[ovirt-users] More 4.1 Networking Questions
Charles Tassell
charles at islandadmin.ca
Mon Apr 10 10:54:35 UTC 2017
Hi Everyone,
Just an update, I installed a new Ubuntu guest VM and it was doing
the same thing regarding the network going down, then I disabled IPv6
and it's been fine for the past 10-15 minutes. So the issue seems to be
IPv6 related, and I don't need IPv6 so I can just turn it off. The eth1
NIC disappearing is still worrisome though.
On 2017-04-10 07:13 AM, Charles Tassell wrote:
> Hi Everyone,
>
> Thanks for the help, answers below.
>
> On 2017-04-10 05:27 AM, Sandro Bonazzola wrote:
>> Adding Simone and Martin, replying inline.
>>
>> On Mon, Apr 10, 2017 at 10:16 AM, Ondrej Svoboda <osvoboda at redhat.com
>> <mailto:osvoboda at redhat.com>> wrote:
>>
>> Hello Charles,
>>
>> First, can you give us more information regarding the duplicated
>> IPv6 addresses? Since you are going to reinstall the hosted
>> engine, could you make sure that NetworkManager is disabled
>> before adding the second vNIC (and perhaps even disable IPv6 and
>> reboot as well, so we have a solid base and see what makes the
>> difference)?
>>
> I disabled NetworkManager on the hosts (systemctl disable
> NetworkManager ; service NetworkManager stop) before doing the oVirt
> setup and rebooted to make sure that it didn't come back up. Or are
> you referring to on the hosted engine VM? I just removed and re-added
> the eth1 NIC in the hosted engine, and this is what showed up in dmesg:
> [Mon Apr 10 06:46:43 2017] pci 0000:00:08.0: [1af4:1000] type 00 class
> 0x020000
> [Mon Apr 10 06:46:43 2017] pci 0000:00:08.0: reg 0x10: [io 0x0000-0x001f]
> [Mon Apr 10 06:46:43 2017] pci 0000:00:08.0: reg 0x14: [mem
> 0x00000000-0x00000fff]
> [Mon Apr 10 06:46:43 2017] pci 0000:00:08.0: reg 0x20: [mem
> 0x00000000-0x00003fff 64bit pref]
> [Mon Apr 10 06:46:43 2017] pci 0000:00:08.0: reg 0x30: [mem
> 0x00000000-0x0003ffff pref]
> [Mon Apr 10 06:46:43 2017] pci 0000:00:08.0: BAR 6: assigned [mem
> 0xc0000000-0xc003ffff pref]
> [Mon Apr 10 06:46:43 2017] pci 0000:00:08.0: BAR 4: assigned [mem
> 0xc0040000-0xc0043fff 64bit pref]
> [Mon Apr 10 06:46:43 2017] pci 0000:00:08.0: BAR 1: assigned [mem
> 0xc0044000-0xc0044fff]
> [Mon Apr 10 06:46:43 2017] pci 0000:00:08.0: BAR 0: assigned [io
> 0x1000-0x101f]
> [Mon Apr 10 06:46:43 2017] virtio-pci 0000:00:08.0: enabling device
> (0000 -> 0003)
> [Mon Apr 10 06:46:43 2017] virtio-pci 0000:00:08.0: irq 35 for MSI/MSI-X
> [Mon Apr 10 06:46:43 2017] virtio-pci 0000:00:08.0: irq 36 for MSI/MSI-X
> [Mon Apr 10 06:46:43 2017] virtio-pci 0000:00:08.0: irq 37 for MSI/MSI-X
> [Mon Apr 10 06:46:43 2017] IPv6: ADDRCONF(NETDEV_UP): eth1: link is
> not ready
> [Mon Apr 10 06:46:43 2017] IPv6: eth1: IPv6 duplicate address
> fe80::21a:4aff:fe16:151 detected!
>
> Then when the network dropped I started getting these:
>
> [Mon Apr 10 06:48:00 2017] IPv6: eth1: IPv6 duplicate address
> 2001:410:e000:902:21a:4aff:fe16:151 detected!
> [Mon Apr 10 06:48:00 2017] IPv6: eth1: IPv6 duplicate address
> 2001:410:e000:902:21a:4aff:fe16:151 detected!
> [Mon Apr 10 06:49:51 2017] IPv6: eth1: IPv6 duplicate address
> 2001:410:e000:902:21a:4aff:fe16:151 detected!
> [Mon Apr 10 06:51:40 2017] IPv6: eth1: IPv6 duplicate address
> 2001:410:e000:902:21a:4aff:fe16:151 detected!
>
> The network on eth1 would go down for a few seconds then come back up,
> but networking stays solid on eth0. I disabled NetworkManager on the
> HE VM as well to see if that makes a difference. I also disabled IPv6
> with sysctl to see if that helps. I'll install a Ubuntu VM on the
> cluster later today and see if it has a similar issue.
>
>
>>
>> What kind of documentation did you follow to install the hosted
>> engine? Was it this page?
>> https://www.ovirt.org/documentation/how-to/hosted-engine/
>> <https://www.ovirt.org/documentation/how-to/hosted-engine/> If
>> so, could you file a bug against VDSM networking and attach
>> /var/log/vdsm/vdsm.log and supervdsm.log, and make sure they
>> include the time period from adding the second vNIC to rebooting?
>>
>> Second, even the vNIC going missing after reboot looks like a bug
>> to me. Even though eth1 does not exist in the VM, can you see it
>> defined for the VM in the engine web GUI?
>>
>>
>> If the HE vm configuration wasn't flushed to the OVF_STORE yet, it
>> make sense it disappeared on restart.
>>
> The docs I used were
> https://access.redhat.com/documentation/en-us/red_hat_virtualization/4.1/html/self-hosted_engine_guide/chap-deploying_self-hosted_engine#Deploying_Self-Hosted_Engine_on_RHEL
> which someone on the list pointed me to last week as being more
> up-to-date than what was on the website (the docs on the website don't
> seem to mention that you need to put the HE on it's own datastore and
> look to be more geared towards bare-metal engine rather than the VM
> self hosted option.)
>
> When I went back into the GUI and looked at the hosted engine config
> the second NIC was listed there, but it wasn't showing up in lspci on
> the VM. I removed the NIC in the GUI and re-added it, and the device
> appeared again on the VM. What is the proper way to "save" the state
> of the VM so that the OVF_STORE gets updated? When I do anything on
> the HE VM that I want to test I just type "reboot", but that powers
> down the VM. I then login to my host and run "hosted-engine
> --vm-start" which restarts it, but of course the last time I did that
> it restarted without the second NIC.
>
>>
>> The steps you took to install the hosted engine with regards to
>> networking look good to me, but I believe Sandro (CC'ed) would be
>> able to give more advice.
>>
>> Sandro, since we want to configure bonding, would you recommend
>> to install the engine physically first, move it to a VM,
>> according to the following method, and only then reconfigure
>> networking?
>> https://www.ovirt.org/documentation/self-hosted/chap-Migrating_from_Bare_Metal_to_an_EL-Based_Self-Hosted_Environment/
>> <https://www.ovirt.org/documentation/self-hosted/chap-Migrating_from_Bare_Metal_to_an_EL-Based_Self-Hosted_Environment/>
>>
>>
>>
>> I don't see why a diret HE deployment couldn't be done. Simone,
>> Martin can you help here?
>>
>>
>>
>> Thank you,
>> Ondra
>>
>> On Mon, Apr 10, 2017 at 8:51 AM, Charles Tassell
>> <ctassell at gmail.com <mailto:ctassell at gmail.com>> wrote:
>>
>> Hi Everyone,
>>
>> Okay, I'm again having problems with getting basic
>> networking setup with oVirt 4.1 Here is my situation. I
>> have two servers I want to use to create an oVirt cluster,
>> with two different networks. My "public" network is a 1G
>> link on device em1 connected to my Internet feed, and my
>> "storage" network is a 10G link connected on device p5p1 to
>> my file server. Since I need to connect to my storage network
>> in order to do the install, I selected p5p1 has the ovirtmgmt
>> interface when installing the hosted engine. That worked
>> fine, I got everything installed, so I used some ssh-proxy
>> magic to connect to the web console and completed the install
>> (setup a Storage domain and create a new network vmNet for VM
>> networking and added em1 to it.)
>>
>> The problem was that when I added a second network device
>> to the HostedEngine VM (so that I can connect to it from my
>> public network) it would intermittently go down. I did some
>> digging and found some IPV6 errors in the dmesg (IPv6: eth1:
>> IPv6 duplicate address 2001:410:e000:902:21a:4aff:fe16:151
>> detected!) so I disabled IPv6 on both eth0 and eth1 in the
>> HostedEngine and rebooted it. The problem is that when I
>> restarted the VM, the eth1 device was missing.
>>
>> So, my question is: Can I add a second NIC to the
>> HostedEngine VM and make it stick, or will it be deleted
>> whenever the engine VM is restarted?
>>
>>
>> When you change something in the HE Vm using the web ui, it has to be
>> saved also on the OVF_STORE to make it permanent for further reboot.
>> Martin can you please elaborate here?
>>
>>
>> Is there a better way to do what I'm trying to do, ie, should
>> I setup ovirtmgmt on the public em1 interface, and then
>> create the "storage" network after the fact for connecting to
>> the datastores and such. Is that even possible, or
>> required? I was thinking that it would be better for
>> migrations and other management functions to happen on the
>> faster 10G network, but if the HostedEngine doesn't need to
>> be able to connect to the storage network maybe it's not
>> worth the effort?
>>
>> Eventually I want to setup LACP on the storage network, but
>> I had to wipe the servers and reinstall from scratch the last
>> time I tried to set that up. I was thinking that it was
>> because I setup the bonding before installing oVirt, so I
>> didn't do that this time.
>>
>> Here are my /etc/sysconfig/network-scripts/ifcfg-* files in
>> case I did something wrong there (I'm more familiar with
>> Debian/Ubuntu network setup than CentOS)
>>
>> ifcfg-eth0: (ovirtmgmt aka storage)
>> ----------------
>> BROADCAST=192.168.130.255
>> NETMASK=255.255.255.0
>> BOOTPROTO=static
>> DEVICE=eth0
>> IPADDR=192.168.130.179
>> ONBOOT=yes
>> DOMAIN=public.net <http://public.net>
>> ZONE=public
>> IPV6INIT=no
>>
>>
>> ifcfg-eth1: (vmNet aka Internet)
>> ----------------
>> BROADCAST=192.168.1.255
>> NETMASK=255.255.255.0
>> BOOTPROTO=static
>> DEVICE=eth1
>> IPADDR=192.168.1.179
>> GATEWAY=192.168.1.254
>> ONBOOT=yes
>> DNS1=192.168.1.1
>> DNS2=192.168.1.2
>> DOMAIN=public.net <http://public.net>
>> ZONE=public
>> IPV6INIT=no
>>
>> _______________________________________________
>> Users mailing list
>> Users at ovirt.org <mailto:Users at ovirt.org>
>> http://lists.ovirt.org/mailman/listinfo/users
>> <http://lists.ovirt.org/mailman/listinfo/users>
>>
>>
>>
>>
>>
>> --
>>
>> SANDRO BONAZZOLA
>>
>> ASSOCIATE MANAGER, SOFTWARE ENGINEERING, EMEA ENG VIRTUALIZATION R&D
>>
>> Red Hat EMEA <https://www.redhat.com/>
>>
>> <https://red.ht/sig>
>> TRIED. TESTED. TRUSTED. <https://redhat.com/trusted>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ovirt.org/pipermail/users/attachments/20170410/5758893d/attachment-0001.html>
More information about the Users
mailing list