1.4: install debian 5.03 freezes with kvm

[fromport]

Today final pieces of hardware came in to finish my new server setup:
supermicro 6026TT.
Install of (final) 1.4 went smooth.
Copied some templates and .iso images (debian-503-amd64-netinst.iso & debian-503-i386-netinst.iso) onto the local harddrive.
Creating a new guest also went fine.
When i started it, the familiar bootup screen of debian came up, i choose expert install and then screen/console goes to "blank" with a underscore cursor in the top left corner. I tried both 64 & 32 bit images: same restults

pveversion -v
pve-manager: 1.4-9 (pve-manager/1.4/4390)
qemu-server: 1.1-6
pve-kernel: 2.6.24-16
pve-qemu-kvm: 0.11.0-2
pve-firmware: 1
vncterm: 0.9-2
vzctl: 3.0.23-1pve3
vzdump: 1.2-3
vzprocps: 2.0.11-1dso2
vzquota: 3.0.11-1
vhost1:/vz/images#

pveperf output:

# 
pveperf  
CPU BOGOMIPS:      76608.34
REGEX/SECOND:      669422
HD SIZE:           73.33 GB (/dev/pve/root)
BUFFERED READS:    112.15 MB/sec
AVERAGE SEEK TIME: 10.82 ms
FSYNCS/SECOND:     1503.09
DNS EXT:           45.52 ms
DNS INT:           19.42 ms

I burned the 64 bit image to a cd-rw to check if it would install on real hardware and it functioned without a hitch. So the image is (imho) not a suspect.

Here is the config of the guest:

vhost1:/etc/qemu-server# cat 101.conf 
name: lenny64
ide2: local:iso/debian-503-amd64-netinst.iso,media=cdrom
vlan0: virtio=9E:68:07:28:01:94
virtio0: local:101/vm-101-disk-1.raw
ostype: l26
memory: 1024
sockets: 1
boot: da
freeze: 0
cpuunits: 1000
acpi: 1
kvm: 1

Anyone some hints/tips/clues ?

 

[dietmar]

Please can you try with ide instead of virtio?

 

[fromport]

Deleted virtio ethernet & disk.
Replaced with ide harddrive & e1000 ethernet

name: lenny64
ide2: local:iso/debian-503-amd64-netinst.iso,media=cdrom
ostype: l26
memory: 1024
sockets: 1
boot: da
freeze: 0
cpuunits: 1000
acpi: 1
kvm: 1
ide0: local:101/vm-101-disk-1.raw
vlan0: e1000=F2:95:02:FE:EB:16

Same behaviour

 

[fromport]

Hate to follow-up on my self, but:

I just uploaded/started a copy of "systemrescuecd" and it has the same behaviour:

http://dth.net/tmp/srcd_screendump.jpg

 

[tom]

Check if you got the latest bios and if VT is enabled.

Debian is known to work without issues on 1.4 (KVM and OpenVZ).

 

[fromport]

Check if you got the latest bios and if VT is enabled.
Debian is known to work without issues on 1.4 (KVM and OpenVZ).

I've been running 1.4beta on different hardware (dell r200) without problems.
It's just this (new) server.

I checked, no more recent bios available and VT is enabled:

http://dth.net/tmp/bios1.jpg
http://dth.net/tmp/bios2.jpg

While trying i had the console open through IPMI console and i caught
an error while the guest froze:

http://dth.net/tmp/error_while_freeze.jpg

Does this reveal any hints/clues to what is happening ?

 

[fromport]

Started memtest and it froze after 34 seconds.

Hardware problems

http://dth.net/tmp/memory_test_freezing.jpg


Will investigate, sorry for wasting brain/cpu cycles.

 

[fromport]

*update*
Mu problems aren't solved (yet).
Went through 3 different support channels. (first memory manufacterer, and twice mobo/server manufacterer)
Eventually even got an experimental bios.
Since there are 4 motherboards in the chassis i can test 3 different bios'es at the moment.
It looks like the hardware isn't the problem.
Memtest86+ ran fine for 7 hours on the hardware.
http://dth.net/supermicro/memtest86_completed_7hr.jpg
Installing an os (i tried ubuntu karmic (officially out in 6 days)) on the bare metal is not a problem. All install/complete without a hitch.

Even a kernel compile on the machine is stable

time make-kpkg kernel-image --initrd
[SNIP]
dpkg-deb: building package `linux-image-2.6.32-rc5-git3' in `../linux-image-2.6.32-rc5-git3_2.6.32-rc5-git3-10.00.Custom_amd64.deb'.
make[2]: Leaving directory `/usr/src/linux-2.6.32-rc5-git3'
make[1]: Leaving directory `/usr/src/linux-2.6.32-rc5-git3'
real 8m27.486s
user 45m56.360s
sys 4m20.510s

On a different HD i installed ubuntu-karmic-amd64 (officially stable in 6 days) and did a quick setup of libvirt. I really gives me the same problems:
The moment i startup a guest i get an "exception" error:
http://dth.net/supermicro/ubuntu_karmic_libvirt_exception_error_after guest_start.jpg
This is on the "standard" ubuntu kernel

# uname -a
Linux ubuntu 2.6.31-14-server #48-Ubuntu SMP Fri Oct 16 15:07:34 UTC 2009 x86_64 GNU/Linux

Complete kern.log of the ubuntu install is here:
http://dth.net/supermicro/kern.log_ubuntu_karmic_2.6.31-14-server

So that brings me back to proxmox.
Although this hardware should be the latest/greatest:
chassis: http://supermicro.com/products/system/2U/6026/SYS-6026TT-BIBQRF.cfm?INF=
Motherboard: http://supermicro.com/products/motherboard/QPI/5500/X8DTT-F.cfm
2 x E5530 cpu, 3x4GB ecc kingston RAM.
It doesn't work

Kern.log of proxmox 1.4 is here: http://dth.net/supermicro/kern.log_proxmox
Since people have probably more experience with this matter, is there anyone that could point me in the right direction ?

_i_ don't know what to do any more apart from abondoning the hardware and switch to another platform.

Danny

 

[dietmar]

http://dth.net/supermicro/kern.log_proxmox[/URL]

I can't see any errors in this log?

 

[fromport]

I can't see any errors in this log?

this is the error in proxmox, which is similar to the one i get in ubuntu karmic with libvirt:

handle_exception: unexpected, vectoring info 0x80000410 intr info 0x80000b0d

I took 2 screendumps:

first one is when everything is ready to rock:

I've defined a guest, to start from a cd image.
It actually is started already and you see the familiar bootscreen of debian-5.03-netinstall.

http://dth.net/supermicro/guest_started.jpg

When i hit "return" or what i usually do: go to expert install mode and then hit return the next thing that happens is this:

http://dth.net/supermicro/guest_hang.jpg

Left top screen is the "console" of the bare-hardware (seen through IPMI)
Left down screen is a ssh session with "dmesg" output after the guest stopped responding.
Right top screen is the vnc output of the guest, obviously not responding.
The kvm process is still running at that point. first 30 seconds it eats a lot of cpu cycles and then stops eating cpu cycles:

vhost1:~# ps auxw |grep kvm
root      5408 33.4  0.0 1149548 17772 ?       Sl   09:50   3:55 /usr/bin/kvm -monitor unix:/var/run/qemu-server/101.mon,server,nowait -vnc unix:/var/run/qemu-server/101.vnc,password -pidfile /var/run/qemu-server/101.pid -daemonize -usbdevice tablet -name lenny64 -smp sockets=1,cores=1 -boot da -vga cirrus -tdf -k en-us -drive file=/var/lib/vz/images/101/vm-101-disk-1.raw,if=ide,index=0 -drive file=/var/lib/vz/template/iso/debian-503-amd64-netinst.iso,if=ide,index=2,media=cdrom -m 1024 -net tap,vlan=0,ifname=vmtab101i0,script=/var/lib/qemu-server/bridge-vlan -net nic,vlan=0,model=e1000,macaddr=F2:95:02:FE:EB:16 -id 101 -cpuunits 1000

So the proxmox server is still responding, no problems there.
I've tested with openvz guests and that runs just fine.
I think i have triggered some kvm "feature" which is specifically to my hardware.
I opened a request at the kvm mailling list yesterday
http://www.spinics.net/lists/kvm/msg24584.html