3:41 p.m.
Hi!
I am trying to use virt-customize to add packages to an image. However
it seems that there is a problem where a resolv.conf is not being
installed and thus package installation fails.
The command is:
virt-customize -v -x -a trusty-server-cloudimg-amd64-disk1.img
--run-command "host openstack.org"
I am having a hard time gathering complete output as the command seems
to corrupt my terminal somehow but I think the relevant lines may be:
[ 3.0] Running: host openstack.org
running command:
exec >>'/tmp/builder.log' 2>&1
host openstack.org
libguestfs: trace: sh "exec >>'/tmp/builder.log' 2>&1\n\n\nhost
openstack.org\n"
guestfsd: main_loop: new request, len 0x60
mount --bind /dev /sysroot/dev
mount --bind /dev/pts /sysroot/dev/pts
mount: mount point /sysroot/dev/pts does not exist
mount --bind /proc /sysroot/proc
mount --bind /sys/fs/selinux /sysroot/selinux
mount: mount point /sysroot/selinux does not exist
mount --bind /sys /sysroot/sys
mount --bind /sys/fs/selinux /sysroot/sys/fs/selinux
mount: mount point /sysroot/sys/fs/selinux does not exist
renaming /sysroot/etc/resolv.conf to /sysroot/etc/g1fbx1fs
cp /etc/resolv.conf /sysroot/etc/resolv.conf
cp: cannot stat '/etc/resolv.conf': No such file or directory
/bin/sh -c exec >>'/tmp/builder.log' 2>&1
host openstack.org
And the command fails due to the missing resolv.conf.
I am using version 1.28.6 build from source on ubuntu 14.04. The same
image file listed above works perfectly when launched as an openstack
instance.
Thanks,
Lee Revell
4:12 p.m.
On Tue, Mar 10, 2015 at 04:41:22PM -0400, Lee Revell wrote:
Hi!
I am trying to use virt-customize to add packages to an image.
However it seems that there is a problem where a resolv.conf is not
being installed and thus package installation fails.
The command is:
virt-customize -v -x -a trusty-server-cloudimg-amd64-disk1.img
--run-command "host openstack.org"
I am having a hard time gathering complete output as the command
seems to corrupt my terminal somehow but I think the relevant lines
may be:
[ 3.0] Running: host openstack.org
running command:
exec >>'/tmp/builder.log' 2>&1
host openstack.org
libguestfs: trace: sh "exec >>'/tmp/builder.log'
2>&1\n\n\nhost
openstack.org\n"
guestfsd: main_loop: new request, len 0x60
mount --bind /dev /sysroot/dev
mount --bind /dev/pts /sysroot/dev/pts
mount: mount point /sysroot/dev/pts does not exist
mount --bind /proc /sysroot/proc
mount --bind /sys/fs/selinux /sysroot/selinux
mount: mount point /sysroot/selinux does not exist
mount --bind /sys /sysroot/sys
mount --bind /sys/fs/selinux /sysroot/sys/fs/selinux
mount: mount point /sysroot/sys/fs/selinux does not exist
renaming /sysroot/etc/resolv.conf to /sysroot/etc/g1fbx1fs
cp /etc/resolv.conf /sysroot/etc/resolv.conf
cp: cannot stat '/etc/resolv.conf': No such file or directory
/bin/sh -c exec >>'/tmp/builder.log' 2>&1
host openstack.org
And the command fails due to the missing resolv.conf.
I am using version 1.28.6 build from source on ubuntu 14.04. The
same image file listed above works perfectly when launched as an
openstack instance.
I'm afraid our handling of /etc/resolv.conf inside the chroot is a
large (and continuing) source of problems. Currently this code is
run:
https://github.com/libguestfs/libguestfs/blob/master/daemon/command.c#L148
There are a couple of thing you could do:
- Don't rely too heavily on name resolution.
- Use --firstboot* scripts instead, since they genuinely run in guest
context, and should therefore work more like you expect.
(and I guess a third one .. suggest a better approach to resolv.conf
handling).
Just for comparison, libguestfs 1.29.29 on Fedora host, plus
virt-builder, with a Fedora guest does this:
----------------------------------------------------------------------
$ virt-builder fedora-21 --install /usr/bin/host --run-command "host
openstack.org"
[ 2.0] Downloading: http://libguestfs.org/download/builder/fedora-21.xz
[ 2.0] Planning how to build this image
[ 2.0] Uncompressing
[ 13.0] Opening the new disk
[ 24.0] Setting a random seed
[ 24.0] Installing packages: /usr/bin/host
[ 49.0] Running: host openstack.org
[ 50.0] Setting passwords
virt-builder: Setting random password of root to K0kkVIZXdtgmKtvw
[ 53.0] Finishing off
Output file: fedora-21.img
Output size: 4.0G
Output format: raw
Total usable space: 5.2G
Free space: 4.3G (83%)
$ guestfish --ro -a fedora-21.img -i
Welcome to guestfish, the guest filesystem shell for
editing virtual machine filesystems and disk images.
Type: 'help' for help on commands
'man' to read the manual
'quit' to quit the shell
Operating system: Fedora release 21 (Twenty One)
/dev/sda3 mounted on /
/dev/sda1 mounted on /boot
<fs> cat /tmp/builder.log
[lots of yum output deleted]
Installed:
bind-utils.x86_64 32:9.9.6-8.P1.fc21
Dependency Installed:
bind-libs.x86_64 32:9.9.6-8.P1.fc21
Dependency Updated:
bind-libs-lite.x86_64 32:9.9.6-8.P1.fc21
bind-license.noarch 32:9.9.6-8.P1.fc21
Complete!
openstack.org has address 162.242.140.107
openstack.org mail is handled by 20 mx2.emailsrvr.com.
openstack.org mail is handled by 10 mx1.emailsrvr.com.
----------------------------------------------------------------------
Pino ^^ any ideas?
Rich.
--
Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones
Read my programming and virtualization blog: http://rwmj.wordpress.com
virt-top is 'top' for virtual machines. Tiny program with many
powerful monitoring features, net stats, disk stats, logging, etc.
http://people.redhat.com/~rjones/virt-top
4:27 p.m.
On 03/10/2015 05:12 PM, Richard W.M. Jones wrote:
I'm afraid our handling of /etc/resolv.conf inside the chroot is
a
large (and continuing) source of problems. Currently this code is
run:
https://github.com/libguestfs/libguestfs/blob/master/daemon/command.c#L148
There are a couple of thing you could do:
- Don't rely too heavily on name resolution.
- Use --firstboot* scripts instead, since they genuinely run in guest
context, and should therefore work more like you expect.
Thanks, I'll try these. I actually tried several other options like:
--upload /home/leer/resolv.conf:/etc/resolv.conf
--write /etc/resolv.conf:"nameserver 172.16.12.1"
--run-command "echo 'nameserver 172.16.12.1' > /etc/resolv.conf"
--copy-in /home/leer/resolv.conf:/etc/resolv.conf
The first three all failed with some variant on "'/etc/resolv.conf': No
such file or directory". The fourth apparently has been removed from
this version or was not compiled in.
Also I believe what I thought was scrollback buffer corruption is
actually virt-customize making changes to the virtual as I noticed the
md5sum of the image changed after each failed attempt.
Thanks again,
Lee
8:11 a.m.
On Tue, Mar 10, 2015 at 09:12:12PM +0000, Richard W.M. Jones wrote:
On Tue, Mar 10, 2015 at 04:41:22PM -0400, Lee Revell wrote:
> Hi!
>
> I am trying to use virt-customize to add packages to an image.
> However it seems that there is a problem where a resolv.conf is not
> being installed and thus package installation fails.
>
> The command is:
>
> virt-customize -v -x -a trusty-server-cloudimg-amd64-disk1.img
> --run-command "host openstack.org"
>
> I am having a hard time gathering complete output as the command
> seems to corrupt my terminal somehow but I think the relevant lines
> may be:
>
> [ 3.0] Running: host openstack.org
> running command:
> exec >>'/tmp/builder.log' 2>&1
>
>
> host openstack.org
>
> libguestfs: trace: sh "exec >>'/tmp/builder.log'
2>&1\n\n\nhost
> openstack.org\n"
> guestfsd: main_loop: new request, len 0x60
> mount --bind /dev /sysroot/dev
> mount --bind /dev/pts /sysroot/dev/pts
> mount: mount point /sysroot/dev/pts does not exist
> mount --bind /proc /sysroot/proc
> mount --bind /sys/fs/selinux /sysroot/selinux
> mount: mount point /sysroot/selinux does not exist
> mount --bind /sys /sysroot/sys
> mount --bind /sys/fs/selinux /sysroot/sys/fs/selinux
> mount: mount point /sysroot/sys/fs/selinux does not exist
> renaming /sysroot/etc/resolv.conf to /sysroot/etc/g1fbx1fs
> cp /etc/resolv.conf /sysroot/etc/resolv.conf
> cp: cannot stat '/etc/resolv.conf': No such file or directory
> /bin/sh -c exec >>'/tmp/builder.log' 2>&1
>
>
> host openstack.org
>
> And the command fails due to the missing resolv.conf.
>
> I am using version 1.28.6 build from source on ubuntu 14.04. The
> same image file listed above works perfectly when launched as an
> openstack instance.
I'm afraid our handling of /etc/resolv.conf inside the chroot is a
large (and continuing) source of problems. Currently this code is
run:
https://github.com/libguestfs/libguestfs/blob/master/daemon/command.c#L148
There are a couple of thing you could do:
- Don't rely too heavily on name resolution.
- Use --firstboot* scripts instead, since they genuinely run in guest
context, and should therefore work more like you expect.
(and I guess a third one .. suggest a better approach to resolv.conf
handling).
We can finally use overlayfs now it is merged
# cat /etc/resolv.conf
nameserver 192.168.1.254
# mkdir /mnt/{overlay,empty,custometc}
# mount -t overlay overlay -o lowerdir=/etc,upperdir=/mnt/overlay,workdir=/mnt/empty
/mnt/custometc
# echo "nameserver 192.168.1.1" > /mnt/custometc/resolv.conf
# cat /mnt/custom/etcresolv.conf
nameserver 192.168.1.1
# cat /etc/resolv.conf
nameserver 192.168.1.254
Regards,
Daniel
--
|: http://berrange.com -o- http://www.flickr.com/photos/dberrange/ :|
|: http://libvirt.org -o- http://virt-manager.org :|
|: http://autobuild.org -o- http://search.cpan.org/~danberr/ :|
|: http://entangle-photo.org -o- http://live.gnome.org/gtk-vnc :|
8:29 a.m.
On 03/10/2015 05:12 PM, Richard W.M. Jones wrote:
Just for comparison, libguestfs 1.29.29 on Fedora host, plus
virt-builder, with a Fedora guest does this:
Hmm, something weird is going in with the output of virt-customize.
leer@eng-leer:~$ virt-customize -a
trusty-server-cloudimg-amd64-disk1.img --firstboot update-resolv-conf.sh
--install mysql-server -v -x > virt-customize.log 2>&1
leer@eng-leer:~$ wc -l virt-customize.log
125 virt-customize.log
leer@eng-leer:~$ head virt-customize.log
Err http://security.ubuntu.com trusty-security InRelease
Err http://archive.ubuntu.com trusty InRelease
Err http://archive.ubuntu.com trusty-updates InRelease
Err http://security.ubuntu.com trusty-security Release.gpg
Could not resolve 'security.ubuntu.com'
Err http://archive.ubuntu.com trusty Release.gpg
Could not resolve 'archive.ubuntu.com'
Lee
12:03 p.m.
On Wed, Mar 11, 2015 at 09:29:29AM -0400, Lee Revell wrote:
On 03/10/2015 05:12 PM, Richard W.M. Jones wrote:
>Just for comparison, libguestfs 1.29.29 on Fedora host, plus
>virt-builder, with a Fedora guest does this:
Hmm, something weird is going in with the output of virt-customize.
What actually happens is that we download the log file using the
guestfs_download API:
https://github.com/libguestfs/libguestfs/blob/master/customize/customize_...
The API opens [in this instance] /dev/stderr with O_TRUNC:
https://github.com/libguestfs/libguestfs/blob/master/src/proto.c#L752
So if you redirect stderr to a file, then the file gets truncated.
A quick fix is to use `tee' (or another program) in the redirection,
since a pipe ignores truncations.
The real fix is to modify the guestfs_download API to allow the
truncation behaviour to be optional.
Rich.
--
Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones
Read my programming and virtualization blog: http://rwmj.wordpress.com
virt-p2v converts physical machines to virtual machines. Boot with a
live CD or over the network (PXE) and turn machines into KVM guests.
http://libguestfs.org/virt-v2v
2:09 p.m.
On 03/10/2015 05:12 PM, Richard W.M. Jones wrote:
Just for comparison, libguestfs 1.29.29 on Fedora host, plus
virt-builder, with a Fedora guest does this:
libguestfs 1.29.29 on a ubuntu 14.04 host with fedora-21 guest seems
have no functioning network, it cannot ping the host. Is there some
configuration required on the host?
leer@eng-leer:~/libguestfs-1.29.29$ virt-builder fedora-21 --run-command
"ping -c 4 172.16.12.200"
[ 1.0] Downloading: http://libguestfs.org/download/builder/fedora-21.xz
[ 1.0] Planning how to build this image
[ 1.0] Uncompressing
[ 28.0] Opening the new disk
[ 31.0] Setting a random seed
[ 31.0] Running: ping -c 4 172.16.12.200
connect: Network is unreachable
virt-builder: error: ping -c 4 172.16.12.200: command exited with an error
Lee
3:35 a.m.
On Wed, Mar 11, 2015 at 03:09:56PM -0400, Lee Revell wrote:
On 03/10/2015 05:12 PM, Richard W.M. Jones wrote:
>Just for comparison, libguestfs 1.29.29 on Fedora host, plus
>virt-builder, with a Fedora guest does this:
>
libguestfs 1.29.29 on a ubuntu 14.04 host with fedora-21 guest seems
have no functioning network, it cannot ping the host. Is there some
configuration required on the host?
leer@eng-leer:~/libguestfs-1.29.29$ virt-builder fedora-21
--run-command "ping -c 4 172.16.12.200"
[ 1.0] Downloading: http://libguestfs.org/download/builder/fedora-21.xz
[ 1.0] Planning how to build this image
[ 1.0] Uncompressing
[ 28.0] Opening the new disk
[ 31.0] Setting a random seed
[ 31.0] Running: ping -c 4 172.16.12.200
connect: Network is unreachable
virt-builder: error: ping -c 4 172.16.12.200: command exited with an error
Because of the non-root networking technology we use (SLIRP), 'ping'
doesn't work and there are other shortcomings. See:
http://libguestfs.org/virt-rescue.1.html#network
If you use the libvirt backend, it is able to set up a real network
without needing root (but using a setuid program called
'qemu-bridge-helper' supplied by qemu), but we don't have that for the
direct backend yet.
http://libguestfs.org/guestfs.3.html#backend
Rich.
--
Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones
Read my programming and virtualization blog: http://rwmj.wordpress.com
virt-builder quickly builds VMs from scratch
http://libguestfs.org/virt-builder.1.html
8:14 a.m.
On 03/12/2015 04:35 AM, Richard W.M. Jones wrote:
> leer@eng-leer:~/libguestfs-1.29.29$ virt-builder fedora-21
> --run-command "ping -c 4 172.16.12.200"
> [ 1.0] Downloading: http://libguestfs.org/download/builder/fedora-21.xz
> [ 1.0] Planning how to build this image
> [ 1.0] Uncompressing
> [ 28.0] Opening the new disk
> [ 31.0] Setting a random seed
> [ 31.0] Running: ping -c 4 172.16.12.200
> connect: Network is unreachable
> virt-builder: error: ping -c 4 172.16.12.200: command exited with an error
Because of the non-root networking technology we use (SLIRP), 'ping'
doesn't work and there are other shortcomings. See:
I think the problem actually goes deeper than name resolution. On a
Ubuntu 14.04 host the firstboot-command doesn't seem to be run at all
and it appears that the network never gets configured. Here is a
complete log. The same command, libguestfs + supermin version, and
images all work perfectly on a Fedora 21 host.
Command tested:
$ ./run virt-builder fedora-21 --firstboot-command "echo 'testing' >
/var/log/test" -v -x 2>&1 | tee virt-builder.log
This appears to succeed as the log shows, but then mounting the image
with guestfish:
$ ./run guestfish -a fedora-21.img
Welcome to guestfish, the guest filesystem shell for
editing virtual machine filesystems and disk images.
Type: 'help' for help on commands
'man' to read the manual
'quit' to quit the shell
<fs> run
<fs> mount /dev/sda3 /
<fs> ls /var/log/
README
anaconda
audit
journal
ppp
<fs> exit
Lee
8:40 a.m.
On Fri, Mar 13, 2015 at 09:14:35AM -0400, Lee Revell wrote:
On 03/12/2015 04:35 AM, Richard W.M. Jones wrote:
>>leer@eng-leer:~/libguestfs-1.29.29$ virt-builder fedora-21
>>--run-command "ping -c 4 172.16.12.200"
>>[ 1.0] Downloading: http://libguestfs.org/download/builder/fedora-21.xz
>>[ 1.0] Planning how to build this image
>>[ 1.0] Uncompressing
>>[ 28.0] Opening the new disk
>>[ 31.0] Setting a random seed
>>[ 31.0] Running: ping -c 4 172.16.12.200
>>connect: Network is unreachable
>>virt-builder: error: ping -c 4 172.16.12.200: command exited with an error
>
>Because of the non-root networking technology we use (SLIRP), 'ping'
>doesn't work and there are other shortcomings. See:
I think the problem actually goes deeper than name resolution. On a
Ubuntu 14.04 host the firstboot-command doesn't seem to be run at
all and it appears that the network never gets configured. Here is a
complete log. The same command, libguestfs + supermin version, and
images all work perfectly on a Fedora 21 host.
Command tested:
$ ./run virt-builder fedora-21 --firstboot-command "echo 'testing' >
/var/log/test" -v -x 2>&1 | tee virt-builder.log
This appears to succeed as the log shows, but then mounting the
image with guestfish:
Do you boot and then shut down the image before examining it?
Rich.
--
Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones
Read my programming and virtualization blog: http://rwmj.wordpress.com
virt-df lists disk usage of guests without needing to install any
software inside the virtual machine. Supports Linux and Windows.
http://people.redhat.com/~rjones/virt-df/
9:59 a.m.
On 03/13/2015 09:40 AM, Richard W.M. Jones wrote:
Do you boot and then shut down the image before examining it?
Sorry, thinko on my part. This did work after all once the image booted.
Here's a log that shows update not working with the Ubuntu 14.04 host.
Command tested:
$ ./run virt-builder fedora-21 --network --update -v -x 2>&1 | tee
virt-builder.log
Lee
10:13 a.m.
On Fri, Mar 13, 2015 at 10:59:15AM -0400, Lee Revell wrote:
On 03/13/2015 09:40 AM, Richard W.M. Jones wrote:
>
>Do you boot and then shut down the image before examining it?
Sorry, thinko on my part. This did work after all once the image booted.
Here's a log that shows update not working with the Ubuntu 14.04 host.
Command tested:
$ ./run virt-builder fedora-21 --network --update -v -x 2>&1 | tee
virt-builder.log
Yes, the log indicates name resolution not working. Please try the
diagnosis steps that I outlined in:
https://www.redhat.com/archives/libguestfs/2015-March/msg00075.html
Rich.
--
Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones
Read my programming and virtualization blog: http://rwmj.wordpress.com
virt-builder quickly builds VMs from scratch
http://libguestfs.org/virt-builder.1.html
10:28 a.m.
On 03/13/2015 11:13 AM, Richard W.M. Jones wrote:
Yes, the log indicates name resolution not working. Please try the
diagnosis steps that I outlined in:
https://www.redhat.com/archives/libguestfs/2015-March/msg00075.html
Sorry, I somehow missed the above message.
$ ./run guestfish get-backend
direct
Here are the outputs of the /dev/tcp test, first doing:
$ ./run guestfish -v -x --network -a /dev/null run : debug sh 'exec 3<>
/dev/tcp/google.com/80 ; echo -e "GET /\n" >&3 ; cat <&3'
And then by IP:
$ ./run guestfish -v -x --network -a /dev/null run : debug sh 'exec 3<>
/dev/tcp/173.194.121.5/80 ; echo -e "GET /\n" >&3 ; cat <&3'
Both fail...
Lee
11:21 a.m.
On Fri, Mar 13, 2015 at 11:28:56AM -0400, Lee Revell wrote:
On 03/13/2015 11:13 AM, Richard W.M. Jones wrote:
>
>Yes, the log indicates name resolution not working. Please try the
>diagnosis steps that I outlined in:
>
>https://www.redhat.com/archives/libguestfs/2015-March/msg00075.html
Sorry, I somehow missed the above message.
$ ./run guestfish get-backend
direct
Here are the outputs of the /dev/tcp test, first doing:
$ ./run guestfish -v -x --network -a /dev/null run : debug sh 'exec
3<> /dev/tcp/google.com/80 ; echo -e "GET /\n" >&3 ; cat
<&3'
And then by IP:
$ ./run guestfish -v -x --network -a /dev/null run : debug sh 'exec
3<> /dev/tcp/173.194.121.5/80 ; echo -e "GET /\n" >&3 ; cat
<&3'
Both fail...
So what's happening is that /etc/resolv.conf in the appliance never
gets created.
*If* /etc/resolv.conf had been created then we should see evidence
that this code had been run:
https://github.com/libguestfs/libguestfs/blob/master/daemon/command.c#L14...
(eg. the "renaming ... to ..." messages should appear).
Question is why does /etc/resolv.conf not get created?
On Fedora, it is created by dhclient running from appliance/init:
https://github.com/libguestfs/libguestfs/blob/master/appliance/init#L88-L94
On Debian/Ubuntu that is supposed to run dhclient if available or
dhcpcd if not.
Run `./run virt-rescue --network --scratch' on the Ubuntu host and see if:
- either dhclient or dhcpcd exists in the appliance
- they work when you run them, or if there is an error message
- if running one of them creates /etc/resolv.conf
I don't see any error in the log from either dhclient or dhcpcd, so I
have to assume that one of them ran, but perhaps there is some other
problem.
Also perhaps the packagelist (appliance/packagelist.in) doesn't list
the right package containing dhclient/dhcpcd.
Also maybe a different dhcp client is needed.
Rich.
--
Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones
Read my programming and virtualization blog: http://rwmj.wordpress.com
virt-builder quickly builds VMs from scratch
http://libguestfs.org/virt-builder.1.html
11:58 a.m.
On 03/13/2015 12:21 PM, Richard W.M. Jones wrote:
So what's happening is that /etc/resolv.conf in the appliance
never
gets created.
Is the /dev/tcp test by IP evidence that the problem goes deeper than
name resolution?
Run `./run virt-rescue --network --scratch' on the Ubuntu host
and see if:
- either dhclient or dhcpcd exists in the appliance
- they work when you run them, or if there is an error message
- if running one of them creates /etc/resolv.conf
<rescue> which dhclient
/sbin/dhclient
<rescue> /sbin/dhclient
<rescue> ls /etc/resolv.conf
ls: cannot access /etc/resolv.conf: No such file
or directory
<rescue> which dhcpcd
<rescue>
Would there be any point in testing latest code from git?
Lee
12:29 p.m.
On Fri, Mar 13, 2015 at 12:58:52PM -0400, Lee Revell wrote:
On 03/13/2015 12:21 PM, Richard W.M. Jones wrote:
>So what's happening is that /etc/resolv.conf in the appliance never
>gets created.
>
Is the /dev/tcp test by IP evidence that the problem goes deeper
than name resolution?
>Run `./run virt-rescue --network --scratch' on the Ubuntu host and see if:
>
> - either dhclient or dhcpcd exists in the appliance
>
> - they work when you run them, or if there is an error message
>
> - if running one of them creates /etc/resolv.conf
>
><rescue> which dhclient
/sbin/dhclient
><rescue> /sbin/dhclient
><rescue> ls /etc/resolv.conf
ls: cannot access /etc/resolv.conf: No such file or directory
I wonder why dhclient doesn't create /etc/resolv.conf .
><rescue> which dhcpcd
><rescue>
Would there be any point in testing latest code from git?
I think you were testing 1.29.29, which is only a few days old. There
haven't been any changes to the appliance.
Rich.
--
Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones
Read my programming and virtualization blog: http://rwmj.wordpress.com
Fedora Windows cross-compiler. Compile Windows programs, test, and
build Windows installers. Over 100 libraries supported.
http://fedoraproject.org/wiki/MinGW
3:59 p.m.
On 03/13/2015 01:29 PM, Richard W.M. Jones wrote:
I wonder why dhclient doesn't create /etc/resolv.conf .
FWIW, I've also found that on the fedora 21 host, supermin passes all 8
"make check" tests. On the Ubuntu 14.04 host, it fails 2 of 8. The log
is attached.
Additionally there are other weird differences, like /etc/passwd and
/etc/shadow and many other /etc files present in the working appliance
are missing on the non-working one.
Lee
7:13 a.m.
On Fri, Mar 13, 2015 at 04:59:32PM -0400, Lee Revell wrote:
On 03/13/2015 01:29 PM, Richard W.M. Jones wrote:
>
>I wonder why dhclient doesn't create /etc/resolv.conf .
>
FWIW, I've also found that on the fedora 21 host, supermin passes
all 8 "make check" tests. On the Ubuntu 14.04 host, it fails 2 of 8.
The log is attached.
The failures almost certainly have nothing to do with the problem at
hand. They seem to happen because you don't have 'augeas-tools'
installed which is required for the test suite.
Additionally there are other weird differences, like /etc/passwd and
/etc/shadow and many other /etc files present in the working
appliance are missing on the non-working one.
"working" == Fedora? I wouldn't pay too much attention to these
differences.
I went back and had a look at the guestfish-network-test*.log files
that you sent previously, and there is one major problem:
virtio_net.ko doesn't exist on your host (or it does exist, but
supermin isn't picking it up for the appliance, but this unlikely).
This explains why the network doesn't work and why dhclient doesn't
work too. I don't know why virtio-net isn't used -- perhaps it isn't
compiled into your host kernel.
Rich.
--
Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones
Read my programming and virtualization blog: http://rwmj.wordpress.com
virt-builder quickly builds VMs from scratch
http://libguestfs.org/virt-builder.1.html
8:14 a.m.
On 03/16/2015 08:13 AM, Richard W.M. Jones wrote:
I went back and had a look at the guestfish-network-test*.log files
that you sent previously, and there is one major problem:
virtio_net.ko doesn't exist on your host (or it does exist, but
supermin isn't picking it up for the appliance, but this unlikely).
This explains why the network doesn't work and why dhclient doesn't
work too. I don't know why virtio-net isn't used -- perhaps it isn't
compiled into your host kernel.
Ok, I'll look into this. What's strange is that I have a virtualized
Openstack environment on the same host using multiple virtual networks
and that all works perfectly.
Lee
3520
days inactive
3526
days old
19 comments
3 participants
participants (3)
-
Daniel P. Berrange -
Lee Revell -
Richard W.M. Jones