Cyconet Blog » xen

Virtualisation on HP ProLiant hardware with squeeze

cyco — Thu, 24 Mar 2011 15:57:10 +0000

In case you are using HP ProLiant G6-series or G7-series with Intel-based hardware and you are thinking about virtualistion (e.g. booting the hypervisor), you should have a look into this customer advisory. To make it short, the following should work:

# grep GRUB_CMDLINE_XEN /etc/default/grub
GRUB_CMDLINE_XEN=”pci=use_crs”
# update-grub

It took me two workdays to discover this solution for a DL180G6. Anyways … I’m still having trouble on a DL160G6 when booting the hypervisor, suggestions are welcome.

linux-image-2.6.32-bpo.5-xen-686 and xen/xen-common from squeeze broken?

cyco — Sun, 31 Oct 2010 14:03:50 +0000

I’m experiencing since some time (out of the blue) spontaneous reboots of some of my dom0s on lenny/i386 (others don’t have this effect). As result of this, I started rolling out linux-image-2.6.32-bpo.5-xen-686 from bpo and backported xen and xen-common from testing:

# dpkg -l | grep xen | grep bpo
ii libxenstore3.0 4.0.1-1~bpo50+1 Xenstore communications library for Xen
ii linux-image-2.6.32-bpo.5-xen-686 2.6.32-26~bpo50+1 Linux 2.6.32 for modern PCs, Xen dom0 suppor
ii xen-hypervisor-4.0-i386 4.0.1-1~bpo50+1 The Xen Hypervisor on i386
ii xen-linux-system-2.6.32-bpo.5-xen-686 2.6.32-26~bpo50+1 Xen system with Linux 2.6.32 on modern PCs (
ii xen-tools 4.1-1~bpo50+1 Tools to manage Debian XEN virtual servers
ii xen-utils-4.0 4.0.1-1~bpo50+1 XEN administrative tools
ii xen-utils-common 4.0.0-1~bpo50+1 XEN administrative tools – common files
ii xenstore-utils 4.0.1-1~bpo50+1 Xenstore utilities for Xen

This fixed the problem of spontaneous reboots for all of the systems I have migrated …. until now! Last week I tried this solution on my private system and it failed in some way.

A full dump of the syslog is also available. It looks like /etc/rcS.d/S03udev is hanging in some way .. the syslog for this looks like:

Oct 29 20:51:07 mordor kernel: [ 17.728120] Code: 04 00 00 00 00 83 3c 24 10 74 32 77 0c 83 3c 24 08 0f 85 9f 00 00 00 eb 12 83 3c 24 20 74 30 83 3c 24 40 0f 85 8d 00 00 00 eb 35 <0f> b6 01 8b 54 24 20 89 02 c7 42 04 00 00 00 00 eb 79 0f b7 01
Oct 29 20:51:07 mordor kernel: [ 17.731554] EIP: [] acpi_ex_system_memory_space_handler+0×189/0×221 SS:ESP 0069:ec135d74
Oct 29 20:51:07 mordor kernel: [ 17.731791] CR2: 00000000eda86000
Oct 29 20:51:07 mordor kernel: [ 17.731881] —[ end trace 7b8af581772a1c55 ]—
Oct 29 20:51:07 mordor kernel: [ 18.116622] input: PC Speaker as /devices/platform/pcspkr/input/input6
Oct 29 20:51:07 mordor kernel: [ 18.427607] Error: Driver ‘pcspkr’ is already registered, aborting…
Oct 29 20:51:07 mordor kernel: [ 232.381814] Adding 10000452k swap on /dev/sda2. Priority:-1 extents:1 across:10000452k

The dom0 is there hanging in some way … I don’t have an exact way to get the system proceed from there … usually I need to wait some seconds and pressing ^C after a while will get the box up.
Anyways … I remember there are some problems with udev >= 150 and older kernels, which is also documented in the squeeze release notes. Maybe there is also a problem with older udev and newer kernels? But why is this combination working on other systems? Booting linux-image-2.6.32-bpo.5-686 works without any issue anyways, so it seems xen related. Any idea is appreciated!
UPDATE: I’ve also reported this via #602109.

[selfnote] Change CD-image while (HVM) domU is running

cyco — Sun, 29 Nov 2009 15:23:15 +0000

xm list | grep <VMName>
xm block-list <VMName_Number>
xm block-detach <VMName_Number> <VMName_Vdev> -f
xm block-attach <VMName_Number> file:///iso/disk.iso \
hdc:cdrom r

New server hardware for svn.kvirc.de

cyco — Fri, 19 Jun 2009 04:15:12 +0000

Maybe some of you have noticed, svn.kvirc.de was running under heavy load recently while it was smashed by search engine crawlers.

First we tried to tweak the trac installation, which caused the heavy load, when running the spiders trough. But it turned out, that this doesn’t has the effect to recover the situation, since the trac tree is growing (with the svn tree and revisions). So the spiders have more to crawle and taking more resources away.

We checked our relations and we where lucky to find with waaf.net a sponsor for new server hardware.

Some features of our new ProLiant DL360 server we baptized ‘baikonur‘:

* Dual Core Processor
* 2 GB RAM
* Compaq Smart Array Raid Controller
* 2x 72.8 GB IBM Ultra3 10K
* 2x Gigabit Ethernet
* Integrated Lights Out

The migration was quite relaxed. svn.kvirc.de is a virtual maschine, so we did setup the new host and proudly racked it up at the bandnew hosting center of TMT Teleservice as first (sponsored) customer. The last step was to migrate the system over to the new host and change the DNS records. The outage was something about 5-10 minutes basically caused by the TTL of 600.

Looking into the Load Average I think we have enought resources for a couple of more svn revisions and search engine spiders.

That’s all, happy ircing, and happy developing.

windows in domU on etch. where is the showstopper?

cyco — Tue, 24 Apr 2007 18:29:17 +0000

Since some days I’m trying to install windows into a domU just for fun. All went fine until accessing discs. Sometime partition is not formatable, sometimes files could not read from cdrom. To be sure, it’s not the disc image, I used a fresh one from a MSDN Subscription, but it didn’t help.

Searching around on web, I found that I’m not the only one, with this issue. The common(s) seems to be etch (and AMD CPU).

I found traces of the issue at SP Team Blog, xen-users and xen-forum (german).

Maybe anybody can enlighten me?

Taking Xen guests (domU) alive via routed network

cyco — Thu, 29 Mar 2007 12:02:46 +0000

The xend config looks like this:

playground:~# grep -v ^# /etc/xen/xend-config.sxp
(network-script ‘network-bridge netdev=dummy0′)
(vif-script vif-bridge)
(dom0-min-mem 196)
(dom0-cpus 0)

Creating volume group for Xen domU (/dev/hdb2 was allready created with id 8e):

playground:~# pvcreate /dev/hdb2
Physical volume “/dev/hdb2″ successfully created
playground:~# vgcreate xen-vol /dev/hdb2
Volume group “xen-vol” successfully created
playground:~# vgscan
Reading all physical volumes. This may take a while…
Found volume group “xen-vol” using metadata type lvm2

Creating first domU:

playground:~# xen-create-image –ip 192.168.222.50 –hostname xetch.blubb.cyconet.org
General Infomation
——————–
Hostname : xetch.blubb.cyconet.org
Distribution : etch
Fileystem Type : ext3
Size Information
—————-
Image size : 4Gb
Swap size : 1024Mb
Image type : full
Memory size : 128Mb
Kernel path : /boot/vmlinuz-2.6.18-4-xen-vserver-686
Initrd path : /boot/initrd.img-2.6.18-4-xen-vserver-686
Networking Information
———————-
IP Address 1 : 192.168.222.50
Netmask : 255.255.255.0
Gateway : 192.168.222.1
Creating ext3 filesystem on /dev/xen-vol/xetch.blubb.cyconet.org-disk
Done
Installing your system with debootstrap mirror ftp://ftp.de.debian.org/debian/
Done
Running hooks
Done
No role script specified. Skipping
Creating Xen configuration file
Done
Setting up root password
Enter new UNIX password:
etype new UNIX password:
passwd: password updated successfully
All done
Logfile produced at:
/var/log/xen-tools/xetch.blubb.cyconet.org.log

Add the following to /etc/network/interfaces:

auto dummy0
iface dummy0 inet static
address 192.168.222.1
netmask 255.255.255.0
network 192.168.222.0
broadcast 192.168.222.255

Bringing up network interface, restarting xen and starting the domU:

playground:~# ifup dummy0
playground:~# /etc/init.d/xendomains stop
playground:~# /etc/init.d/xend restart
playground:~# /etc/init.d/xendomains start
playground:~# xm create xetch.blubb.cyconet.org.cfg -c
Using config file “/etc/xen/xetch.blubb.cyconet.org.cfg”.
…

If you want to autostart your domUs, you have to symlink the configfile to /etc/xen/auto:

playground:~# mkdir /etc/xen/auto/
playground:~# ln -s /etc/xen/xetch.blubb.cyconet.org.cfg /etc/xen/auto/xetch.blubb.cyconet.org.cfg

Getting started with XEN on etch

cyco — Wed, 28 Mar 2007 16:32:10 +0000

Today I thought it’s time to look deeper into the Xen hype. So I decided to install xen on etch. I followed Ganneffs recommandation

playground:~# aptitude install bridge-utils iproute \
xen-ioemu-3.0.3-1 xen-utils-3.0.3-1\
xen-linux-system-2.6.18-4-xen-vserver-686 libc6-xen

Booting the dom0 leeds me into kernel panic, cause no root device was found. Hmm … there seems to be the driver missing. Right – there wasn’t any /boot/initrd.img-2.6.18-4-xen-vserver-686.
Waldi did hint me to /var/lib/dpkg/info/linux-image-2.6.18-4-xen-vserver-686.postinst and I got my ramdisk this way:

playground:~# update-initramfs -c -t -k \
2.6.18-4-xen-vserver-686
playground:~# command -v update-grub > \
/dev/null && update-grub

After booting dom0 successfully, my serial console disappeared. After some digging around, I got the clue I need to tell xen, that there is one.
I modified /boot/grub/menu.lst:

# xenhopt=com1=115200,8n1 console=com1
..
# xenkopt=console=vga

Dom0 is running now, lets see what we can do next, maybe:

playground:~# aptitude install xen-tools