Creating KVMs kills host network

Submitted by nabab on Tue, 02/26/2019 - 10:02 Pro Licensee

Hi,

Whenever I create a new KVM the host machine crashes. I have tried this with lots of different configurations. I am pretty sure this is a network issue but I can't figure it out. But in the end I did everything with Cloudmin on fresh installs, and I think this should have worked, but it didnt.

I did the following steps:

  • installed Cloudmin on a brand new CentOS 7, lets call him PC1, with nothing else.
  • created another physical brand new CentOS 7, PC2
  • added it in Cloudmin in PC1 as physical machine connecting through SSH key.
  • installed Webmin on PC2 from the left menu in Cloudmin.
  • added PC2 as a KVM host
  • got offered in Cloudmin to install all the virtualization softwares and bridge, and it did successfully
  • added properly PC2 as KVM host
  • rebooted PC2 and refreshed status in Cloudmin, all OK
  • created a new KVM instance (with CentOS 7)

I get the following result:

Creating virtual system with KVM ..
.. creation started with IP x.x.x.x.
Waiting for creation to complete ...............
.. creation has completed successfully.

Fixing interface names in Grub configuration file ..
.. no Grub configuration file found!

Creating swap file of 2 GB ..
.. done

Removing missing disks from fstab file ..
.. done

Expanding filesystem to 64 GB ..
.. done

Mounting new instance's filesystem ..
.. mounted on /mnt/kvm-xxxx.xxx.xx.io

Setting root password ..
.. done

Updating configuration files with hostname and IP address ..
.. done

Updating network configuration files ..
.. done

Modifying /etc/fstab file ..
.. done

Allowing SSH logins by root ..
.. already allowed

Setting SSH server port ..
.. done

Un-mounting instance's filesystem ..
.. done

Adding DNS entry xxxx.xxx.xx.io. for IP address x.x.x.x ..
.. done

Starting up new KVM instance ..
.. failed to start : KVM instance was started OK, but the graphical console password could not be set : Timeout connecting

Fetching current status ..
.. status successfully retrieved (Parent host is down)

Re-fetching current status of host system x.x.x.x ..
.. status successfully retrieved (Down)

Enabling system at host boot time ..
.. done

When I create the KVM I give the IPv4 and the MAC address.

At the moment of the crash I found this in the log messages:

Feb 26 16:24:26 sd-xx kernel: L1TF CPU bug present and SMT on, data leak possible. See CVE-2018-3646 and https://www.kernel.org/doc/html/latest/admin-guide/l1tf.html for details.
Feb 26 16:24:26 sd-xx kvm: 1 guest now active
Feb 26 16:24:26 sd-xx kernel: tun: Universal TUN/TAP device driver, 1.6
Feb 26 16:24:26 sd-xx kernel: tun: (C) 1999-2004 Max Krasnyansky <maxk@qualcomm.com>
Feb 26 16:24:26 sd-xx kernel: br0: port 2(tap0) entered blocking state
Feb 26 16:24:26 sd-xx kernel: br0: port 2(tap0) entered disabled state
Feb 26 16:24:26 sd-xx kernel: device tap0 entered promiscuous mode
Feb 26 16:24:26 sd-xx kernel: br0: port 2(tap0) entered blocking state
Feb 26 16:24:26 sd-xx kernel: br0: port 2(tap0) entered forwarding state
Feb 26 16:24:32 sd-xx kernel: kvm [11371]: vcpu0 disabled perfctr wrmsr: 0xc1 data 0xffff
Status: 
Active

Comments

Submitted by nabab on Tue, 02/26/2019 - 10:03 Pro Licensee

Title: Creating KVMs kills host config » Creating KVMs kills host network

Submitted by JamieCameron on Thu, 02/28/2019 - 23:15

When you say the host crashes, do you mean it completely reboots, or is it just inaccessible on the network?

Submitted by nabab on Tue, 05/05/2020 - 12:16 Pro Licensee

Hello there,

I am sorry I hadn't replied back then but I got really annoyed with these problems and I just stopped creating VMs with Cloudmin in the past year.

However I took a new server and decided to give it a new try, and I get the exact same problem than I did last year. When I said the host crashes it means it needs a physical reboot to be reachable again.

The /var/log/messages ends like this (just before the crash):

May  5 18:46:43 sd-71876 kernel: L1TF CPU bug present and SMT on, data leak possible. See CVE-2018-3646 and https://www.kernel.org/doc/html/latest/admin-guide/hw-vuln/l1tf.html for details.
May  5 18:46:43 sd-71876 kvm: 1 guest now active
May  5 18:46:43 sd-71876 kernel: tun: Universal TUN/TAP device driver, 1.6
May  5 18:46:43 sd-71876 kernel: tun: (C) 1999-2004 Max Krasnyansky <maxk@qualcomm.com>
May  5 18:46:43 sd-71876 NetworkManager[725]: <info>  [1588697203.7503] manager: (tap0): new Tun device (/org/freedesktop/NetworkManager/Devices/5)
May  5 18:46:43 sd-71876 kernel: br0: port 2(tap0) entered blocking state
May  5 18:46:43 sd-71876 kernel: br0: port 2(tap0) entered disabled state
May  5 18:46:43 sd-71876 kernel: device tap0 entered promiscuous mode
May  5 18:46:43 sd-71876 NetworkManager[725]: <info>  [1588697203.7552] device (tap0): state change: unmanaged -> unavailable (reason 'connection-assumed', sys-iface-state: 'external')
May  5 18:46:43 sd-71876 kernel: br0: port 2(tap0) entered blocking state
May  5 18:46:43 sd-71876 kernel: br0: port 2(tap0) entered forwarding state
May  5 18:46:43 sd-71876 dbus[632]: [system] Activating via systemd: service name='org.freedesktop.nm_dispatcher' unit='dbus-org.freedesktop.nm-dispatcher.service'
May  5 18:46:43 sd-71876 NetworkManager[725]: <info>  [1588697203.7563] device (tap0): state change: unavailable -> disconnected (reason 'none', sys-iface-state: 'external')
May  5 18:46:43 sd-71876 systemd: Starting Network Manager Script Dispatcher Service...
May  5 18:46:43 sd-71876 dbus[632]: [system] Successfully activated service 'org.freedesktop.nm_dispatcher'
May  5 18:46:43 sd-71876 nm-dispatcher: req:1 'connectivity-change': new request (3 scripts)
May  5 18:46:43 sd-71876 nm-dispatcher: req:1 'connectivity-change': start running ordered scripts...
May  5 18:46:43 sd-71876 systemd: Started Network Manager Script Dispatcher Service.
May  5 18:46:49 sd-71876 kernel: kvm [5084]: vcpu0 disabled perfctr wrmsr: 0xc1 data 0xffff