Rebooting KVM host looses it's guest.

ipsl · 04-01-2016, 09:35 AM

Running ubuntu 14.04 for the entire stack.

The host machine has 4 drives:
sda 64GB SSD (Main OS)
sdb 64GB SSD (dd copy of sda)
sdc 1.2 TB SSD (1st guest)
sdd 1.2 TB SSD (2nd guest)

When the host machine reboots I will sometimes loose the connection to the guest VM

Code:

tsadmin@lw-kvm-00:~$ virsh -c qemu:///system list
 Id    Name                           State
----------------------------------------------------

tsadmin@lw-kvm-00:~$

but the XML files are there

Code:

tsadmin@lw-kvm-00:~$ ls -l /etc/libvirt/qemu/
total 16
drwxr-xr-x 2 root root 4096 Mar 31 13:38 autostart
-rw------- 1 root root 2760 Apr  1 07:10 lkr-test-elastic00.xml
-rw------- 1 root root 2760 Mar 31 12:59 lkr-test-elastic01.xml
drwxr-xr-x 3 root root 4096 Mar 31 11:58 networks

if I try to redefine using the xml

Code:

tsadmin@lw-kvm-00:~$ sudo virsh define /etc/libvirt/qemu/lkr-test-elastic00.xml
error: Failed to define domain from /etc/libvirt/qemu/lkr-test-elastic00.xml
error: unknown OS type hvm

tsadmin@lw-kvm-00:~$

The xml file

Code:

<!--
WARNING: THIS IS AN AUTO-GENERATED FILE. CHANGES TO IT ARE LIKELY TO BE
OVERWRITTEN AND LOST. Changes to this xml configuration should be made using:
  virsh edit lkr-test-elastic00
or other application using the libvirt API.
-->

<domain type='kvm'>
  <name>lkr-test-elastic00</name>
  <uuid>e6da59a7-708b-0bf1-5642-bf43200b6011</uuid>
  <memory unit='KiB'>32768000</memory>
  <currentMemory unit='KiB'>32768000</currentMemory>
  <vcpu placement='static'>8</vcpu>
  <os>
    <type arch='x86_64' machine='pc-i440fx-trusty'>hvm</type>
    <boot dev='hd'/>
  </os>
  <features>
    <acpi/>
    <apic/>
    <pae/>
  </features>
  <clock offset='utc'/>
  <on_poweroff>destroy</on_poweroff>
  <on_reboot>restart</on_reboot>
  <on_crash>restart</on_crash>
  <devices>
    <emulator>/usr/bin/kvm-spice</emulator>
    <disk type='block' device='disk'>
      <driver name='qemu' type='raw'/>
      <source dev='/dev/sdc1'/>
      <target dev='vda' bus='virtio'/>
      <address type='pci' domain='0x0000' bus='0x00' slot='0x06' function='0x0'/>
    </disk>
    <disk type='block' device='cdrom'>
      <driver name='qemu' type='raw'/>
      <target dev='hdc' bus='ide'/>
      <readonly/>
      <address type='drive' controller='0' bus='1' target='0' unit='0'/>
    </disk>
    <controller type='usb' index='0'>
      <address type='pci' domain='0x0000' bus='0x00' slot='0x01' function='0x2'/>
    </controller>
    <controller type='pci' index='0' model='pci-root'/>
    <controller type='ide' index='0'>
      <address type='pci' domain='0x0000' bus='0x00' slot='0x01' function='0x1'/>
    </controller>
    <interface type='bridge'>
      <mac address='52:54:00:e9:1d:79'/>
      <source bridge='br0'/>
      <model type='virtio'/>
      <address type='pci' domain='0x0000' bus='0x00' slot='0x03' function='0x0'/>
    </interface>
    <interface type='bridge'>
      <mac address='52:54:00:de:ac:69'/>
      <source bridge='br1'/>
      <model type='virtio'/>
      <address type='pci' domain='0x0000' bus='0x00' slot='0x04' function='0x0'/>
    </interface>
    <serial type='pty'>
      <target port='0'/>
    </serial>
    <console type='pty'>
      <target type='serial' port='0'/>
    </console>
    <input type='mouse' bus='ps2'/>
    <input type='keyboard' bus='ps2'/>
    <graphics type='vnc' port='-1' autoport='yes'/>
    <sound model='ich6'>
      <address type='pci' domain='0x0000' bus='0x00' slot='0x05' function='0x0'/>
    </sound>
    <video>
      <model type='cirrus' vram='9216' heads='1'/>
      <address type='pci' domain='0x0000' bus='0x00' slot='0x02' function='0x0'/>
    </video>
    <memballoon model='virtio'>
      <address type='pci' domain='0x0000' bus='0x00' slot='0x07' function='0x0'/>
    </memballoon>
  </devices>
</domain>

Each guest is an entire 1.2TB drive. About the only thing I can think of is doing the dd to clone the host hard drive, but how or why is beyond me.

dyasny · 04-03-2016, 06:14 AM

virsh list will only show the running VMs. On reboot, is a VM is not set to autostart, it will stay down until you start it. virsh list --all, will give you all the defined VMs, so you can start the ones that are down. or you can do virsh autostart VMNAME to set it to autostart on host boot

ipsl · 04-04-2016, 08:47 AM

Quote:

tsadmin@lw-kvm-00:~$ virsh list --all
Id Name State
----------------------------------------------------

The system doesn't see them, but the data in the harddrives is there and the xml files are also there and set in the auto-start directory.

Quote:

tsadmin@lw-kvm-00:~$ ls -l /etc/libvirt/qemu/autostart/
total 0
lrwxrwxrwx 1 root root 40 Mar 31 13:38 lkr-test-elastic00.xml -> /etc/libvirt/qemu/lkr-test-elastic00.xml
lrwxrwxrwx 1 root root 40 Mar 31 13:38 lkr-test-elastic01.xml -> /etc/libvirt/qemu/lkr-test-elastic01.xml

jefro · 04-05-2016, 06:35 PM

How do you start them then? Or are you saying that all is not working?

It is possible that the filesystem is not behaving well on shutdowns. When you say reboot, do you have all clients properly shut down?

I might start with a live media and do a file system check.

ipsl · 04-06-2016, 07:41 AM

I could build the host, install and configure kvm. Create several guest via virt-manager, set the guest to autostart, and dd the host disk. This would work, then on a reboot, it wouldn't.

At this point my boss said to blow it out and throw ESXi on them so not only was it frustrating, it's now moot. :/

Slax-Dude · 04-06-2016, 10:16 AM

Looks like kvm kernel module might not be loading after reboot, for some reason.

But since you gave up (or were forced to give up), we will never know why.
Please mark this thread as "Solved".

dyasny · 04-07-2016, 09:28 AM

Nothing to do with the kernel module. If libvirt failed to locate the files after reboot, you probably had a permissions or SELinux misconfiguration. Never happened to me, but then again, I am smart enough not to use ubuntu for KVM

Slax-Dude · 04-08-2016, 02:45 AM

Quote:

Originally Posted by ipsl

if I try to redefine using the xml

Code:

tsadmin@lw-kvm-00:~$ sudo virsh define /etc/libvirt/qemu/lkr-test-elastic00.xml
error: Failed to define domain from /etc/libvirt/qemu/lkr-test-elastic00.xml
error: unknown OS type hvm

tsadmin@lw-kvm-00:~$

This looks like a missing kvm kerlel module.
Even if livbirt 'forgot' where the file is, here the OP is specifically telling it where to find it and to define a new VM with it...

But since the OP will not pursue further info we'll never know.