Host Network Not Starting

[finish06]

As stated in the title, the host network will not start. Upon using IPMI and being on the host machine, I have run the following commands:
1. ifconfig
Only results in network "lo"
2. service networking.service restart
Starts the network and the host can be pinged at this point, however if I reboot, I it will not reinitiate the network.
3. Upon logging into host as root, the following error shows: "-bash: ovs-vsctl: command not found"
...

I don't know what else to try...

Thanks!

 

[baggar11]

Can confirm. I noticed openvswitch-switch and openvswitch-common updated, during the update process, never got the prompt back via SSH. Had to IPMI to the host and noticed similar results. Wondering how I can get interfaces back up.

 

[baggar11]

I was able to run the following to get an IP back on the management lan.

ifup vmbr0 up
ifconfig vmbr0 192.168.x.x netmask 255.255.255.0
route add default gw 192.168.10.1 vmbr0

Although, vmbr1 doesn't seem to be appearing in my ifconfig list. Even though the bridge appears to be functioning still. I'll leave it in its current state until someone else chimes in on a proper fix. I'm a little worried the openvswitch updates will cause the network not to come up properly even after a reboot.

 

[finish06]

I am glad that worked for you! Unfortunately, for me it only puts out another error.
Specifically: "ifup: failed to open lockfile /run/network/.ifstate.lock: No such file or directory"

Thanks for coming back though and posting what worked!

 

[charminator]

confirmed, too. Yesterday i installed updates on a 4 node cluster. Every node does not start networking after the updates were applied.
The cluster config seems to be broken, too.
Corosync does not run on any node any more and each node claims there is no cluster.

EDIT: We do not use openvswitch

 

[DemomanCA]

I'm also experiencing this. Updated openVswitch this morning over SSH, which kicked me out once it was in the process of updating, and then I couldn't get back in. On reboot, no network interfaces were brought up, but a "/etc/init.d/networking restart" got me back online. I dropped the openvswitch config and switched back to linux bridge/bond and still had the same issue. No network on restart, but networking restart brings it back up.

The only thing of interest in syslog seems to be


May 13 08:32:16 hypervisor systemd[1]: Found dependency on networking.service/stop
May 13 08:32:16 hypervisor systemd[1]: Breaking ordering cycle by deleting job networking.service/stop
May 13 08:32:16 hypervisor systemd[1]: Job networking.service/stop deleted to break ordering cycle starting with sysinit.target/stop

It would appears something has broken the systemd ordering of starting the networking service?

 

[baggar11]

Well interesting....

I was hoping more people would come across this issue, I guess we're special.

Since losing networking to the openvswitch packge upgrades and getting the management ip restored in a hacky fashion, I figured I'd try downgrading openswitch to 2.3.2-3. This didn't appear to bring back my storage bridge on vmbr1 even though my vm's seemed to happily be utilizing it still.

So then I powered down all my vm's, upgrade the openvswitch packages again(which killed networking again) and rebooted. To my surprise, networking was restored on boot. I'm not sure what you guys are running for configs, but mine is fairly simple. Both hosts have 1x10g and 1x1g and bridged via openvswitch.

 

[DemomanCA]

I can confirm uninstalling openvswitch-common and openvswitch-switch and reinstalling seems to have fixed the problem

 

[hasz78]

Hi,
Informative as someone has a problem with this issue. Working properly "apt-get purge" instead of "remove".

 

[linkstat]

Hello.

I have this issue in only one (of four) nodes. I've using open-vswitch, and the same network config in all nodes. The software upgrades is the same level for all of them. But only one node start with the network down (and in consequence, with all pve-related.service failed). But if I start the network service with systemctl start networking.service the network goes up without any problem. Of course, then I have manually restart the pve services: systemctl start pvestatd.service and systemctl start pvedaemon.service for all working.

The question would be... is a secure soluction purge the open-vswitch related packages and reinstall them? Any special attention in the procedure (like rebooting in a determinate moment or something like that)? Or simple run aptitude purge packages and the aptitude install package(s) will do the work?
Feedback, please.
Thanks!

 

[finish06]

I am still experiencing this error... must manually restart networking.service upon reboot for OVS network to start up. Does anyone know anything about this bug?

Thanks!

 

[linkstat]

I can say, that uninstalling, and then reinstalling the openvswitch-switch package fix the problem for me.

What I do:

1. Backup the /etc/network/interfaces:
cp /etc/network/interfaces /etc/network/interfaces.Backup

2. Download the packages that you will uninstall:
cd /var/cache/apt/archives
aptitude download openvswitch-switch openvswitch-common libatomic1
(after that you uninstall the open-vswitch packages, you lost network connectivity, so, sure that you want to get the packages just downloaded, ready for install).

3. Purge the openvswitch packages:
aptitude purge openvswitch-switch
(that completely uninstall and remove the configfiles and the dependencies required by the package, automatically installed)

4. Install the openvswitch-switch packages:
aptitude install openvswitch-switch
(for some reason, if apt would need more packages, you must download them from another machine and copy in a usb stick and then put the downloaded file(s) in /var/cache/apt/archives)

5. Restore the config files:
cp /etc/network/interfaces.Backup /etc/network/interfaces

Now, you can reboot and test if that works.


Greetings!