incubator-tashi-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Hunter Pitelka <hpite...@andrew.cmu.edu>
Subject Re: Issues with CreateVM
Date Mon, 24 Oct 2011 13:56:06 GMT
Kushal,

We have pretty strict firewall policies installed by default on all 
those machines, so you might want to run `iptables -F` just to clear them 
out.

Once you have the machine nothing should be stopping it from connecting to 
the DHCP server on the internal network.  Just make sure that the virtual 
bridge is set up correctly and has an ip address.

If you can't get a DHCP addr, try just manually assigning it an address 
and then pinging the ip address of the bridge. If you can't ping the 
physical host, then your tap device is not connected to the bridge (brctl 
will help you out there).

Also, what Michael said about getting tcpdump running helped me 
tremendously when debugging networking issues on virtual machines.

~Hunter

On Sun, 23 Oct 2011, Kushal Dalmia wrote:

> Hi,
>
> We tried the vnc option and when we see the VM console while the VM boots
> up, we see that the IP configuration for eth0 fails (being a RHEL image,
> marked by a [FAILED] during the boot process). We think this might be the
> cause for not being able to ping/ssh into the VM. Our qemu-if script looks
> like:
>
> #!/bin/sh
> /sbin/ifconfig $1 0.0.0.0 up
> /usr/sbin/brctl addif mybridge $1
> exit 0
>
> Are we supposed to add anything else to this?
>
> Thanks,
> Kushal
>
> On Sun, October 23, 2011 9:06 pm, Richard Gass wrote:
>> You can try to use the vmmspecificcall to start vnc to see the console
>> of the vm.
>>
>> #tashi vmmspecificcall --instance vmname --arg startVnc
>>
>>
>> Then you use vnc
>>
>>
>> vncviewer Vmname
>>
>> Give that a go.
>>
>>
>> Thanks
>>
>>
>> --
>> Richard
>>
>>
>> Sent from my mobile device
>>
>>
>> On Oct 24, 2011, at 2:30, Kushal Dalmia <kdalmia@andrew.cmu.edu> wrote:
>>
>>
>>> Hi Micheal,
>>>
>>>
>>> Thanks for that. We were missing one of the tun-tap modules. Now we can
>>>  create a VM which shows up in "getInstances".
>>>
>>> However we are not able to SSH into it(as a matter of fact, we are not
>>> able to ping the host as well). We have disabled DHCP and DNS in our
>>> tashi config since all our tashi setup is on a single machine.The IP
>>> address it is getting is from the range 192.168.1.xx.
>>>
>>> We have setup a network object in the config with network id=1 and
>>> name='global'. Do we need any further setup to be able to ssh into the
>>> VM?
>>>
>>>
>>> Thanks,
>>> Kushal
>>>
>>>
>>> On Sun, October 23, 2011 7:53 pm, Michael Stroucken wrote:
>>>
>>>> Kushal Dalmia wrote:
>>>>
>>>>
>>>>> Please see inline comments.
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>> Keep in mind that there is some debug output put in /tmp by the
>>>>>> nodemanager currently (like /tmp/20037.{out|err}). It should have
>>>>>> something there, because I see that the nodemanager tries to
>>>>>> create VMs.
>>>>>>
>>>>>>
>>>>>>
>>>>>
>>>>> My /tmp/<pid>.err file looks like this:
>>>>> /etc/qemu-ifup.1: could not launch network script
>>>>> qemu-system-x86_64: -net
>>>>> tap,ifname=tashi4.0,vlan=1,script=/etc/qemu-ifup.1: Device 'tap'
>>>>> could not be initialized
>>>>>
>>>>> The /etc/qemu-ifup.1 script looks like this:
>>>>> #!/bin/sh
>>>>>
>>>>>
>>>>>
>>>>> /sbin/ifconfig $1 0.0.0.0 up
>>>>> /usr/sbin/brctl addif mybridge $1
>>>>> exit 0
>>>>>
>>>>>
>>>>>
>>>>
>>>> Is /etc/qemu-ifup.1 executable (chmod 755 /etc/qemu-ifup.1)?
>>>>
>>>>
>>>>
>>>> Are you running as root?
>>>>
>>>>
>>>>
>>>> Do you have the tun module loaded?
>>>>
>>>>
>>>>
>>>> http://turbolinux.org/2011/05/etcqemu-ifupcould-not-launch-network-sc
>>>> ript -could-not-initialize-device-tap/
>>>>
>>>>
>>>>
>>>> Greetings,
>>>> Michael.
>>>>
>>>>
>>>>
>>>>
>>>
>>>
>>
>>
>
>
>
>

Mime
View raw message