Recently I had to setup Enfold Server on Windows XP and as I run Kubuntu on my notebook, the virtual environment I've chosen to run Windows under is QEMU. I've been running QEMU for some time now and it provides all the features I need in a virtual machine, especially with kqemu installed. When I was in the process of setting up Enfold Server, though, I accidentally loaded the same QEMU image twice and the 2nd instance started running a CHKDSK on the drive and writing to it. After closing the 2nd instance down and restarting the already running Windows XP QEMU instance, Windows XP would get as far as the loading Windows XP screen and then cyclicly boot. I then attempted to mount the QEMU image (as it was in the RAW format) but when I did, I got the following error message: "primary boot sector is invalid". This was when alarm bells really started to ring as I'd put quite a lot of time and effort into this Windows XP QEMU image (including setting up all the Internet Explorer browsers and Safari for Windows), and unfortunately due to a file size error I was getting when attempting to backup the QEMU image, I didn't have a backup! After fruitlessly googling around for a resolution, I decided I'd attempt to boot off the Windows XP cdrom and attempt to recover the image that way. The following are the steps I took to restore my QEMU Windows XP Environment:

1. Booted off the Windows XP cdrom by typing the following at the command prompt:
   qemu winxp.img -cdrom /dev/cdrom -boot d
2. When the "Welcome to Setup" screen appeared, pressed R to start the Recovery Console.
3. Executed the FIXMBR command from the Recovery Console.
4. Ran a CHKDSK to fixup any other problems that may have existed.

I had the honour of giving the first Plone speech at the Academia Sinica in Taipei, Taiwan recently. TsungWei Hu, who arranged for me to give the speech, manages the Open Source Software Foundry (OSSF) project, whose key objectives are to promote Open Source Software in Taiwan. TsungWei mentioned that he had been meaning to ask Paul Everitt to give a speech regarding Plone at the Academia Sinica (Paul - hint, hint) but hadn't, so I got in first ;-)

This speech was the last in a series of Plone presentations I have been giving. The first of which was in Tokyo, Japan, and the next was for the Taipei Open Source Software User Group (TOSSUG). At the TOSSUG meetup I was able to meet Sean Moss-Pultz and Harald Welte of the OpenMoko project. Sean is the General/Technical Manager for the OpenMoko project, while Harald is the Lead System Level Architect. As I'm an Open Source fan and enjoy hacking my own computer, I'm really excited about what Sean, Harald and the rest of the OpenMoko Core Team are doing as mobile phones in general are pretty much closed devices nowadays.

N.B. The examples assume you are using a Linux-based operating system.

1. Do a subversion checkout of the instancemanager product into an appropriate location, by entering the following:
   

   svn co http://svn.plone.org/svn/collective/instancemanager/trunk/ instancemanager

2. Install instancemanager by entering the following command from within the instancemanager folder (this will ensure instancemanager runs from the checked out directory and can be easily updated):
   

   sudo python setup.py develop

3. The first time you run instancemanager (just type instancemanager and press Enter), it creates an .instancemanager directory in your home directory. There you'll find a userdefaults.py that you can customise to your local needs.
   The most relevant configuration directives you may want to change are the following:
   

   zope_instance_template (where your zope instances will reside)
   zope_location_template (the location where your zope server software resides - N.B. this can use python string substitution, e.g. /opt/zope/zope%(zope_version)s)
   user (default admin username)
   password (default admin password)

4. Following that, create a file called PROJECTNAME.py (can be empty) in the .instancemanager directory, customising that with project-specific settings (just copy the few items that you want to change from userdefaults.py.
   

   archive_basedir_template (the directory where all your zope/plone compressed product files are stored)
   symlink_basedir_template (the directory where you store your svn product checkouts)
   archivebundle_basedir_template (the directory where you store your zope/plone compressed product bundles - usually the same as archive_basedir_template)
   symlinkbundle_basedir_template (the directory where you store your svn bundle checkouts - usually the same as symlink_basedir_template)
   python (the path to your zope instance's python binary)
   zope_version (the version of the zope instance that hosts your Plone site)

5. Instancemanager expects to find the python/zope server software in a structured folder hierarchy (e.g. /opt/zope/zope2.9.6 and /opt/python/python2.4.4). I've found the easiest way to set this up is to use the Universal Installers listed on plone.org and modify the configuration directives in the install.sh bash script to point to those locations (for example, instead of pointing to /opt/Plone-2.5/ have them point to /opt/zope/zope2.9.6 and /opt/python/python2.4.4 respectively).
6. Copy the Data.fs with your Plone site's test data into the
   zope_instance_template/datafs folder, ensuring it's renamed to PROJECTNAME.fs.

   cp Data.fs ~/instances/datafs/PROJECTNAME.fs

7. Create the zope instance.
   

   instancemanager fresh PROJECTNAME

8. All going well, you should now have a new zope instance created for your Plone site accessible via the standard methods, i.e. http://localhost:8080/manage
9. Some further documentation regarding instancemanager can be found here.
   Some of the commands I often use are:
   

   instancemanager fresh PROJECTNAME (creates a new zope instance, including setting up all the Plone products as specified in the project file and also creates a new Plone site)
   instancemanager --products PROJECTNAME (just recreates the Products directory and all contained products)
   instancemanager soft PROJECTNAME (stops and starts the zope instance and reinstalls the products specified in the main_products configuration directive in the project file)
   instancemanager test PROJECTNAME (runs the tests for the main Plone product specified in the project file)
   instancemanager --test MyPloneProduct PROJECTNAME (runs the tests for the specified Plone product, i.e. MyPloneProduct)
   

Recently I had to setup Xen 3 for a client, and the following are the steps I undertook to install it (N.B. As a guide for the installation, I followed the instructions listed here).

The most pertinent hardware specifications for the server we installed Xen 3 on are as follows:

• 2x Intel Xeon 1.6 Ghz CPUs
• 4GB DDR2-667 RAM
• 3ware 9550SX-4LP PCI-X-to-SATA II RAID controller
• 4x Seagate Barracuda ES SATA 3.0GB/s 250GB hard drives (configured as RAID 5)
  

1. Firstly, I downloaded and burned the Debian testing (etch) netinst CD image to CD.
2. Following this, I began installing Debian etch on the server using the netinst CD.
   At the disk partitioning screen, I setup the hard drive as follows:
   Disk /dev/sda: 749.9 GB, 749966721024 bytes

255 heads, 63 sectors/track, 91178 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

Device Boot Start End Blocks Id System
/dev/sda1 * 1 365 2931831 83 Linux
/dev/sda2 366 1581 9767520 8e Linux LVM
/dev/sda3 1582 4013 19535040 82 Linux swap / Solaris
/dev/sda4 4014 91178 700152862+ 8e Linux LVM

   N.B. The /dev/sda4 LVM partition is where the logical volumes for the Xen guest domains (virtual servers) are to be stored.
   I left the rest of the installation steps at their setup defaults.
   
3. I then installed Xen 3 on the server by entering the following command:
   apt-get install xen-linux-system-2.6.18-3-xen-686 xen-tools \\
xen-docs-3.0 libc6-xen file xen-hypervisor-3.0.3-1-i386-pae \\
xen-linux-system-2.6.18-3-xen-vserver-686 bridge-utils
4. Following this, I made the following change in the /etc/xen/xend-config.sxp file and rebooted the server:
   replace
   (network-script network-dummy)
   with
   (network-script 'network-bridge netdev=eth0')
5. I now made the following changes to the /etc/xen-tools/xen-tools.conf file:
lvm = vservers
debootstrap = 1
size = 10Gb
memory = 512Mb
swap = 1Gb
fs = reiserfs
dist = etch
image = sparse
kernel = /boot/vmlinuz-2.6.18-3-xen-vserver-686
initrd = /boot/initrd.img-2.6.18-3-xen-vserver-686
mirror = http://debian.orcon.net.nz/debian
   
6. Following this, I created a new guest domain (or virtual server) on the server by entering the following command:
   xen-create-image --hostname myservername.mydomain.co.nz --dhcp
7. I then booted the guest domain by entering the following:
   xm create /etc/xen/myservername.mydomain.co.nz.cfg
   
8. I now connected a virtual console to my newly created guest domain by entering the following and then logged on as root:
   xm console myservername.mydomain.co.nz
9. Finally, I set the root password and then logged off the
   virtual console by typing CTRL+] and then made a ssh connection to the guest domain and started setting up the services on it.
   

This may be my last post as a sysadmin as like some other people, I've discovered that being a sysadmin has its own challenges, most especially the need to be constantly available, as computer systems can fall over at any time. Something, though, that proved invaluable in my previous role was some Systems Monitoring software called Zenoss (another attraction was that it runs on Zope!).
After ensuring I had the prerequisite software, the installation wasn't difficult. It was a bit of a nuisance that Zenoss creates its own Zope instance but the specific python products it installed within the Zope instance really required this for ease of setup.
Once the Zenoss server software was installed, I then had to configure the network devices to notify Zenoss via SNMP of their current status. Though the organisation where I worked had a few varied operating systems and devices, this wasn't a problem. SNMP Informant (with the MBM module for CPU, case, and component monitoring), Syslog, and Zenwin provided monitoring for the Windows 2003 server (there were some gotchas with the setup of the SNMP service but a big thanks to #zenoss as they helped me resolve the particular issues I had). The snmp package took care of my Linux boxes and Zenoss' builtin Cisco support provided monitoring for the Cisco router.
Once I had added all the network devices, I then setup Zenoss to send notifications via email whenever specified Zenoss events occured.
When Zenoss is notified of an event it categorises the event by such things as type of event, severity, etc. When Zenoss is unsure of how to categorise an event, you can categorise it manually. Zenoss can use this criteria for its event notification system.
Something that really impressed me about the Zenoss project was that when I subscribed to the mailing list, I requested a Zenoss t-shirt and though I live at the bottom of the world in New Zealand they sent me one!