[precise] virbr0 not created during fresh install when dnsmasq is running with --except-interface=virbr0

I'm afraid I can't reproduce this. Can you say exactly how you installed libvirt-bin? Was it listed in the preseed? Installed with 'virtualization host' in tasksel?

Does /etc/libvirt/qemu/networks/autostart/ have default.xml? What does 'virsh net-list' show? And is there anything of interest in /var/log/libvirt/libvirtd.log?

@serge

I had installed it via "apt-get install vertinst". I was trying to get koan working.

However, since encountering this bug and some other issues, I reinstalled this server from scratch with Oneiric server. I followed the exact same process. I first installed ubuntu-orchestra-server. After verifying that was working fairly well I installed vertinst and qemu-kvm and koan.

At this point I ran into a different issue and was instructed to install the Precise version of the vertinst package which I did by downloading the .deb and using "dpkg -i" to install it.

I'm very motivated to get this resolved so if there is anything I can do to assist in debugging it, let me know.

So you installed the precise virtinst package on an uptodate oneiric system? That oneiric system already had libvirt fully installed?

I can try to reproduce based on that. Please let me know if I misunderstood.

@serge,

You have the correct. Andres told me to do that due to another bug that I ran into which had been fixed in the Precise version but was not backported/SRU'd for Oneiric.

@serge,

I'm attaching the default.xml file the libvirtd.log file.

Running "virsh net-list" returns:

> virsh net-list
Name State Autostart
-----------------------------------------

> sudo su -
[sudo] password for bradf:
root@red:~# virsh net-list
Name State Autostart
-----------------------------------------

root@red:~# virsh net-list --all
Name State Autostart
-----------------------------------------
default inactive yes

root@red:~# virsh net-autostart default
Network default marked as autostarted

root@red:~#

Quoting Brad Figg (<email address hidden>):
> > sudo su -
> [sudo] password for bradf:
> root@red:~# virsh net-list
> Name State Autostart
> -----------------------------------------
>
> root@red:~# virsh net-list --all
> Name State Autostart
> -----------------------------------------
> default inactive yes
>
> root@red:~# virsh net-autostart default
> Network default marked as autostarted

Sorry, I meant

 virsh net-start default

here. It was already marked autostart. We want to see if libvirt is
able to start it at this point.

From libvirtd.log:

14:14:46.066: 17275: error : virCommandWait:1991 : internal error Child process (dnsmasq --strict-order --bind-interfaces --pid-file=/var/run/libvirt/network/default.pid --conf-file= --except-interface lo --listen-address 192.168.122.1 --dhcp-range 192.168.122.2,192.168.122.254 --dhcp-leasefile=/var/lib/libvirt/dnsmasq/default.leases --dhcp-lease-max=253 --dhcp-no-override) status unexpected: exit status 2

The dnsmasq manpage says error 2 means :
       A problem with network access occurred (address in use, attempt to use privileged ports without permission).

(I tried to reproduce this on a canonistack instance, by installing
libvirt-bin and then installing
 http://archive.ubuntu.com/ubuntu/pool/main/v/virtinst/virtinst_0.600.0-1ubuntu4_all.deb
with dpkg -i, then doing 'apt-get install -f', but after reboot virbr0 did come up)

Is it possible to do 'apport-collect 918807' from that machine?

Otherwise, please post the output of 'ps -ef', 'dpkg -l | grep dnsmasq', 'dpkg -l libvirt-bin', 'ps -ef | grep libvirt', 'ps -ef | grep dnmasq', 'ifconfig -a', and the files /etc/network/interfaces and /etc/libvirt/libvirtd.conf.

Thanks.

> apport-collect 918807
Package libvirt not installed and no hook available, ignoring

*** Updating problem report

No additional information collected.

Press any key to continue...

No pending crash reports. Try --help for more information.

[~]
> dpkg -l | grep dnsmasq
ii dnsmasq 2.57-1ubuntu1 A small caching DNS proxy and DHCP/TFTP server
ii dnsmasq-base 2.57-1ubuntu1 A small caching DNS proxy and DHCP/TFTP server

[~]
> dpkg -l libvirt-bin
Desired=Unknown/Install/Remove/Purge/Hold
| Status=Not/Inst/Conf-files/Unpacked/halF-conf/Half-inst/trig-aWait/Trig-pend
|/ Err?=(none)/Reinst-required (Status,Err: uppercase=bad)
||/ Name Version Description
+++-======================-=========================-====================================
ii libvirt-bin 0.9.2-4ubuntu15.1 the programs for the libvirt library

[~]
> sudo ps -ef | grep libvirt
root 1017 1 0 Jan25 ? 00:00:00 /usr/sbin/libvirtd -d
bradf 2508 2137 0 10:28 pts/0 00:00:00 grep --color=auto libvirt

[~]
> sudo ps -ef | grep dnsmask
bradf 2511 2137 0 10:28 pts/0 00:00:00 grep --color=auto dnsmask

So if you do

   virsh net-start default

does that succeed and bring up virbr0?

What does 'ifconfig -a' show?

[~]
> sudo virsh net-start default
error: Failed to start network default
error: internal error Child process (dnsmasq --strict-order --bind-interfaces --pid-file=/var/run/libvirt/network/default.pid --conf-file= --except-interface lo --listen-address 192.168.122.1 --dhcp-range 192.168.122.2,192.168.122.254 --dhcp-leasefile=/var/lib/libvirt/dnsmasq/default.leases --dhcp-lease-max=253 --dhcp-no-override) status unexpected: exit status 2

[~]
> ifconfig -a
eth0 Link encap:Ethernet HWaddr 00:23:54:9e:5e:c8
          inet addr:192.168.1.6 Bcast:192.168.1.255 Mask:255.255.255.0
          inet6 addr: fe80::223:54ff:fe9e:5ec8/64 Scope:Link
          UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
          RX packets:7266810 errors:0 dropped:0 overruns:0 frame:0
          TX packets:3971503 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:1000
          RX bytes:3480412144 (3.4 GB) TX bytes:6937280997 (6.9 GB)
          Interrupt:18

eth1 Link encap:Ethernet HWaddr 00:23:54:9e:5e:c7
          BROADCAST MULTICAST MTU:1500 Metric:1
          RX packets:0 errors:0 dropped:0 overruns:0 frame:0
          TX packets:0 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:1000
          RX bytes:0 (0.0 B) TX bytes:0 (0.0 B)
          Interrupt:17

lo Link encap:Local Loopback
          inet addr:127.0.0.1 Mask:255.0.0.0
          inet6 addr: ::1/128 Scope:Host
          UP LOOPBACK RUNNING MTU:16436 Metric:1
          RX packets:252185 errors:0 dropped:0 overruns:0 frame:0
          TX packets:252185 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:0
          RX bytes:30386239 (30.3 MB) TX bytes:30386239 (30.3 MB)

Thanks. So it's not a race with networking setup - dnsmasq simply refuses
to run. Odd!

I'll try a bit more to reproduce with a vm with two nics and with an
interfaces file like yours.

Still could not reproduce.

Looking at ps -ef output,

/usr/sbin/dnsmasq -x /var/run/dnsmasq/dnsmasq.pid -u dnsmasq -7 /etc/dnsmasq.d,.dpkg-dist,.dpkg-old,.dpkg-new

I'm convinced that's the problem. Is there a reason this dnsmasq is running, and not binding just certain interfaces?

If you can change this dnsmasq to run with "--except-interface=virbr0", that should suffice.

There you go. 'sudo apt-get install dnsmasq' lets me reproduce.

Setting 'except-interface=virbr0' in /etc/default/dnsmasq *should* fix it. But it doesn't. I'm not sure whether we should consider it a bug in libvirt that it is trying to grab more bridges than it should.

You should be able to work around this at any rate by doing "apt-get purge dnsmasq"

Following text in libvirt source explains why libvirt does it:

    /*
     * --interface does not actually work with dnsmasq < 2.47,
     * due to DAD for ipv6 addresses on the interface.
     *
     * virCommandAddArgList(cmd, "--interface", ipdef->bridge, NULL);
     *
     * So listen on all defined IPv[46] addresses
     */

I'm changing the priority to low because there is a workaround. However, it's possible that the workaround is not always acceptable, in which case priority here should be raised.

For precise, we coudl actually fix this (at the cost of a delta from upstream and debian), since precise has dnsmasq 2.59.

Indeed, purging dnsmasq fixed this problem.

I think we can mark this a dup of bug 928524 ?