which seems to work for me though there are weird things going on with the 3.3 kernels, see Bug 806107 https://bugzilla.redhat.com/show_bug.cgi?id=806107 which might cause you problems with this kernel as well.
Michael Young
On Sat, Mar 24, 2012 at 05:07:23PM -0400, Chris Jones wrote:
I installed Fedora 16 and then ran:
sudo yum install xen
Xen installed without errors and I rebooted. In the Grub2 list I selected Xen 4.1.2 and then selected the option:
Fedora Linux, with Xen 4.1.2 and Linux 3.3.0-4.fc16.x86_64
The boot process starts (i notice that loading ramdisk is very short) and then shortly thereafter the boot fails and the system reboots without any error.
It wouldn't be this one: https://bugzilla.redhat.com/show_bug.cgi?id=804347 ?
I tried checking the boot log but I'm not sure I'm looking at the right log.
The only problem I see in the boot is
Started SYSV: sandbox, xguest and other apps that want to use pam_names$ Failed to start Xenstored - daemon managing xenstore file system
-- xen mailing list xen@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/xen
Thank you both for your response.
That's probably the bug. I was hoping to find more answers here as after I found that bug report.
@M.A Young I saw that fix but wasn't sure how to apply it. I'm not really sure how to use the special builds people mention.
Do either of you know a stable platform to run Xen on? I tried Xen Cloud Platform but found the templates limiting. I wouldn't mind going back to Xen Clout Platform if I could figure out how to run config files for installation instead of templates. I mainly want to use the hypervisor for standing up test servers and desktop environments.
On Sat, Mar 24, 2012 at 5:39 PM, Konrad Rzeszutek Wilk < konrad.wilk@oracle.com> wrote:
On Sat, Mar 24, 2012 at 05:07:23PM -0400, Chris Jones wrote:
I installed Fedora 16 and then ran:
sudo yum install xen
Xen installed without errors and I rebooted. In the Grub2 list I selected Xen 4.1.2 and then selected the option:
Fedora Linux, with Xen 4.1.2 and Linux 3.3.0-4.fc16.x86_64
The boot process starts (i notice that loading ramdisk is very short) and then shortly thereafter the boot fails and the system reboots without any error.
It wouldn't be this one: https://bugzilla.redhat.com/show_bug.cgi?id=804347 ?
I tried checking the boot log but I'm not sure I'm looking at the right log.
The only problem I see in the boot is
Started SYSV: sandbox, xguest and other apps that want to use pam_names$ Failed to start Xenstored - daemon managing xenstore file system
-- xen mailing list xen@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/xen
n Sat, Mar 24, 2012 at 08:59:30PM -0400, Chris Jones wrote:
Thank you both for your response.
That's probably the bug. I was hoping to find more answers here as after I found that bug report.
Ok, well, the patches should appear shortly in the FC16 kernel builds (I hope?) They were Acked-by Ingo, let me attach the patches to the BZ.
Josh, is there a specific process for those patches to appear in F16? Should I email them to some kernel-backport@fedora.org email list?
Thanks!
@M.A Young I saw that fix but wasn't sure how to apply it. I'm not really sure how to use the special builds people mention.
Do either of you know a stable platform to run Xen on? I tried Xen Cloud Platform but found the templates limiting. I wouldn't mind going back to Xen Clout Platform if I could figure out how to run config files for installation instead of templates. I mainly want to use the hypervisor for standing up test servers and desktop environments.
On Sat, Mar 24, 2012 at 5:39 PM, Konrad Rzeszutek Wilk < konrad.wilk@oracle.com> wrote:
On Sat, Mar 24, 2012 at 05:07:23PM -0400, Chris Jones wrote:
I installed Fedora 16 and then ran:
sudo yum install xen
Xen installed without errors and I rebooted. In the Grub2 list I selected Xen 4.1.2 and then selected the option:
Fedora Linux, with Xen 4.1.2 and Linux 3.3.0-4.fc16.x86_64
The boot process starts (i notice that loading ramdisk is very short) and then shortly thereafter the boot fails and the system reboots without any error.
It wouldn't be this one: https://bugzilla.redhat.com/show_bug.cgi?id=804347 ?
I tried checking the boot log but I'm not sure I'm looking at the right log.
The only problem I see in the boot is
Started SYSV: sandbox, xguest and other apps that want to use pam_names$ Failed to start Xenstored - daemon managing xenstore file system
-- xen mailing list xen@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/xen
Hi
I can confirm that M A Young's Xen kernel fixes the problem of booting Linux Kernel as Dom0 in FC16.
With latest Fedora 16 kernel "kernel-3.3.0-4.fc16.x86_64" Dom0 will not boot, but after showing a stack trace, desktop computer reboots. With M A Young's Dom0 kernel kernel-3.3.0-5.may.fc16.x86_64 the problem is fixed.
Fixing this problem is now a bit more important, because old working kernels have been removed by old kernel removal logic.
Xen paravirtual virtual guest machine works too with SELinux disabled.
Best regards, Marko Ristola
On 03/26/2012 05:53 PM, Konrad Rzeszutek Wilk wrote:
n Sat, Mar 24, 2012 at 08:59:30PM -0400, Chris Jones wrote:
Thank you both for your response.
That's probably the bug. I was hoping to find more answers here as after I found that bug report.
Ok, well, the patches should appear shortly in the FC16 kernel builds (I hope?) They were Acked-by Ingo, let me attach the patches to the BZ.
Josh, is there a specific process for those patches to appear in F16? Should I email them to some kernel-backport@fedora.org email list?
Thanks!
@M.A Young
On Tue, Mar 27, 2012 at 06:42:46PM +0300, Marko Ristola wrote:
Hi
I can confirm that M A Young's Xen kernel fixes the problem of booting Linux Kernel as Dom0 in FC16.
With latest Fedora 16 kernel "kernel-3.3.0-4.fc16.x86_64" Dom0 will not boot, but after showing a stack trace, desktop computer reboots. With M A Young's Dom0 kernel kernel-3.3.0-5.may.fc16.x86_64 the problem is fixed.
Fixing this problem is now a bit more important, because old working kernels have been removed by old kernel removal logic.
Xen paravirtual virtual guest machine works too with SELinux disabled.
Why did you need to disable SELinux? Is there a BZ for that?
On 03/27/2012 07:07 PM, Konrad Rzeszutek Wilk wrote:
On Tue, Mar 27, 2012 at 06:42:46PM +0300, Marko Ristola wrote:
Xen paravirtual virtual guest machine works too with SELinux disabled.
Why did you need to disable SELinux? Is there a BZ for that?
BZ entry is https://bugzilla.redhat.com/show_bug.cgi?id=749172
pygrub called by Xend reads grub settings from within the disk image. Thus xend needs for virtual disk image "xen_image_t" SELinux type.
I use qemu-dm for the DomU disk access after booting. /usr/lib/xen/bin/qemu-dm works with "virt_image_t" SELinux type.
I can't put those both SELinux types for the disk image.
I don't know what kind of SELinux policy / binary file labeling change is needed to fix it.
Regards, Marko Ristola
On Tue, Mar 27, 2012 at 11:07:09PM +0300, Marko Ristola wrote:
On 03/27/2012 07:07 PM, Konrad Rzeszutek Wilk wrote:
On Tue, Mar 27, 2012 at 06:42:46PM +0300, Marko Ristola wrote:
Xen paravirtual virtual guest machine works too with SELinux disabled.
Why did you need to disable SELinux? Is there a BZ for that?
BZ entry is https://bugzilla.redhat.com/show_bug.cgi?id=749172
pygrub called by Xend reads grub settings from within the disk image. Thus xend needs for virtual disk image "xen_image_t" SELinux type.
I use qemu-dm for the DomU disk access after booting. /usr/lib/xen/bin/qemu-dm works with "virt_image_t" SELinux type.
I can't put those both SELinux types for the disk image.
I don't know what kind of SELinux policy / binary file labeling change is needed to fix it.
I thought it was xen_disk_t? There are some slides by Walsh about what your need for LVM and Xen - see if Google comes up with anything.
Regards, Marko Ristola
Thanks for taking time for this. I have swithced on the Dom0 server into latest Fedora 17 RC this weekend.
I had severe problems with non-Xen i386 non-PCIE rack server with NFS etc: IBM xSeries 306. It halted with latest Fedora 16 kernel about two minutes after successful boot. Before these problems, NFS server could livelock, if for example compiling or copying Kernel over NFS.
So I took the Dom0 server as the NFS server and it has now Fedora 17. Motherboard: FUJITSU SIEMENS A8NE-FM 1011-005, two CPUs. On Fedora 17, without Dom0 the computer seems to work fine (once locked up during software RAID5 resync). While running as Dom0, using the NFSv4 kernel server (with idmap) will cause a kernel oops and reboot instantly.
I haven't got the kernel message log yet.
So the SELinux problem isn't now at the topmost on my list.
(With another non-PCIE AMD motherboard disabling MSI interrupts and making motherboard use only interrupts < 16, made it work with Fedora 16 without Dom0. As Dom0 it doesn't work).
Regards, Marko
On 05/10/2012 06:42 PM, Konrad Rzeszutek Wilk wrote:
On Tue, Mar 27, 2012 at 11:07:09PM +0300, Marko Ristola wrote:
On 03/27/2012 07:07 PM, Konrad Rzeszutek Wilk wrote:
On Tue, Mar 27, 2012 at 06:42:46PM +0300, Marko Ristola wrote:
Xen paravirtual virtual guest machine works too with SELinux disabled.
Why did you need to disable SELinux? Is there a BZ for that?
BZ entry is https://bugzilla.redhat.com/show_bug.cgi?id=749172
pygrub called by Xend reads grub settings from within the disk image. Thus xend needs for virtual disk image "xen_image_t" SELinux type.
I use qemu-dm for the DomU disk access after booting. /usr/lib/xen/bin/qemu-dm works with "virt_image_t" SELinux type.
I can't put those both SELinux types for the disk image.
I don't know what kind of SELinux policy / binary file labeling change is needed to fix it.
I thought it was xen_disk_t? There are some slides by Walsh about what your need for LVM and Xen - see if Google comes up with anything.
Regards, Marko Ristola
Michael Young's latest Xen hypervisor build 4.1.3-rc1 seems to fix my Dom0 NFS server crash problem. Dom0 NFS server is as stable as with vanilla kernel. I booted Dom0 (standard Fedora 17 kernel) with serial console, with all debug flags turned on.
Old problematic Xen hypervisor was the current Fedora 17 one.
Regards, Marko
On 05/13/2012 08:42 PM, Marko Ristola wrote: ...
So I took the Dom0 server as the NFS server and it has now Fedora 17. Motherboard: FUJITSU SIEMENS A8NE-FM 1011-005, two CPUs. On Fedora 17, without Dom0 the computer seems to work fine (once locked up during software RAID5 resync). While running as Dom0, using the NFSv4 kernel server (with idmap) will cause a kernel oops and reboot instantly.
...
Regards, Marko
...
On Mon, May 14, 2012 at 08:11:24PM +0300, Marko Ristola wrote:
Michael Young's latest Xen hypervisor build 4.1.3-rc1 seems to fix my Dom0 NFS server crash problem. Dom0 NFS server is as stable as with vanilla kernel.
That's good to hear. Xen 4.1.3 fixes many bugs, and obviously those fixes include something you were hitting :)
I booted Dom0 (standard Fedora 17 kernel) with serial console, with all debug flags turned on.
Old problematic Xen hypervisor was the current Fedora 17 one.
Yeah, the "default" version currently is Xen 4.1.2.
-- Pasi
Regards, Marko
On 05/13/2012 08:42 PM, Marko Ristola wrote: ...
So I took the Dom0 server as the NFS server and it has now Fedora 17. Motherboard: FUJITSU SIEMENS A8NE-FM 1011-005, two CPUs. On Fedora 17, without Dom0 the computer seems to work fine (once locked up during software RAID5 resync). While running as Dom0, using the NFSv4 kernel server (with idmap) will cause a kernel oops and reboot instantly.
...
Regards, Marko
...
-- xen mailing list xen@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/xen
-
Chris Jones -
Konrad Rzeszutek Wilk -
M A Young -
Marko Ristola -
Pasi Kärkkäinen