Libguestfs - Jun 2014 - Re: Script to read systemd journal of a guest to find its IP address

Heya,

Thanks Rich for the hints on IRC about journal bindings for libguestfs.
The below script is what I ended up with. At least I learnt how to
iterate over the systemd journal entries.

    $ sudo python find-ip.py 
    192.162.122.118

Obviously, this only searches for the _first_ occurance of the IP
address and doesn't take into account the guest may have acquired a new
IP on a later reboot. (I guess I'd have to do some thing like 
$ journalctl --lines=500 | egrep -i 'dhclient.*bound').

-------------------------------------------------------------------
#!/usr/bin/python
#-----------------------------------------------------------------
# PURPOSE: To find what IP address a guest has acquired from DHCP
#-----------------------------------------------------------------

image = "/var/lib/libvirt/images/ostack-controller.qcow2"

root_filesystem = "/dev/fedora/root"

import os
import sys
import re
import guestfs

g = guestfs.GuestFS ()

g.add_drive (image, format='qcow2')

g.launch ()
#print g.list_partitions()

# Set trace
#g.set_trace (1)

g.mount_options ("ro", root_filesystem, "/")

# Open the journal
g.journal_open ("/var/log/journal")


# Loop over the journal to find a specific string
count = 0
prog = re.compile("dhclient.*bound")
while g.journal_next():
    count += 1
    output = g.journal_get()
    #print output
    flag = 0
    for i in output:
        if i['attrname']=="MESSAGE" and
i['attrval'].startswith("bound"):
            print i['attrval'].split()[2]
            flag = 1
            break
    if flag == 1:
        break

# Close the journal handle
g.journal_close ()

# Unmount all the file system(s)
g.umount_all ()

g.sync ()

sys.exit(0)
-------------------------------------------------------------------

-- 
/kashyap

On Sat, Jun 14, 2014 at 11:41:03AM +0530, Kashyap Chamarthy
wrote:
> g.add_drive (image, format='qcow2')
You must use:

  g.add_drive (image, readonly=1)

otherwise you risk disk corruption if the guest is live (eg. if the
mount command has to replay the journal).
> g.launch ()
> #print g.list_partitions()
> 
> # Set trace
> #g.set_trace (1)
> 
> g.mount_options ("ro", root_filesystem, "/")
What I'd do differently here is to use inspection to detect the root
(and other) filesystems.  You never know if /var is on a separate
partition ...

Here is the example code (see call to "g.inspect_os" and the code with
the comment "Mount up the disks like guestfish -i").

http://libguestfs.org/guestfs-python.3.html#example-2:-inspect-a-virtual-machine-disk-image
> g.sync ()
No need to sync if you've added the drive readonly.

Thanks,

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/

On Sat, Jun 14, 2014 at 10:37:12AM +0100, Richard W.M. Jones
wrote:
> On Sat, Jun 14, 2014 at 11:41:03AM +0530, Kashyap Chamarthy wrote:
> > g.add_drive (image, format='qcow2')
> 
> You must use:
> 
>   g.add_drive (image, readonly=1)
> 
> otherwise you risk disk corruption if the guest is live (eg. if the
> mount command has to replay the journal).
Ah. I did use 'readonly=1' while testing, but forgot to mention it while
writing this. Thanks for catching.
> > g.launch ()
> > #print g.list_partitions()
> > 
> > # Set trace
> > #g.set_trace (1)
> > 
> > g.mount_options ("ro", root_filesystem, "/")
> 
> What I'd do differently here is to use inspection to detect the root
> (and other) filesystems.  You never know if /var is on a separate
> partition ...
> 
> Here is the example code (see call to "g.inspect_os" and the code
with
> the comment "Mount up the disks like guestfish -i").
> 
>
http://libguestfs.org/guestfs-python.3.html#example-2:-inspect-a-virtual-machine-disk-image
> 
> > g.sync ()
> 
> No need to sync if you've added the drive readonly.
Noted.

Thanks for these useful review comments, I'll test with these changes.

-- 
/kashyap

I've pushed a new tool called "virt-log" upstream, so you can now
just do:

  virt-log -d Guest

or:

  virt-log -a disk.img

It will automatically fetch the journal or traditional syslog file as
appropriate, and we can add Windows support in future.

Unfortunately the journal_* API calls are rather slow because they
have to fetch each single line from the journal.  So there's some work
to be done to make that fast.

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