dawgiestyle

Xen on Debian Etch

Wed, 18 Apr 2007 01:25:59 -0230

Well, it looks like I may have jumped the gun with that Xen on Debian Sarge post. I guess I didn't realize that "Etch" had already gone stable. But hey...this is good news!
Now, with Etch, we have all the xen goodies we need right in the Debian repositories, and apt-getable (or aptitudeable, if you prefer). This will certainly make for a much more easily maintainable system, and it proved to be much easier to set up as well.

Most of what I did came from one of the howto's that were previously referenced in Xen on Debian Sarge, namely, The Perfect Xen 3.0.3 Setup For Debian Sarge, with a little from Xen from Backports on Debian Sarge. Just for the record, my system has a single NIC, and is set up to use the network-bridge method. This is probably the easiest way to get xen up and running initially, until you determine whether or not you need either of the other methods.First, the installation of Debian Etch. I downloaded the "netinst" image from http://cdimage.debian.org/debian-cd/4.0_r0/i386/iso-cd/. I mainly accepted defaults for the most part, but I chose to set up my partitions manually as follows (from The Perfect Xen 3.0.3 Setup):

boot 100 MB (Primary) (Location for the new partition: Beginning) (ext3) (Bootable flag: on <-- important, otherwise your system will not boot!)

swap 1GB (Logical) (Location for the new partition: Beginning)

/ 2GB (Logical) (Location for the new partition: Beginning) (ext3)

/vserver the rest (Logical) (Location for the new partition: Beginning) (ext3)

...with one exception. Instead of /vserver, I set up the rest of the disk as a LVM physical volume, and created a volume group named vserver.

Next came package selection. The default selections are Desktop Environment and Standard System. I unselected Desktop Environment and continued, and before long, the base system was installed.

First thing after the reboot, keeping in line with The Perfect Xen 3.0.3 Setup, I set up my static IP address in /etc/network/interfaces, like so:

iface eth0 inet static
address 192.168.1.254
netmask 255.255.255.0
network 192.168.1.0
broadcast 192.168.1.255
gateway 192.168.1.2

Then the hostname:

# echo host.domain > /etc/hostname
# /bin/hostname -F /etc/hostname

I then ran the following commands, as indicated in The Perfect Xen 3.0.3 Setup:

# apt-get remove exim4 exim4-base lpr nfs-common portmap pidentd pcmcia-cs pppoe pppoeconf ppp pppconfig
# apt-get install iproute bridge-utils python-twisted gcc-3.3 binutils make zlib1g-dev python-dev transfig bzip2 screen ssh debootstrap libcurl3-dev libncurses5-dev x-dev

...and this is where I departed from The Perfect Xen 3.0.3 Setup. With Etch, this becomes possible:

# apt-get install xen-linux-system-2.6.18-4-xen-686 xen-tools

...which installs all necessary files for running xen and creating images.
When the installation was finished, I had to make a couple of changes to the default /etc/xen/xend-config.sxp file (This from Xen from Backports on Debian Sarge):

uncomment (network-script network-bridge)
comment (network-script network-dummy)
and
(vif-script vif-bridge) should already be uncommented.

Then, reboot.

To create images, I first had to edit the /etc/xen-tools/xen-tools.conf file. I uncommented lvm = and entered vserver to match my lvm volume group, uncommented debootstrap = 1, changed the default distribution to etch, and changed some of the default size and memory components.

After this, a new xen image can be created with a single command:

# xen-create-image --hostname=newimage.domain --ip=192.168.1.200

One last thing. This bugged the hell out of me until I went searching for the answer. The xendomains init script is supposed to save the virtual machines when shutting down the server, but for some reason, it wasn't working.
I found the answer here. Apparently, when xendomains is parsing the virtual machine hostnames, it was only set to read 17 characters, so with my fully qualified domain names being longer than 17 characters, it was messing everything up.
Hats off to Petter Reinholdtsen for coming up with this patch:

--- /etc/init.d/xendomains.orig 2007-03-10 14:40:09.000000000 +0100
+++ /etc/init.d/xendomains 2007-03-10 14:40:21.000000000 +0100
@@ -128,9 +128,9 @@

parseln()
{
- name=`echo "$1" | cut -c0-17`
+ name=`echo "$1" | cut -c0-25`
name=${name%% *}
- rest=`echo "$1" | cut -c18- `
+ rest=`echo "$1" | cut -c26- `
read id mem cpu vcpu state tm < <(echo "$rest")
}

After these 2 changes, everything is happy. My most painless xen install to date :)

Xen on Debian Sarge

Wed, 11 Apr 2007 03:04:27 -0230

Over a year ago, I installed Xen on my Debian Sarge server. There were many howto's available, but I still had a helluva time getting it up and running. I just chalked it up to inexperience and didn't think too much about it after.

Fast forward to a couple weeks ago. I inherited a new machine to use as my server, so I wanted to take the opportunity to do everything right, with time not being a factor, as I still have my old server running.
This time I followed 3 different howto's to a tee, right from the initial Debian install. Of the 3 methods I tried, only 1 worked - building xen from source.First attempt: The Perfect Xen 3.0.3 Setup For Debian Sarge from howtoforge.com (using pre-compiled binaries section).
The Debian install was smooth enough, as was the installation of the pre-compiled xen binaries, but upon rebooting into the xen kernel, I immediately got a kernel panic where the root partition was not detected. I played around with it for a bit, but to no avail.

Second attempt: Xen from Backports on Debian Sarge from debian-administration.org.
This method involved enabling the backports repository (xen packages are available for Debian Testing) and installing xen using apt-get. Same problem when booting the xen kernel. Kernel panic because the root partition was not detected.

Third attempt: The Perfect Xen 3.0.3 Setup For Debian Sarge from howtoforge.com (using "build from source" section).
This is the one that did the trick. After following the instructions, I had no problem at all booting the xen kernel, and was able to create new virtual servers without any trouble.

Each attempt I made was on a fresh install, formatted after each failed attempt. I assume that there must be very limited hardware support in the stock xen kernels (though my root partition was on an ide drive using ext3).
Once I booted into the new xen kernel, I used xen-tools to create virtual servers with a single command. Very slick!

So if you're planning to give xen an honest go, and are running Debian, stick with building from source. It may seem more tedious and un-debian-like, but trust me, it can save you a headache.

I am still currently getting my virtual servers configured just right so I can let them take over the functions of my current "production" virtual servers on my older box. This will involve, at the final stage, swapping my software RAID1 array over to the new box for use with the virtual file server (I hope to post details on how I did this). Once completed, I will have the old server for testing purposes, and can hopefully generate some useful howtos for all to benefit from.

Feel free to post any Debian/xen experiences; or if you know why the pre-compiled kernels might not be working for me, I'd love to hear it.

NucleusCMS

Tue, 10 Apr 2007 13:55:33 -0230

From B2evolution, to Wordpress, to Drupal, and now to NucleusCMS, I seem to be slowly making my way through all available blog software.

This latest transition comes after a very long period of inactivity. I installed Drupal about a year ago and made a total of 2 posts since then. This actually worked to my benefit, as there doesn't seem to be a Drupal -> NucleusCMS conversion script available anywhere. Luckily, I still had my old Wordpress database, so I was able to convert from that and only lose 2 entries.

I still have some messing around to do with the categories, as Wordpress and Drupal both allowed for entries to belong to multiple categories and NucleusCMS does not, but all in all, it was a very painless conversion.
Drupal really wasn't right for me. It was a bit heavy for my purposes, and I really dislike the available themes, many of which use table-based layouts, which I oppose.

As silly a reason as it is, I chose NucleusCMS because there seems to be a large number of themes that are pretty suitable "out of the box". I really just don't feel like creating my own, or heavily modifying existing themes anymore.

I am hoping to get this site going again, starting with this entry. I have done a lot of different things with my home servers in the past year, and would really like to write a few howto's. Guess we'll see what happens...

Information Security with Colin Percival

Thu, 28 Jul 2005 21:42:17 -0230

The recent disclosure of side-channel techniques to retrieve cryptographic secrets on hyperthreading machines caused stirs in security and operating system development communities. Colin Percival, a FreeBSD security officer, reported the vulnerability and weathered the questions and criticisms. Michael W. Lucas recently interviewed him on this vulnerability, vendors' responses, and security research.

Read about it here.

Wordpress Permalinks for B2evolution Converts

Thu, 28 Apr 2005 09:13:49 -0230

One thing I failed to mention about my Wordpress conversion is the permalink behavior. This is important for anyone who uses trackbacks, or anyone who get search engine hits to particular posts. I noticed that Jeff had the same problem, so I've made a few modifications to the script I wrote to fix it for myself, and am now offering it here. B2evolution's permalink style replaces spaces and other special characters in post titles with underscores, making a permalink something like www.domain.com/2005/02/20/post_title_here. Wordpress, on the other hand, uses dashes, like www.domain.com/2005/02/20/post-title-here. This is what you end up with when you use the b2evo2wp.php script mentioned in the last post. Once I noticed this, I knew all my permalinks would be broken, so I made a quick script to change the existing permalinks back to match the b2evolution ones. This will not change future behavior, so any posts made after the conversion with be like "post-title-here" but we don't care since nobody is linking to those posts yet. This script is only meant to make the permalinks to existing posts consistent with any links pointing to those posts. Requirements: The original b2evolution tables (or at least the "posts" table) used to do the conversion must still exist. Instructions:

Download the file here
Unzip the permalink_fix.zip
Open permalink_fix.php in a text editor and edit the config information. Note that you must enter all information for both the b2evolution and wordpress tables, even if this info is the same.
Upload permalink_fix.php to your webserver
Type the url to permalink_fix.php in your browser. If your mysql information is right, you will be informed of any titles that need to be edited manually (seems to be only those with single quotes in the title).

I will be submitting this script to the author of the b2evo2wp.php script for possible inclusion, but I will also keep it here for those that have already converted. Hope it helps :) [Edit] I should say that this was tested with b2evolution version 0.9.0.10 and Wordpress version 1.5. I cannot guarantee that it will work with other versions.

From b2evolution to Wordpress

Sat, 23 Apr 2005 17:19:10 -0230

Craving change, and having read up on Wordpress, I finally decided to give it a shot. The biggest catalyst for this change was probably the referrer spam and the occasional comment/trackback spam that I was getting with b2evolution.

The installation was a breeze, taking no more than 5 minutes. The problem, however, was getting my b2evolution data imported into my Wordpress tables. Wordpress has a number of importer scripts, including one for the original b2, but unfortunately b2evolution is different enough that this one doesn't work. After some searching on the Wordpress support forum, I found this thread linking to this file, which once renamed to b2evo2wp.php, did the trick.

Wordpress also seems to handle spam better, with some more configurable options for comments, and a plugin I found that bounces referrer spam, so hopefully there will be a noticeable difference in the next few days.

Overall, Wordpress looks to be a very nice blogging engine, though I cannot knock b2evolution either. Both have their benefits, but I think Wordpress is a better fit for me. Maybe now somebody googling to find out how to import b2evo data will find this.

Helpful "make" Tips

Sat, 26 Mar 2005 16:14:44 -0330

make for Nonprogrammers by Dru Lavigne -- If you're a typical FreeBSD user, you may never have compiled C source code on your own. Yet if you've ever issued a make command, it's compiled code for you. How does it do that? What does it do, anyway? And what else can it do? Dru Lavigne answers all of these questions.

Especially useful is this great way to help automate updating your system sources.

On my test system, I already had cvsup up and running and had created a sup file in /root/cvs-supfile. So, I added these lines to /etc/make.conf:

SUP_UPDATE= yes SUP= /usr/local/bin/cvsup SUPFLAGS= -g -L 2 SUPFILE= /root/cvs-supfile

Note: In order for this to work, you must have cvsup-without-gui installed and have configured a SUPFILE at the specified location. If you've installed cvsup-without-gui but haven't yet created a SUPFILE, replace that SUPFILE line with these:

SUPHOST= cvsup.ca.freebsd.org SUPFILE= /usr/share/examples/cvsup/standard-supfile PORTSSUPFILE= /usr/share/examples/cvsup/ports-supfile DOCSUPFILE= /usr/share/examples/cvsup/doc-supfile

When filling in SUPHOST=, please choose a mirror geographically close to you. Also, review the three files in /usr/share/examples/cvsup to pick and choose which parts of the operating system, ports, and docs you wish to update.

When you've finished, run make update from /usr/src to update the specified sources.

I highly recommend adding OnLAMP to your RSS reader, if only for Dru's FreeBSD articles. Always great information.

Novell Is Doing It Right

Sat, 26 Mar 2005 11:02:59 -0330

Having never used NLD (Novell Linux Desktop), I was a little taken aback when I saw such a beaming review.

Of the desktop offerings we have piloted, NLD offered the broadest range of functionality in what we designated as the desktop and mobility space. From the desktop point of view, the system provides a user-friendly environment, ease of administration and an excellent enhancement of the OpenOffice.org productivity suite.

I don't believe there was even one negative point in the review. Tom Adelstein also talks about how Novell is in the progress of switching their entire staff to Linux. It is, afterall, probably tough to sell linux from a windows box.

I remain mostly happy with Ubuntu, except for the occasionally annoying thing where the boot process hangs at "Configuring network devices" (usually solved by hitting the reset button, though I don't know the cause). In an office environment though, NLD probably fits better than any solution to date.

Ubuntu Linux "Hoary Hedgehog": My Experience

Mon, 14 Mar 2005 12:32:10 -0330

Ubuntu Linux is a fairly new linux distribution based on Debian GNU/Linux. Here's a description from Ubuntu's website:

Ubuntu is a Linux distribution that starts with the breadth of Debian and adds regular releases (every six months), a clear focus on the user and usability (it should "Just Work", TM) and a commitment to security updates with 18 months of support for every release. Ubuntu ships with the latest Gnome release as well as a selection of server and desktop software that makes for a comfortable desktop experience off a single installation CD.

The latest "Preview" release, 5.04, or Hoary Hedgehog (Hoary) was released just a few days ago, and, not liking to get too comfortable with any one distro, I decided to give it a shot. I started by downloading the live cd, just to make sure I wasn't going to have any problems. It started up with no problems, and it really impressed me with how clean it seemed. I had read beforehand that it included only the Gnome desktop environment. This was my main reason for wanting to try the live cd first. Before settling on Xfce4 on my Suse desktop, I was a KDE guy. I was a little surprised to find how far Gnome had come since the last time I used it (Redhat 7.3). On to the installation!

The install, not surprisingly, is not graphical, but ncurses based like Debian's. It differs from Debian's though in that it minimizes the amount of user input required. This is a big plus for newbies. I didn't get a real feel for partitioning from the installation since my partitions were already set up. All I had to do was select hda2 and set its mount point to /boot, and select hda4 and set its mount point to /. The installation detected my swap partition and assigned it automatically. The installation as a whole was very straight-forward.

One thing I don't like about the default installation is that the root account is disabled. I read about this beforehand as well, so I was ready for it. This is actually a good thing to have done, since all the launchers that require root access are run with sudo, prompting for the user's own password, but I do think it is still necessary to educate the user about sudo on the command line, just in case.

I had been using Suse long enough that I expected to be able to set up my LDAP authentication and nfs mounted /home right away. This software had to be added manually though. "That's ok", I thought. A perfect opportunity to check out synaptic, the gtk frontend for Debian's apt/dpkg package manager. I am impressed. I wasn't sure if it would stack up to Suse's YAST for installing/updating packages, but I believe it exceeds it easily.

I did run into some problems with LDAP and nfs, but admittedly, this was due to misconfiguration of my FreeBSD server (my bad). Most of the problem coming from the lack of nfs file locking capability. Enabling rpc.statd and rpc.lockd on the nfs server took care of those problems, and after editing 4 pam files, my LDAP authentication was up too.

Most of the software I regularly use was already loaded. Rhythmbox took the place of XMMS, though I did have to install the gstreamer-mad package for mp3 support. No problem. The default browser is Firefox, and other default software includes Evolution, Gaim, Xchat, the Gimp, Openoffice.org, and even a terminal server client. One disappointment was that I found I couldn't use Evolution due to yet more file locking issues stemming from my nfs mounted /home. I could compile it myself and disable file locking, but being used to Thunderbird anyway, I just decided to use it instead.

This is a preview release, which is a nice way to say open beta, so bugs are to be expected, but overall, Ubuntu is well on its way to becoming a very excellent desktop linux. As far as I'm concerned, it already is if you're like me and enjoy fixing problems as they pop up. It's how we learn. It is also very nice to see a distribution that doesn't try to please everyone by including every known window manager and desktop environment. It's not unheard of, but I would definitely like to see more of this. Ubuntu immediately struck me as a very cohesive experience, while avoiding severe bloat. It feels both very simple and very refined. I highly recommend this one to anybody looking for a new distro to try.

SCO v. IBM: The Short Version, by toads_for_all

Sat, 5 Mar 2005 15:17:23 -0330

For anybody like myself who is too busy to keep up with the SCO/IBM saga, here is a funny little summary.

Sure it's sort of a joke, but this is pretty much what happened.

IBM: But Caldera contributed parts of the original to Linux. Aren't we just doing what Caldera did, treating the derivatives the same way Caldera did the original by contributing parts to Linux?

SCO: Um, I mean, well, I guess.....Did I say contracts? I meant it's about copyrights.

NOVELL: But you don't own all the copyrights, we do.

SCO: Do not!

Read the rest here.