On Tue, Sep 13, 2011 at 2:25 AM, Steven Timm <[email protected]> wrote:
> On Mon, 12 Sep 2011, James Kelly wrote:
>
> Hello SL users,
>>
>> When I install sl5.6 and sl5.7 64 bit guests on a KVM sl6.1 64 bit host I
>> am
>> seeing the following errors (when the virtual machine starts or restarts):
>>
>> Sep 11 20:00:34 kvm-sl6x kernel: kvm: 2022: cpu0 unimplemented perfctr
>> wrmsr: 0xc0010005 data 0x0
>> Sep 11 20:00:34 kvm-sl6x kernel: kvm: 2022: cpu0 unimplemented perfctr
>> wrmsr: 0xc0010001 data 0x130076
>> Sep 11 20:00:34 kvm-sl6x kernel: kvm: 2022: cpu0 unimplemented perfctr
>> wrmsr: 0xc0010005 data 0xffffffffffde6d26
>> Sep 11 20:00:34 kvm-sl6x kernel: kvm: 2022: cpu0 unimplemented perfctr
>> wrmsr: 0xc0010001 data 0x530076
>>
>
> I'm pretty sure the wrmsr errors above are harmless..we have seen
> them on FermiCloud under both SLF5 and SLF6. The error below is
> unconnected.
>
> Thanks for your reply Steven. I was also told something similar on the KVM
irc channel, but it is good to have this confirmed.
>
>
>> When i tried an sl5.6 32 bit guest the virsh console connection from the
>> host to the guest crashed with following error:
>>
>> *** glibc detected *** /usr/bin/virsh: corrupted double-linked list:
>> 0x000000000101b1c0 ***
>>
>> The install had finished and the guest was being rebooted when this
>> error occurred. I had to issue a virsh destroy command from a second ssh
>> connection to get get rid of the frozen virsh console connection.
>> Strangely
>> the vm did boot when i later issued a virsh start command.
>>
>
> What do you see when you bring up the VM with virt-viewer or virt-manager,
> sometimes you can see some other messages there before the virsh console
> is available.
>
I don't use virt-viewer or virt-manager. The host and guest do not have any
x or gui stuff on them. I use putty/ssh to install the vm's with
virt-install using a script such as the following:
virt-install \
--name=vm_data \
--ram=1024 \
--vcpus=1 \
--os-type=linux \
--os-variant=rhel5.4 \
--disk
path=/dev/vg_one_libvirt/lv_data,device=disk,bus=virtio,sparse=true,format=raw
\
-w bridge=br0,model=virtio \
--nographics \
-x console=ttyS0 \
--location=http://linuxsoft.cern.ch/scientific/56/x86_64/
The sever has software raid1 and the vm's are stored in an libvirt created
storage pool:
# virsh -c qemu:///system pool-list
Name State Autostart
-----------------------------------------
pool_one active yes
pool_two active yes
Each storage pool is on a seperate pair of raid1 partitions and the logical
volumes of the host are as follows:
# lvs
LV VG Attr LSize Origin Snap% Move Log Copy%
Convert
LogVolDevShm vg_kvmsl6x -wi-a- 500.00m
LogVolHome vg_kvmsl6x -wi-ao 996.00m
LogVolRoot vg_kvmsl6x -wi-ao 2.93g
LogVolSwap vg_kvmsl6x -wi-ao 2.93g
LogVolTmp vg_kvmsl6x -wi-ao 2.93g
LogVolVar vg_kvmsl6x -wi-ao 3.91g
LogVolVarLog vg_kvmsl6x -wi-ao 5.37g
lv_data vg_one_libvirt -wi-ao 19.00g
lv_smb vg_two_libvirt -wi-a- 193.00g
The guests are installed with similar partitioning to the KVM host.
The virt-install script I use, starts a vm and connects to its console. I
select the vnc install option part way the the text based install. When the
install is finished click 'reboot' in the tightvnc vnc viewer this closes
the tightvnc connection. I then watch the console that is still running in
my putty/ssh connection to the kvm host server.
I have seen the glibc error on this console as the vm tries to reboot follow
a fresh install. I have also seen this glibc error while using;
virsh -c qemu:///system console vm_guest_name
to connect to a vm and watch the console output while I rebooted the vm from
a second ssh connection (connected directly to the vm itself).
In each case the vm shuts down ok and then hangs whilst trying to restart. I
tap enter to see what is happening and I get the glibc error displayed.
If I then destroy the vm and start it again it boots ok!
Unfortunately, I am not a linux desktop user, but if you think it will help
I will find a spare computer, install sl6.1, and try and connect to the kvm
host with virt-manager / ssh. Would it be worth doing this?
>
>> Installing an sl6.1 64 bit guest on the sl6.1 64 bit host I get the this
>> error as the vm starts or restarts:
>>
>> Sep 11 14:05:14 kvm-sl6x kernel: kvm: 1928: cpu0 unhandled rdmsr:
>> 0xc0010001
>>
>> What are your thoughts on this error?
> Can anyone offer any advice on whether I have hit a bug somewhere of
>> whether
>> I have done something wrong with the configuration of the host/guest?
>>
>> jk
>>
>>
