Re: VM networking stops working

[Mike <mike-kcIQ9yavhvsV/sYkb9DZNQ@public.gmane.org>]

Izik Eidus schrieb:
> Mike wrote:
>> Izik Eidus schrieb:
>>  
>>> Mike wrote:
>>>    
>>>> Izik Eidus schrieb:
>>>>  
>>>>      
>>>>> Mike wrote:
>>>>>           
>>>>>> Izik Eidus schrieb:
>>>>>>  
>>>>>>               
>>>>>>> Mike wrote:
>>>>>>>                      
>>>>>>>> Hello,
>>>>>>>> I have a problem with my virtual machines. I am using kvm53
>>>>>>>> on an AMD Athlon(tm) 64 X2 Dual Core Processor 5000+.
>>>>>>>> As distribution I use Ubuntu server 7.10.
>>>>>>>> As kernel I use Linux 2.6.22-14-server #1 SMP Sun Oct 14 
>>>>>>>> 22:09:15 GMT 2007 x86_64 GNU/Linux.
>>>>>>>>
>>>>>>>> I have 3 different vms running. At the beginning they all work 
>>>>>>>> without a problem.
>>>>>>>> Then, after one moment to another one or more stop working.
>>>>>>>> I did a ping test to all the machines, the vm with the failure 
>>>>>>>> stops pinging.
>>>>>>>> The result from external ping: Destination host unreachable.
>>>>>>>> The same result from a ping onto the main server.
>>>>>>>>
>>>>>>>> Here are my configuration files:
>>>>>>>> cat /etc/qemu-ifup
>>>>>>>> #!/bin/sh
>>>>>>>> /sbin/ifconfig $1 0.0.0.0 promisc up
>>>>>>>> /usr/sbin/brctl addif br0 $1
>>>>>>>> sleep 2
>>>>>>>> ------------------------------------------------------------------------ 
>>>>>>>>
>>>>>>>> cat /etc/network/interfaces
>>>>>>>> auto lo
>>>>>>>> iface lo inet loopback
>>>>>>>>
>>>>>>>> auto br0
>>>>>>>> iface br0 inet static
>>>>>>>> address 195.24.77.169
>>>>>>>> netmask 255.255.255.0
>>>>>>>> gateway 195.24.77.1
>>>>>>>> bridge_ports eth0
>>>>>>>> bridge_stp off
>>>>>>>> bridge_maxwait 5
>>>>>>>> ------------------------------------------------------------------------ 
>>>>>>>>
>>>>>>>> I use scripts for starting the vms:
>>>>>>>> kvm -hda apache.img -hdb apache_storage.img -m 512 -boot c -net 
>>>>>>>> nic,vlan=0,macaddr=00:16:3e:00:00:01 -net tap -nographic 
>>>>>>>> -daemonize
>>>>>>>> kvm -hda postfix.img -m 512 -boot c -net 
>>>>>>>> nic,vlan=0,macaddr=00:16:3e:00:00:12 -net tap -nographic 
>>>>>>>> -daemonize
>>>>>>>> kvm -hda openvpn.img -m 256 -boot c -net 
>>>>>>>> nic,vlan=0,macaddr=00:16:3e:00:00:03 -net tap -nographic 
>>>>>>>> -daemonize
>>>>>>>> ------------------------------------------------------------------------ 
>>>>>>>>
>>>>>>>> The tapN devices are being created at the startup of the vms.
>>>>>>>>
>>>>>>>> When I run dmesg I have this:
>>>>>>>> [  112.390943] br0: port 2(tap1) entering learning state
>>>>>>>> [  112.391919] br0: port 3(tap0) entering learning state
>>>>>>>> [  112.397605] br0: port 4(tap2) entering learning state
>>>>>>>> [  122.674313] tap0: no IPv6 routers present
>>>>>>>> [  122.729185] tap1: no IPv6 routers present
>>>>>>>> [  122.843901] tap2: no IPv6 routers present
>>>>>>>> [  123.336081] kvm: emulating exchange as write
>>>>>>>> [  127.367940] br0: topology change detected, propagating
>>>>>>>> [  127.367949] br0: port 4(tap2) entering forwarding state
>>>>>>>> [  127.362971] br0: topology change detected, propagating
>>>>>>>> [  127.362977] br0: port 2(tap1) entering forwarding state
>>>>>>>> [  127.362978] br0: topology change detected, propagating
>>>>>>>> [  127.362980] br0: port 3(tap0) entering forwarding state
>>>>>>>>
>>>>>>>> No indication of a problem, nowhere in the log files.
>>>>>>>>
>>>>>>>> I don't know if the vms are freezing, or the network forwarding 
>>>>>>>> stops working.
>>>>>>>> I also activated "net.ipv4.conf.default.forwarding=1"
>>>>>>>>
>>>>>>>> I googled days for that problem, but without a result.
>>>>>>>> I hope that maybe you have a solution for this problem. That 
>>>>>>>> would be great, because I prefer using kvm instead of other 
>>>>>>>> virtualization systems.
>>>>>>>>
>>>>>>>> Thank you very much in advance,
>>>>>>>> Greetings from Luxembourg.
>>>>>>>>
>>>>>>>> Have a nice day,
>>>>>>>> Mike
>>>>>>>>                               
>>>>>>> it might be not the network that die, but the whole vm
>>>>>>> can you run the vms with -vnc ? (first vm -vnc :1 secoend vm 
>>>>>>> -vnc :2 third vm -vnc :3)
>>>>>>> and try to connect to them with vncviewer and see if they are 
>>>>>>> alive?
>>>>>>>
>>>>>>>                         
>>>>>> Thank you for your reply.
>>>>>> I can't connect to the vm if it has died, even with vnc. I tried 
>>>>>> this already, no reply from the host.
>>>>>> I don't have physical access to the server because its located in 
>>>>>> a dc. Only ssh to the host.
>>>>>>                   
>>>>> ok this mean there is bug in kvm for amd and it not realted to the 
>>>>> network
>>>>> what is the guest that you use? is it 64/32?
>>>>>
>>>>>             
>>>> I'm using Debian Etch 32 Bit for all guests
>>>> with Kernel Linux apache 2.6.18-5-486 #1 Tue Oct 2 23:38:54 UTC 
>>>> 2007 i686 GNU/Linux
>>>>
>>>> ------------------------------------------------------------------------- 
>>>>
>>>>         
>>> ok but what more interesting me is:
>>> if you run just one vm
>>> is it work?
>>>     
>> Yes. If I run only 1 vm, it runs days without problems.
>> I have sometimes have peaks in reply time from the guest.
>> e.g.
>> bytes=32 time=34ms TTL=57
>> bytes=32 time=34ms TTL=57
>> bytes=32 time=119ms TTL=57
>> bytes=32 time=400ms TTL=57
>> bytes=32 time=540ms TTL=57
>> bytes=32 time=947ms TTL=57
>> bytes=32 time=754ms TTL=57
>> bytes=32 time=850ms TTL=57
>> bytes=32 time=925ms TTL=57
>> bytes=32 time=2032ms TTL=57
>> bytes=32 time=897ms TTL=57
>> bytes=32 time=197ms TTL=57
>> bytes=32 time=34ms TTL=57
>> bytes=32 time=34ms TTL=57
>>
>> but if I ping the host server the same time, the reply time is 
>> normal, around 34ms
>>
>> ------------------------------------------------------------------------- 
>>
>> This SF.net email is sponsored by: Microsoft
>> Defy all challenges. Microsoft(R) Visual Studio 2005.
>> http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
>> _______________________________________________
>> kvm-devel mailing list
>> kvm-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org
>> https://lists.sourceforge.net/lists/listinfo/kvm-devel
>>   
> ok sound interesting,
> you are sure you are using kvm53 modules?
> (one issue like that solved at kvm53, but maybe we have another one...)
>

> I had similar problems for my first week or so of trying to use KVM
> for my virtualization needs. My problem was solved by adding a
> "model=rtl8139" to the first -net option on the kvm/qemu command.
> The default ne2k emulation seems to be broken. Another possibility
> is problems with macaddr. In my testing I found that several of my
> machines were getting the same macaddr by default so now I specify
> them explicitly for each VM (or generate a dummy one based on the
> time for testing).
>
> An example from one of my KVMMs (hopefully line wrapping won't
> interfere):
>
> /usr/local/bin/qemu-system-x86_64 -name DOM9-BSD62 -hda /vm_img/
> bsd_in1/dom9-disk1.qcow \
> -hdb /vm_img/bsd_in1/dom9-disk2.raw -m 512 -boot c -smp 2 -usb -
> usbdevice tablet -vnc :9 \
> -serial telnet::4109,server,nowait -monitor telnet::
> 4209,server,nowait \
> -net nic,vlan=0,macaddr=52:54:09:12:34:56,model=rtl8139 \
> -net tap,vlan=0,ifname=tap1,script=/etc/kvm/kvm-ifup
>
>
> Lynn Kerby
Thank you for this hint,
I'll try this immediately.



-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/