[one-users] VM cannot connect to outside (internet)
Sangram Rath
sangram.rath at gmail.com
Sat May 10 13:10:37 PDT 2014
Thanks Valentin. That worked.
I have another networking question though.
I have 5 fixed IPs that need to go to 5 vms I have inside opennebula. I
have 1 physical ethernet port and it is connected to a bridge now (br0)
that has one of those fixed IPs mentioned in it. (Using Cent OS)
What is the best way to have traffic route to these VMs based on IP
address.
On Sat, May 10, 2014 at 8:46 PM, Valentin Bud <valentin.bud at gmail.com>wrote:
> Hi Sangram,
>
> On Sat, May 10, 2014 at 1:51 PM, Sangram Rath <sangram.rath at gmail.com>wrote:
>
>> Hi Valentin,
>>
>> Thanks for getting back, really appreciate.
>> A bit of background:
>> Had opennebula 3.8 (all working fine), performed upgrade to 4.6 and
>> things screwed up. So ended up reinstalling (fresh install of 4.6). And we
>> are using images from the old installation. So contextualization was
>> working before and everything was fine. No changes done to file system
>> inside images till now. All images are runing Ubuntu 12.04 / Ubuntu 12.10.
>> One things to note here is that in all images, in file
>> /etc/network/interfaces eth0 is commented. And this was working before in
>> 3.8 (I did not setup 3.8). Right now I have manually uncommented eth0 so
>> that interface comes up.
>>
>
> I am wondering if the vmcontext script is really activated in the VM.
> I think you can check that with service --status-all on Ubuntu.
>
> If I were you I would update the one-context package on all VMs to 4.6.
> You can follow Basic Contextualization documentation [1] for that.
>
>
>>
>> 1 - Virtual network called "public" uses virbr0. A vm started with this
>> network gets an IP (192.168.122.x), different IP in sunstone and different
>> inside the VM (same ip range and if I uncomment eth0 manually). And I am
>> able to ping to internet.
>>
>
> Another clue that contextualization is not happening and that because of
> more
> than one reason as you'll below. But it's good that your network setup is
> working.
>
>
>>
>> 2 - The vms (5) should connect to internet. They will run different web
>> servers and we have 5 static IPs. If I type a web server name from browser
>> it should access that specific vm. Rest of the vms may or may not access
>> internet. So I will have a private network as well.
>>
>> 3 - *onevnet list*
>>
>> ID USER GROUP NAME CLUSTER TYPE BRIDGE
>> LEASES
>> 0 oneadmin oneadmin net0 - F
>> br0 0
>> 4 oneadmin oneadmin Public - R
>> virbr0 2
>> 5 oneadmin oneadmin Internet - R
>> br0 0
>> 6 oneadmin oneadmin Private - R
>> br1 0
>>
>> *************
>> *onevnet show 4*
>>
>> VIRTUAL NETWORK 4 INFORMATION
>> ID : 4
>> NAME : Public
>> USER : oneadmin
>> GROUP : oneadmin
>> CLUSTER : -
>> TYPE : RANGED
>> BRIDGE : virbr0
>> VLAN : No
>> USED LEASES : 2
>>
>> PERMISSIONS
>> OWNER : um-
>> GROUP : ---
>> OTHER : ---
>>
>> VIRTUAL NETWORK TEMPLATE
>> BRIDGE="virbr0"
>> DESCRIPTION=""
>> DNS="8.8.8.8"
>> GATEWAY="192.168.122.1"
>> NETWORK_ADDRESS="192.168.122.10"
>> NETWORK_MASK="255.255.255.0"
>> PHYDEV=""
>> VLAN="NO"
>> VLAN_ID=""
>>
>> RANGE
>> IP_START : 192.168.122.1
>> IP_END : 192.168.122.254
>>
>> LEASES ON HOLD
>> LEASE=[ MAC="02:00:c0:a8:7a:01", IP="192.168.122.1",
>> IP6_LINK="fe80::400:c0ff:fea8:7a01", USED="1", VID="-1" ]
>>
>> USED LEASES
>> LEASE=[ MAC="02:00:c0:a8:7a:02", IP="192.168.122.2",
>> IP6_LINK="fe80::400:c0ff:fea8:7a02", USED="1", VID="48" ]
>>
>> VIRTUAL MACHINES
>>
>> ID USER GROUP NAME STAT UCPU UMEM
>> HOST TIME
>> 48 oneadmin oneadmin Sfout staging ( runn 0 2G localhost
>> 0d 23h06
>>
>>
>> 4 - Yes the template has a NETWORK section.
>>
>> 5 - *onetemplate show 5*
>>
>> TEMPLATE 5 INFORMATION
>> ID : 5
>> NAME : Sfout staging (Ubuntu Server 11.10)
>> USER : oneadmin
>> GROUP : oneadmin
>> REGISTER TIME : 05/06 13:35:47
>>
>> PERMISSIONS
>> OWNER : um-
>> GROUP : ---
>> OTHER : ---
>>
>> TEMPLATE CONTENTS
>> CONTEXT=[
>> HOSTNAME="sfout.dev.redeyeelectronics.com" ]
>> CPU="4"
>> DISK=[
>> IMAGE="Sfout staging image",
>> IMAGE_UNAME="oneadmin" ]
>> GRAPHICS=[
>> LISTEN="0.0.0.0",
>> TYPE="VNC" ]
>> MEMORY="2048"
>> NIC=[
>> NETWORK="Public",
>> NETWORK_UNAME="oneadmin" ]
>>
>
> Your CONTEXT section is missing the NETWORK=YES bit to write the
> networking information inside the CONTEXT CD-ROM. I am almost sure that
> if you mount /dev/disk/by-label/CONTEXT into /mnt and read
> /mnt/context.sh, in the VM,
> you'll notice that it lacks any kind of networking information (ETH0_X
> variables).
>
> See Advanced Contextualization documentation [2]. Especially the Network
> Configuration
> section.
>
>
>>
>> 6- *onevm show *
>>
>> VIRTUAL MACHINE 48 INFORMATION
>> ID : 48
>> NAME : Sfout staging (Ubuntu Server 11.10)-48
>> USER : oneadmin
>> GROUP : oneadmin
>> STATE : ACTIVE
>> LCM_STATE : RUNNING
>> RESCHED : No
>> HOST : localhost
>> CLUSTER ID : -1
>> START TIME : 05/09 05:49:06
>> END TIME : -
>> DEPLOY ID : one-48
>>
>> VIRTUAL MACHINE MONITORING
>> NET_RX : 5M
>> USED MEMORY : 2G
>> USED CPU : 0
>> NET_TX : 165K
>>
>> PERMISSIONS
>> OWNER : um-
>> GROUP : ---
>> OTHER : ---
>>
>> VM DISKS
>> ID TARGET IMAGE TYPE SAVE SAVE_AS
>> 0 hda Sfout staging image file YES -
>>
>> VM NICS
>> ID NETWORK VLAN BRIDGE IP MAC
>> 0 Public no virbr0 192.168.122.2
>> 02:00:c0:a8:7a:02
>> fe80::400:c0ff:fea8:7a02
>>
>> VIRTUAL MACHINE HISTORY
>> SEQ HOST ACTION DS START TIME
>> PROLOG
>> 0 localhost none 0 05/09 05:49:13 0d 23h09m
>> 0h00m01s
>>
>> VIRTUAL MACHINE TEMPLATE
>> AUTOMATIC_REQUIREMENTS="!(PUBLIC_CLOUD = YES)"
>> CONTEXT=[
>> DISK_ID="1",
>> HOSTNAME="sfout.dev.redeyeelectronics.com",
>> TARGET="hdb" ]
>> CPU="4"
>> GRAPHICS=[
>> LISTEN="0.0.0.0",
>> PORT="5948",
>> TYPE="VNC" ]
>> MEMORY="2048"
>> TEMPLATE_ID="5"
>> VMID="48"
>>
>>
>> 7 - OS running inside VM is Ubuntu 12.04 / Ubuntu 12.10. The VM has a
>> network interface but it is commented in /etc/network/interfaces. When I
>> took over this setup it was like this and working. Of course in 3.8
>>
>
> The latest contextualization packages are supported on your OS so
> I see no reason why you shouldn't update them :).
>
>
>>
>>
>> Let me know if you need anything else.
>>
>>
> [1]:
> http://docs.opennebula.org/4.6/user/virtual_machine_setup/bcont.html#bcont
> [2]: http://docs.opennebula.org/4.6/user/virtual_machine_setup/cong.html
>
> Best,
> Valentin
>
>>
>>
>> On Sat, May 10, 2014 at 1:09 PM, Valentin Bud <valentin.bud at gmail.com>wrote:
>>
>>> Hi Sangram,
>>>
>>> On Fri, May 9, 2014 at 9:39 PM, Sangram Rath <sangram.rath at gmail.com>wrote:
>>>
>>>> Hi,
>>>>
>>>> Virtual machine gets an IP through contextualization, however virtual
>>>> machine is not able to connect to internet.
>>>> Also from inside the VM, I do not see any other interface apart from
>>>> lo. Is this normal in contextualization?
>>>>
>>>
>>> I wouldn't call it normal contextualization because the VM is missing
>>> the primary Ethernet
>>> interface, eth0. Let's try figure out why.
>>>
>>>
>>>>
>>>> I am able to ping the VM from same host. Host is Cent OS 6.1.
>>>> Host has br0 connected to interface eth0. And virbr0.
>>>>
>>>
>>> Have you defined a virtual network in OpenNebula? Does that network use
>>> br0 on virbr0?
>>>
>>> Where would you want your VMs to connect to, br0 or vribr0? If you want
>>> to isolate
>>> the VMs in a private network defined on virbr0 you have to enable IP
>>> forwarding on the
>>> host and either NAT or route the virbr0 network to the outside world.
>>>
>>> It would help in troubleshooting if you can post the output of onevnet
>>> list and onevnet show
>>> <the name of your virtual network or it id>.
>>>
>>> Has the template that you instantiate the VM from, a NETWORK section?
>>>
>>> Can you share the output of onetemplate show <name of template or id>?
>>>
>>> One more thing that can help is the output of onevm show <name of VM or
>>> id>.
>>>
>>> What OS are you running inside the VM? It's strange that the VM doesn't
>>> have
>>> a eth0 interface. You can also check the boot logs and search for
>>> Ethernet adapters.
>>>
>>> I also think that lspci output would help you. Where do you have the VM
>>> image
>>> from? Have you built it yourself? Maybe the udev rules are still present
>>> and the interface
>>> doesn't show up because of that.
>>>
>>> Best,
>>> Valentin
>>>
>>
>>
>>
>> --
>> Thanks,
>> Sangram Rath
>>
>
>
>
--
Thanks,
Sangram K Rath
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opennebula.org/pipermail/users-opennebula.org/attachments/20140511/0c6a0446/attachment-0002.htm>
More information about the Users
mailing list