I guess that almost all desktop systems in the office or home are derived from one prototype, the IBM PC (personal computer), which was built around the concept of an open architecture that IBM introduced 30 years ago. It’s true that the company itself didn’t quite understand how revolutionary open hardware architecture would be a few years later, in the mid-80s. Whatever the case, the fact remains that the PC became popular because many independent manufacturers could produce a wide range of compatible peripherals. This, in turn, allowed a competitive market to develop, eventually making desktop PCs available to everyone.

The open development model of Linux first embraced the desktop market of x86 computers, and then proceeded to “capture” other architectures as well (now more than 20), ranging from x86, ARM and MIPS, to Alpha, SPARC, PowerPC, IA64 and S390.

While “desktop” platforms (x86 and ARM) have many different Linux distributions and other operating systems that can run on them, the “pure” server architectures like IA64, S390 and Power cannot boast of this — the number of distributions is much, much smaller. This fact is understandable. Quality and the cost of server hardware for industrial applications is immeasurably higher than for the usual PC. Moreover, not every Linux distribution vendor can find the resources to build and provide long-term support for major server hardware architecture — even if they are willing to.

Some time ago, I got my hands on an IBM IntelliStation POWER 285 workstation, considered the closest model to large industrial servers. The main brain of the station is a dual-core POWER 5+ processor.

Although announced back in 2005, the system is serious stuff even now. Judge for yourself: 4 GB RAM (max capacity 32 GB), with a disk subsystem that has 2×146 GB SCSI drives (max disk storage 1.2 TB). It has a self-diagnostic system and memory “self-healing” technology known as ChipKill, a remote HMC (Hardware Management Console) and a pair of NICs working at 1 Gbit. This system is unbeatable for an entry-level enterprise server and simply forces you to take another respectful look at IBM  — Big Blue creates innovative and striking products.

Well, that’s enough of an introduction. Let’s learn Linux installation and operation on IBM hardware. And later decide how it is possible to effectively exploit Linux on Power.

Remote access

IBM produces equipment that has OpenFirmware (OF) inside. This is analogous to the traditional PC BIOS used on x86 systems. With OF’s help, you can assign which source should be used to start an OS, perform a number of diagnostic procedures, etc. OF itself uses a ServiceProcessor — a special unit embedded into the big tower. When you plug in the power cord, first the ServiceProcessor starts. That initially sets up a remote access option, and then awaits the user’s response.

The user has to press the hardware Start button on the front panel. So, when plugged in, IntelliStation is never completely turned off — even with the main CPUs off, the ServiceProcessor doesn’t sleep. The scheme is reminiscent of a similar product — iLO, developed by Compaq/HP. iLO also lets one start and control a server remotely.

It’s worth noting that IBM also has additional functions run by HMC (Hardware Management Console), among which are planning and allocation of physical resources to logical partitions (LPAR/DLPAR), splitting main CPU performance onto logical CPUs (Capacity On Demand) and related actions that provide virtualisation management on POWER-based processors. Unfortunately, I can only mention these capabilities, and not go any deeper into the subject, because that would be beyond the scope of this review.

So, in order to establish an access link to the ServiceProcessor and later install Linux, you will need an RS232(F)<->RS232(F) cable and a dedicated workstation, from which you can directly control your IntelliStation 285. As a console, any PC with an installed command-line tool like minicom or surecom (a GUI alternative) will suit. Set the communication speed to 19,200 bit/s, the standard default rate for IBM hardware, plug in the power cord and wait for the login prompt after running minicom -D /dev/ttyS0 -s 19200 -o. See Figure 1.

Figure 1: Control Power hardware via RS232 cable

To start the whole system and gain access to OF, from where you can select an OS source,  choose: 1. Power/Restart Control -> 1. Power On/Off System -> 8. Power on. After that, in approximately two minutes, you will hear the noise of the fans — the big system has started initialisation and a minute later the hardware will be completely initialised, signalled via two speaker beeps.

At the same time, on the console, you should see a message from SMS (System Management Services) as in Figure 2, where you choose 1 to enter the system OF menu and then select 1. Select Install/Boot Device to choose the boot device (which you will run the installer from).

Figure 2: SMS greeting before entering OpenFirmware menu

You should see something like Figure 3, where I chose 4. IDE, because the built-in CD-ROM is connected to the IDE connector. In your version, it might be 3. CD/DVD.

Figure 3: Choose installation source

In the subsequent menu (Figure 4), I chose 2. Normal Mode Boot, which is suitable for installation, and confirmed exit from SMS. My SLES 11′s Linux kernel correctly identified the IntelliStation 285.

Figure 4: Choose boot mode

The installation process will proceed as follows:

Welcome to SuSE Linux Enterprise 11!

Type  "install"  to start the YaST installer on this CD/DVD
Type  "slp"      to start the YaST install via network
Type  "rescue"   to start the rescue system on this CD/DVD

Welcome to yaboot version r22.8-r1190.SuSE
booted from '/pci@800000020000003/pci@2,3/ide@1/disk@0:1,\suseboot\yaboot.ibm'
running with firmware 'IBM,SF240_403' on model 'IBM,9111-285', serial 'IBM,0206ABB30', partition '0'
Enter "help" to get some basic usage information
boot:
Please wait, loading kernel...
Allocated 01800000 bytes for executable @ 03000000
Elf32 kernel loaded...

SuSE Linux zImage starting: loaded at 03000000-0469cd30 (4000000/0/02039a68; sp: 02c9fd50)
uncompressing ELF header done. (00000100 bytes)
Allocated 0164c090 bytes for kernel @ 04800000
Leave 0122a5ba bytes for initrd @ 034668bb
uncompressing kernel done. (00f5a920 bytes)
entering kernel at 04810000(34668bb/122a5ba/02039a68)

The installation procedure reminds me of what happens on the x86, with just one difference — in our case, control is via a serial line in text VT102 mode, so some characters may not be displayed properly as you can see in Figure 5.

Figure 5: SUSE Linux installation could be fun

Several menu items, such as the network address, may require you to return to re-edit these fields. We can use Tab/Alt+Tab to move between text interface fields. These keyboard shortcuts are functional even in this very minimal mode.

Embedded Linux on ServiceProcessor

In addition to control via a serial line, IntelliStation (and in general, all IBM products) offers management via an ordinary Web browser. This Web control mechanism (Advanced System Management — see Figure 6) is provided by a tiny computer within the IntelliStation, with Linux and BusyBox utilities.

Figure 6: Management via Web interface

This ServiceProcessor also uses Power technology — a PowerPC 405 CPU from IBM [Datasheet PDF] but with a much smaller form factor. This uses a low-power processor FSP (Flexible Service Processor) running at about 200 MHz (see Figure 7).

Figure 7: Embedded Linux runs on PowerPC 405 from IBM

ASM provides the same controls as the serial-line SMS but in a more usable manner (Figure 8). It lets you manage the system remotely — start/stop it, view logs, allocate resources (like Capacity on Demand parameters), etc.

Figure 8: ASM's user-friendly controls presentation

Let’s briefly see how this subsystem is launched. When you connect the mains power to the box, the ServiceProcessor automatically starts the embedded Linux. After initially testing  some components of a larger block, ServiceProcessor displays on a front panel the readiness status for full system start. When you hit the “Power On” button, OpenFirmware loads and ultimately passes control to your Linux boot loader, yaboot.

Like with the large system, this small subsystem has two independent Ethernet controllers, referred to as HMC 1 and HMC 2. It makes sense to connect them to separate LANs/VLANs for a special management network. First, it helps access restriction and security. Second, you can avoid a network traffic overload on a single LAN.

IP addresses for these ports can be supplied by a DHCP server on a management network. If one isn’t found, default addresses are automatically assigned: 192.168.2.147 (port C7-T1, on HMC1) and 192.168.3.147 (port C7-T2, HMC2). You can log in via telnet, with the default credentials admin/admin.

$ mount
/dev/mtdblock/0/3 on / type cramfs (rw)
none on /dev type devfs (rw)
proc on /proc type proc (rw)
/dev/mtdblock/0/1 on /opt/extucode type jffs2 (rw)
/dev/mtdblock/0/1 on /etc type jffs2 (rw)
/dev/mtdblock/tmpram/tmpram on /tmp type jffs2 (rw)
none on /var type tmpfs (rw)
/dev/mtdblock/pcscore/pcscore on /core type jffs2 (rw)
/dev/mtdblock/pcsram/pcsram on /opt/p1 type jffs2 (rw)
shm on /var/shm type shm (rw)
/dev/mtdblock/1/1 on /alt/extucode type jffs2 (rw)
/dev/mtdblock/2/3 on /opt/fips type cramfs (rw)
/dev/mtdblock/nvram/nvram on /opt/p2 type zofs (rw,sync)
$ uptime
8:49am  up 1 day, 14:53, load average: 0.52, 0.20, 0.07
$ uname -a
Linux fsp 2.4.18-rc4 #1 Wed Mar 25 15:01:22 UTC 2009 ppc unknown
$ cat /proc/version
Linux version 2.4.18-rc4 (root@mcpbuild6) (gcc version 3.2.3 20030401 (prerelease)) #1 Wed Mar 25 15:01:22 UTC 2009

As we see, Embedded Linux uses JFFS/CRAMFS filesystems, which are common in tiny products. An interesting feature is NVRAM memory mounted as storage (/dev/mtdblock/nvram/nvram). Reading data from its files gives the status of different subsystems. You can also change some values using commands from /opt/fips/bin — particularly, you can even try to launch several subsystems (or maybe the full system box) with /opt/fips/bin/powr_fsp_spcn, which checks the current system status, and accordingly launches /bin/powr_L4 or /bin/powr_sqh. The ASM Web server (named webs), when you click the button Save settings and power on does the same.

Available distros for POWER

Originally, I planned to use Red Hat Enterprise Linux as the main OS on this system. However, neither RHEL 5.7 nor 6.2 Beta, nor even 6.2 started correctly. RHEL 6.2 Beta and a more recent version 6.2 spews a kernel trap error during initial startup. Yes, the IntelliStation 285 wasn’t in the supported hardware list, but neither was it for SLES 10/11. Still, SLES 10 and 11 support it unofficially. Let us see how the installation process goes, and whether we can bring the system to production mode.

In addition to RHEL and SLES, numerous systems based on the POWER processor even supported Ubuntu 10.04 (LTS edition). However, my experience proved that this model wasn’t suitable for Ubuntu — it cannot start properly. Moreover, newer versions of Ubuntu come without any POWER/PowerPC architecture support at all, unfortunately.

I checked if the most recent release (6.0.3) of Ubuntu’s parent distribution, Debian, could boot. It wasn’t able to start — it drops directly into OpenFirmware. Another failure! Well, let’s not despair. Quite possibly, the Debian community will fix it.

Oddly enough, a completely amateurish project called Crux PPC has a fully working distribution — Crux PPC 2.7a, which not only loads on Power hardware, but also provides video output via correctly-set framebuffer mode. None of the other distributions mentioned can boast of that! All provided only a serial-line text mode. In general, Crux PPC runs as a LiveCD and has no installer, so I advise professionals to use it. Again, this topic is out of the scope of this article, so I will leave it for later.

Performance

As I mentioned a bit earlier, IBM products are one level, or a “head above” their competitors. Comparing the very different Power and x86 architectures is hard, because there are no objective criteria for such a comparison. However, from an ordinary user’s perspective, we can try the 7Zip archiver in its benchmark mode. This program is completely open source, so anyone can compile it.

Let us compare IntelliStation based on the POWER5+ processor with 2 cores and 4 GB RAM, manufactured in 2005, with an Intel Core 2 Duo (2 cores with hyper-threading) and 4 GB RAM, but released three years later (2008). First, the IntelliStation:

linux:~/src/p7zip_9.20.1/bin # ./7za b

7-Zip (A) 9.20  Copyright (c) 1999-2010 Igor Pavlov  2010-11-18
p7zip Version 9.20 (locale=en_US.UTF-8,Utf16=on,HugeFiles=on,2 CPUs)

RAM size:    3632 MB,  # CPU hardware threads:   2
RAM usage:    425 MB,  # Benchmark threads:      2

Dict        Compressing          |        Decompressing
      Speed Usage    R/U Rating  |    Speed Usage    R/U Rating
       KB/s     %   MIPS   MIPS  |     KB/s     %   MIPS   MIPS

22:    1931    99   1891   1878  |    24309   100   2198   2194
23:    1871   100   1916   1907  |    24132   100   2212   2209
24:    1816    99   1964   1953  |    23917   100   2221   2219
25:    1780    99   2043   2032  |    23525   100   2213   2212
----------------------------------------------------------------
Avr:           99   1954   1942               100   2211   2209
Tot:          100   2082   2076

Next, let us test the Core 2 Duo system:

Intel(R) Core(TM)2 Duo CPU     E6550  @ 2.33GHz
7-Zip 9.04 beta  Copyright (c) 1999-2009 Igor Pavlov  2009-05-30
p7zip Version 9.04 (locale=en_US.UTF-8,Utf16=on,HugeFiles=on,2 CPUs)

RAM size:    4006 MB,  # CPU hardware threads:   2
RAM usage:    425 MB,  # Benchmark threads:      2

Dict        Compressing          |        Decompressing
      Speed Usage    R/U Rating  |    Speed Usage    R/U Rating
       KB/s     %   MIPS   MIPS  |     KB/s     %   MIPS   MIPS

22:    2588   138   1831   2518  |    33730   165   1850   3045
23:    2532   136   1903   2580  |    32265   159   1853   2954
24:    2588   142   1958   2783  |    32828   165   1843   3046
25:    2540   144   2011   2900  |    30299   155   1838   2849
----------------------------------------------------------------
Avr:          140   1925   2695               161   1846   2974
Tot:          150   1886   2834

As you can see, the performance doesn’t differ much. Even a slightly outdated IBM station can compete with a relatively new x86 product.

Despite the fact that equipment based on the Power processor is marketed by IBM as a corporate power-horse for commercial operating systems like AIX and IBM i (produced by IBM itself), we have seen that it can also be used with Linux. Thus, virtually the entire software stack that successfully works on x86 can also be run on Power — Apache, Java, WebSphere, MySQL, etc.

Taking into account the fact that IBM hardware usually comes with unique technologies like the hardware management console and the hardware virtualisation hypervisor, you can easily create an effective solution by combining the best features from both the hardware and software sides — the corporate and the public world.

‘Oneiric’ means ‘dreamy’ and in his blog, Mark Shuttleworth writes at length about choosing a suitable alliterative adjective to describe the ocelot, a small wild leopard. He even quotes from T.S. Eliot’s famous poem on “The Naming of Cats”. He finally settles on “Oneiric Ocelot” because it seems to capture how innovation happens — “part daydream, part discipline.”

This release is cosmetically pleasing, and gives developers even more options to create efficient interfaces that are aesthetically delightful also.

The Ubuntu installation — ‘Say Cheese’

The re-tweaked Ubuntu installer is going to amaze you with its new picture-taking abilities. If you’re installing Ubuntu 11.10 on a notebook or netbook equipped with a camera, it asks you to take a picture of yourself; this picture will recognise you in account settings, the login screen and the lock screen.

Ubuntu Install 'Say Cheese'

The rest of the installation is the same as in the previous release. While installing, keep your Internet connection on, or your installer will get stuck. Also don’t forget to tick “Install 3rd party codecs/restricted drivers”. The installer is very fast, and won’t let you wait for long.

A lightweight ‘LightDM’

As soon as the system reboots, you will be greeted with the slick, light-weight LightDM login screen. This new débutante is a very fast desktop manager, and offers you options for Unity 3D (default) and Unity 2D sessions. If your card doesn’t support Unity 3D, you can choose Unity 2D from LightDM. The Gnome Shell is completely dropped from Ubuntu, but you can install it from the Ubuntu Software Centre, or by using apt-get install. (See the box with suggestions at the end of the article on 11 things to install on a fresh Oneiric install.)

LightDM login screen

The ‘Lenses’ included in Unity Dash

After receiving lots of complaints, Ubuntu developers refined and re-polished the Unity interface, and added lenses for App Launcher, File Search and Music Search. App lenses let you find stuff faster. File lenses do all the remembering, so you don’t have to. It gives you myriad filtering options. For example, Music lenses keep track of your music collections, what you have and what you don’t.

The Unity sidebar now shades itself according to the desktop wallpaper. The interface is translucent, with bigger icons. Compiz also hides a lot of oomph under the hood.

Unity Dash

Music Lenses

Unity Sidebar

Multi-task in style

A new addition to this release is a modified Alt-Tab switcher with preview features, in which you can use the down arrow key to see a preview. This really cool feature adds a plus to Ubuntu’s “coolness rating”. The workspace switcher also got touched up; it displays an orange border around the active workspace. You can also move applications across different workspaces.

Alt-Tab Switcher preview

Workspace Switcher

A completely revamped USC

The Ubuntu Software Centre got a complete makeover — and with this release, Ubuntu dropped Synaptic completely. There are thousands of apps to choose from. The USC also adds new “top-rated” views to the main category and all sub-category pages, along with significant speed improvements for standalone .deb installer files. It allows you to edit and delete your own software reviews.

Another new addition worth mentioning is “One Conf”; enable it from the File menu –> Sync between Computers, and it keeps your installed apps in sync between computers! This feature comes handy when you have more than one Ubuntu device. It also gives options for adding applications to the launcher right after install finishes. With this release, USC is upgraded to 5.0.

Ubuntu Software Centre home screen

A few welcome new defaults

With every incremental Ubuntu release, new software defaults come as a surprise. Thunderbird 7 is the new emailing client, because it has got better plugin support, and includes better menu and launcher integration. The Déjà Dup application is included as a default backup tool. It makes it easy to upload backups to Ubuntu One, the cloud storage initiative by Ubuntu.

Déjà Dup backup application

Déjà Dup options

Ubuntu One ‘graduated’

Ubuntu One debuted with Ubuntu 9.04 Jaunty Jackalope, and as promised, Ubuntu developers delivered it completely integrated with Android, iOS and Windows devices. Ubuntu One provides you with documents, pictures and music syncing across computers and mobile devices, apart from the Ubuntu One music store to stream the tracks you have to your mobile devices (and buy the music you don’t have). The Ubuntu One Bookmark syncing feature has been discontinued with this release, due to performance-related issues.

Ubuntu One comes with 5 GB of free space, and fulfils all your requirements. You can also opt for a 20 GB pack, by spending approximately US$ 30 per year.

Go social with Gwibber

Gwibber, the much-talked-about social-networking client, is completely rewritten from scratch, to give you a hassle-free social experience online. It’s got improved performance, and a new interface using the most recent GNOME technologies.

The nitty-gritty

There are some details worth mentioning, such as a better universal bar, windows controls hidden by default on maximised apps (they reappear when you hover your mouse on the top left-hand side).

The power button is reincarnated and is much more useful now; you can access the Update Manager and GNOME Control Centre, and can switch between accounts without much hassle. Indicators got a refresh, as they are ported to the GTK3-based indicator stack. Launchers and panels also show improved performance, as they too are ported to GTK3.

Power button menu

A windows-snapping feature is also included in the new avatar of Compiz, which mimics Windows ‘Aero snap’. There is multi-architecture support, which enables you to install 32-bit apps to 64-bit installs. This version also has better short-cut key integrations. The boot time from a live USB was a mere 29 seconds on a Core2Duo laptop, and one minute 6 seconds on my netbook. Battery backup performance still requires a lot of work, though.

Window snapping

Application window preview

“Ubuntu” is an ancient African word that means “Humanity to others.” It also means, “I am what I am because of who we all are,” and Ubuntu OS actively works towards delivering the spirit of Ubuntu to the world of computers with its every release. Enjoy the new release of Ubuntu!

After this release, Mark Shuttleworth also announced the Ubuntu 12.04 LTS release, codenamed “Precise Pangolin”.

VirtualBox

Sun Microsystems, and now Oracle, oversee the development of the emulation project called VirtualBox. However, this package couldn’t be found in the OpenIndiana (OI) repositories. To install it in OI, grab the gzipped package here and manually install it. The archive is called “VirtualBox 4.1 for Solaris hosts x86/amd64″. Please unpack it, and then run pkgadd:

root@openindiana:/home/anton/Install# pkgadd -d VirtualBox-4.1.0-SunOS-r73009.pkg

The following packages are available:
1  SUNWvbox     Oracle VM VirtualBox
                  (i386) 4.1.0,REV=2011.07.19.12.47.73009

Select package(s) you wish to process (or 'all' to process
all packages). (default: all) [?,??,q]: a

Processing package instance <SUNWvbox> from </home/anton/Install/VirtualBox-4.1.0-SunOS-r73009.pkg>

[...snip...]

The following files are being installed with setuid and/or setgid
permissions:
/opt/VirtualBox/amd64/VBoxBFE <setuid root>
/opt/VirtualBox/amd64/VBoxHeadless <setuid root>
/opt/VirtualBox/amd64/VBoxNetAdpCtl <setuid root>
/opt/VirtualBox/amd64/VBoxNetDHCP <setuid root>
/opt/VirtualBox/amd64/VBoxSDL <setuid root>
/opt/VirtualBox/amd64/VirtualBox <setuid root>
Do you want to install these as setuid/setgid files [y,n,?,q] y

This package contains scripts which will be executed with super-user
permission during the process of installing this package.

Do you want to continue with the installation of <SUNWvbox> [y,n,?] y

Installing Oracle VM VirtualBox as <SUNWvbox>

[...snip...]

Installation of <SUNWvbox> was successful.

Running VirtualBox in OpenIndiana

File operations, and the general VirtualBox used in OI, are no different than on Linux. You can even exchange VMs created in OI with Linux, and vice-versa.

QEMU

However, some developers need to deal with platforms other than x86 — for example, ARM and PowerPC. So, what’s the alternative to VirtualBox? The solution is straightforward — QEMU. It can emulate x86 systems and a dozen other architectures; e.g., Linux for ARM with qemu-system-arm, or MacOS X with qemu-system-ppc64.

The build process for QEMU 0.15 requires the gettext@0.5.11 package, as well as developer packages that contain definitions of mathematical functions. I performed the following operations:

root@openindiana:/home/anton# pkg info | egrep "math|gettext"
Name: developer/gnome/gettext
FMRI: pkg://openindiana.org/developer/gnome/gettext@0.5.11,5.11-0.151:20110523T143743Z
Name: system/library/math
FMRI: pkg://openindiana.org/system/library/math@0.5.11,5.11-0.151:20110523T150249Z
Name: system/library/math/header-math
FMRI: pkg://openindiana.org/system/library/math/header-math@0.5.11,5.11-0.151:20110523T150250Z

I was able to build QEMU 0.15 without problems. However, I have to disable GuestAgents support, mainly because QEMU 0.15 calls functions with O_ASYNC parameters, which isn’t supported on Solaris.

QEMU 0.15 compiled on OI; VM with Ubuntu Server

Windows 2003 EE in QEMU 0.15

You could compile the previous version, 0.14, or do a quick hack on line 493 in qemu-ga.c, from:

fd = qemu_open(s->path, O_RDWR | O_NONBLOCK | O_ASYNC);

…to:

fd = qemu_open(s->path, O_RDWR | O_NONBLOCK);

Or, as mentioned earlier, compile it without GuestAgents support:

./configure --prefix=/opt/qemu-0.15 –disable-guest-agent
make && make install ## run as root

After that, you should prepare a file image for the virtual OS to install to, and run it as follows:

/opt/qemu-0.15/bin/qemu-img create ubuntu2.img 2000M
/opt/qemu-0.15/bin/qemu -hda ubuntu2.img -cdrom /home/anton/ISO/ubuntu-10.10-server-i386.iso -M pc -m 768 -vga cirrus -no-acpi

Similarly, you can use other OSs, for example, Windows 2003 EE:

/opt/qemu-0.15/bin/qemu -hda windows2003.img -cdrom /home/anton/ISO/win2003_sp2.iso -M pc -m 1024 -vga cirrus

Disk management

Let’s imagine that you have a new drive to be used for data storage. It could be a special, dedicated database storage pool. This drive should be initialised first, partitioned with a file format that is used by OI. You can use the format command (it can be used in interactive mode):

root@openindiana:/home/anton# format
Searching for disks...
Failed to inquiry this logical diskdone

AVAILABLE DISK SELECTIONS:
       0. c1d0 < cyl 2868 alt 2 hd 255 sec 63>
          /pci@0,0/pci-ide@1,1/ide@0/cmdk@0,0
       1. c2d1 <QEMU HAR-QM0000-0001-2.00GB>
          /pci@0,0/pci-ide@1,1/ide@1/cmdk@1,0

We can format the whole drive with the format option in the menu (not shown), or assign a label using volname.

The last step is to add the drive to the ZFS pool—and the new storage space is ready to use:

root@openindiana:/home/anton# zpool create pool c2d1

You can list the available pools and see that both
pools (i.e., disks) are online:

oot@openindiana:/home/anton# zpool list
NAME    SIZE  ALLOC   FREE    CAP  DEDUP  HEALTH  ALTROOT
pool   1.98G   124K  1.98G     0%  1.00x  ONLINE  -
rpool  21.9G  9.17G  12.7G    41%  1.00x  ONLINE  -

Zones management

In addition to VirtualBox and QEMU virtualisation, OI (and OpenSolaris) offers fairly well-developed virtualisation within the OS, called Solaris Zones. Unfortunately, OI does not yet support zones that can run Linux (though it was available in the previous OpenSolaris version). OI supports mere native zones — when the system runs zones (or containers) with OI inside them. Let’s explore how to create them. Begin by using the following code:

root@openindiana:/home/anton# zonecfg -z NewZone
NewZone: No such zone configured
Use 'create' to begin configuring a new zone.
zonecfg:NewZone> create
zonecfg:NewZone> add net
zonecfg:NewZone:net> set physical = rtls0
zonecfg:NewZone:net> set address = 192.168.254.200
zonecfg:NewZone:net> end
zonecfg:NewZone> set zonepath=/pool/NewZone
zonecfg:NewZone> commit
zonecfg:NewZone> exit

You have created a new zone. It is stored under /pool/NewZone on the disk you just attached and initialised. For a zone to be complete, you must install the bare OI system into this area:

root@openindiana:/home/anton# zoneadm -z NewZone install
A ZFS file system has been created for this zone.
   Publisher: Using openindiana.org (http://pkg.openindiana.org/dev-il/ ).
       Image: Preparing at /pool/NewZone/root.
Refreshing catalog 1/1 openindiana.org
Caching catalogs ...
       Cache: Using /var/pkg/download.
Sanity Check: Looking for 'entire' incorporation.
  Installing: Core System (output follows)

               Packages to install:     1
           Create boot environment:    No

[...snip...]

Postinstall: Copying SMF seed repository ... done.
Postinstall: Applying workarounds.
        Done: Installation completed in 771.344 seconds.

Next Steps: Boot the zone, then log into the zone console (zlogin -C)
              to complete the configuration process.

The system is installed into an isolated container. Let’s run it!

root@openindiana:/home/anton# zoneadm -z NewZone boot

zone 'NewZone': WARNING: rtls0:1: no matching subnet found in netmasks(4): 192.168.254.200; using default of 255.255.255.0.

Now, perform a minimal configuration of this zone.
Connect to a virtual console with zlogin -C NewZone:

root@openindiana:/home/anton# zlogin -C NewZone
[Connected to zone 'NewZone' console]
61/99
85/99
99/99
Hostname: NewZone

Loading smf(5) service descriptions: 3/3

What type of terminal are you using?
 1) ANSI Standard CRT
 2) DEC VT100
 3) PC Console
 4) Sun Command Tool
 5) Sun Workstation
 6) X Terminal Emulator (xterms)
 7) Other
Type the number of your choice and press Return: 6
Creating new rsa public/private host key pair
Creating new dsa public/private host key pair
Configuring network interface addresses: rtls0.

System identification is completed.

rebooting system due to change(s) in /etc/default/init

[NOTICE: Zone rebooting]

SunOS Release 5.11 Version oi_151 64-bit
Copyright (c) 1983, 2010, Oracle and/or its affiliates. All rights reserved.
Hostname: NewZone

NewZone console login: root
Password:
Aug  8 14:09:54 NewZone login: ROOT LOGIN /dev/console
OpenIndiana (powered by illumos)    SunOS 5.11    oi_151    July 2011

The one-time configuration of the new zone is done. Review the new zone’s available resources. They’re completely isolated, like the whole environment, in general:

root@NewZone:~# df
/                  (pool/NewZone/ROOT/zbe): 3007372 blocks  3007372 files
/dev               (/dev              ):       0 blocks        0 files
/proc              (proc              ):       0 blocks    29917 files
/system/contract   (ctfs              ):       0 blocks 2147483588 files
/etc/mnttab        (mnttab            ):       0 blocks        0 files
/system/object     (objfs             ):       0 blocks 2147483418 files
/etc/svc/volatile  (swap              ): 5920888 blocks   578264 files
/lib/libc.so.1     (/usr/lib/libc/libc_hwcap1.so.1): 3007372 blocks  3007372 files
/dev/fd            (fd                ):       0 blocks        0 files
/tmp               (swap              ): 5920888 blocks   578264 files
/var/run           (swap              ): 5920888 blocks   578264 files

Restarting NewZone does not interrupt the core system;
rebooting takes less than a minute:

root@NewZone:/# reboot
Aug  8 14:15:37 NewZone reboot: initiated by root on /dev/console

[NOTICE: Zone rebooting]

SunOS Release 5.11 Version oi_151 64-bit
Copyright (c) 1983, 2010, Oracle and/or its affiliates. All rights reserved.
Hostname: NewZone

NewZone console login: root
Password:
Aug  8 14:16:18 NewZone login: ROOT LOGIN /dev/console
Last login: Mon Aug  8 14:09:54 on console
OpenIndiana (powered by illumos)    SunOS 5.11    oi_151    July 2011
root@NewZone:~# date
Mon Aug  8 14:16:22 CEST 2011

I propose to build a QEMU package, so it can be installed later inside NewZone. Why is this so important? It could sometimes be essential to have a standalone package, if it’s new and not in the OI repository yet. Or, in the opposite case — for your own corporate program which you don’t wish to redistribute. As an example, let’s analyse how to create a package from the set of programs in /opt/qemu-0.14.

The first step is to prepare two files, prototype and pkginfo.

# echo 'i pkginfo' > prototype
# pkgproto /opt/qemu-0.14/  >> prototype

In prototype, you should specify those files and directories that the future package will contain. The pkginfo file will contain all information about this package:

root@openindiana:~/pkg# cat pkginfo
BASEDIR=/opt/qemu-0.14
TZ=PST
PATH=/sbin:/usr/sbin:/usr/bin:/usr/sadm/install/bin
PKG=qemu
ARCH=x86
NAME=qemu
VERSION=0.14
CATEGORY=application
DESC=QEMU virtualization
VENDOR=QEMU.ORG
EMAIL=qemu@qemu.org
PKGSAV=/var/sadm/pkg/qemu/save

Begin building the package itself:

root@openindiana:~/pkg# pkgmk -b /opt/qemu-0.14 -r /
## Building pkgmap from package prototype file.
## Processing pkginfo file.
WARNING: missing directory entry for </opt>
WARNING: parameter <PSTAMP> set to "openindiana20110814071557"
WARNING: parameter <CLASSES> set to "none"
## Attempting to volumize 84 entries in pkgmap.
part  1 -- 123944 blocks, 428 entries
## Packaging one part.
/var/spool/pkg/qemu/pkgmap
/var/spool/pkg/qemu/pkginfo
/var/spool/pkg/qemu/root/opt/qemu-0.14/bin/qemu
[...snip...]
/var/spool/pkg/qemu/root/opt/qemu-0.14/share/qemu/vgabios-stdvga.bin
/var/spool/pkg/qemu/root/opt/qemu-0.14/share/qemu/vgabios-vmware.bin
/var/spool/pkg/qemu/root/opt/qemu-0.14/share/qemu/vgabios.bin
## Validating control scripts.
## Packaging complete.

The package is ready, and is stored in /var/spool/pkg. The last step is to convert it to a special pkg format:

root@openindiana:~/pkg# pkgtrans -s /var/spool/pkg/ qemu.pkg

The following packages are available:
  1  qemu     qemu
              (x86) 0.14

Select package(s) you wish to process (or 'all' to process
all packages). (default: all) [?,??,q]: 1
Transferring <qemu> package instance
root@openindiana:~/pkg# ls -l /var/spool/pkg/
total 58668
drwxr-xr-x 3 root root        5 2011-08-14 08:15 qemu
-rw-r--r-- 1 root root 59996160 2011-08-14 08:19 qemu.pkg

Now, transfer qemu.pkg to the isolated area, and install it in NewZone:

root@NewZone:/export# pkgadd -d qemu.pkg

The following packages are available:
  1  qemu     qemu
              (x86) 0.14

Select package(s) you wish to process (or 'all' to process
all packages). (default: all) [?,??,q]: 1

Processing package instance <qemu> from </export/qemu.pkg>

qemu(x86) 0.14
QEMU.ORG
## Processing package information.
## Processing system information.
## Verifying disk space requirements.
## Checking for conflicts with packages already installed.
## Checking for setuid/setgid programs.

Installing qemu as <qemu>

## Installing part 1 of 1.
/opt/qemu-0.14/bin/qemu
...
/opt/qemu-0.14/share/qemu/vgabios.bin
[ verifying class <none> ]

QEMU installation in the zone was successful. You’ll also notice that the core system now has a new virtual device rtls0:1, which will filter all traffic in and out of the isolated container:

root@openindiana:~# ifconfig rtls0
rtls0: flags=1000843<UP,BROADCAST,RUNNING,MULTICAST,IPv4> mtu 1500 index 3
        inet 10.1.0.136 netmask ffffff00 broadcast 10.1.0.255
        ether e2:62:0:a4:d5:24
root@openindiana:~# ifconfig rtls0:1
rtls0:1: flags=1000843<UP,BROADCAST,RUNNING,MULTICAST,IPv4> mtu 1500 index 3
        zone NewZone
        inet 192.168.254.200 netmask ffffff00 broadcast 192.168.254.255

OI combines tools from the Linux world (ls, bash, gcc) and those native to Solaris (svcadm, pfexec, newtask, beadm). The CDDL licence (similar to GPL) made possible the emergence of the Nexenta distribution, which uses the OS/Net (OpenSolaris) core, much like OI. However, Nexenta has only GNU applications, and isn’t built on mixed technology like OI.

However, as we have already seen, this mixed environment is closer to Linux than to Solaris. The use of interesting options like ZFS and zones lets you look differently at existing solutions in Linux — for instance, LVM and lxc. Who knows, maybe one day, we’ll see a Linux distribution offering ZFS as the default root filesystem!

Once upon a time, there was Sun Microsystems. Not just an IT industry flagship, but also a legendary firm. Famous for SPARC processors, the Java programming language, and for the decades it spent developing its own UNIX OS, Solaris. Solaris’ successor is the OpenIndiana project.

Solaris and Sun

The history of Sun Microsystems is truly amazing. Started in 1982 to further develop the project born in Stanford University, the acronym (SUN) itself stands for Stanford University Network. Over time, the company was involved in many interesting projects. Slogans such as, “A computer is a network” and “Write Once, Run Everywhere” reflect fundamental milestones in the development of modern technologies that we are today so familiar with.

Commercially, Sun marketed SunOS, and later Solaris. Note that another of its achievements, the Berkeley Software Distribution (BSD), has been used not only in Solaris, but in a number of other commercial implementations like AIX/HP-UX, and open systems like FreeBSD/NetBSD/OpenBSD. To some extent, all BSD-derived systems (read, modern *NIX) and Linux share several fundamental concepts and features.

BSD is the root of many modern operating systems

In the first half of the last decade, Sun was too slow in turning Solaris from a closed to an open source product, i.e., OpenSolaris. This lost the company the server segment in enterprise-level deployments. Some studies indicate that if Sun hadn’t been so conservative in the 2000s, it is possible that a large part of the server segment would now be OpenSolaris hosts, rather than Linux.

OpenSolaris took another blow when Sun, in a bad financial condition, was bought by Oracle in 2010. The latter can hardly be called an open source friendly company, so it wasn’t surprising when the OpenSolaris community decided to develop their own fork, independent of Oracle’s corporate games. This resulted in the 2010 announcement of OpenIndiana.

As mentioned, OpenIndiana is the Illumos core from OpenSolaris, which is being independently developed by Oracle, plus GNU user-land tools. It is worth mentioning that in 2007, Sun specially hired Ian Murdock (the Debian founder) for the development of “Project Indiana”, to transform Solaris into OpenSolaris, and combine the latter with a number of open source products, such as GNOME. In general, we can assume that Project Indiana has successfully grown into OpenIndiana.

Installation

Right, so let’s try OpenIndiana. I downloaded the Server ISO with the development release 151 and loaded it on a regular x86_64 machine with 2 GB of RAM (later increased to 4 GB). The server needs you to use the command-line via the console or SSH. However, there’s a desktop version as well — the “Live DVD desktop ISO” for those who love GUIs: a choice of GNOME 2.30/XFCE 4.8/LXDE 0.5/KDE 4.7. I burned the ISO on a DVD and booted it, chose the Installer menu option “Install OpenIndiana” and began.

OpenIndiana has a pretty simple installer

Though the OS needs only about 3 GB, I allocated the entire HDD (22 GB). The rest of the allocated space will be used by /home and swap partitions. Please remember, the swap space consists of three different items!

Disk selection for installation

I set the hostname to “openindiana”, set a password for root, and created a single user for myself (“anton”). Network configuration was left as automatic; you can later change settings after logging in. Next, I hit F2 (Install) and waited for the installation to complete. After reviewing the installation log, I chose option “5 Reboot”, to try my very first steps with OpenIndiana!

Installation Log

OpenIndiana reminds you of the old UNIX console

Post logging in, you can configure a static IP address. First disable nwam (NetWork Auto Magic):

# svcadm disable network/physical:nwam

Add a name server (DNS) IP to /etc/resolv.conf, and add your current gateway to /etc/defaultrouter. Also, manually add the gateway to the routing table (necessary only for the current session) with route -p add default GATEWAY. Assign the static IP address with ifconfig, and start the network service:

# svcadm enable network/physical:default

Now, when you restart you workstation/server, this static network configuration will be applied.

Package management

The first thing that comes to mind is: how does one install extra packages.

Personally, I like Midnight Commander — a simple console file manager. To install it, I relied on the pkg package manager. As root (sudo bash), check if you have registered repositories:

# pkg publisher
PUBLISHER                             TYPE     STATUS   URI
openindiana.org          (preferred)  origin   online   http://pkg.openindiana.org/dev-il/

Yes, we can install new packages. But before that, let’s do a system update:

root@openindiana:/home/anton# pkg image-update
Creating Plan
Packages to update:   332
Create boot environment:   Yes
DOWNLOAD                                  PKGS       FILES    XFER (MB)
system/extended-system-utilities        19/332     57/2077    0.4/103.9
library/libtecla                        39/332    121/2077    0.6/103.9
system/network/ppp/pppdump              65/332    231/2077    2.3/103.9
system/library                          75/332    302/2077    4.0/103.9
system/kernel                          326/332   2063/2077  102.2/103.9
Completed                              332/332   2077/2077  103.9/103.9

PHASE                                        ACTIONS
Removal Phase                                120/120
Install Phase                                114/114
Update Phase                               5868/5868

PHASE                                          ITEMS
Package State Update Phase                   664/664
Package Cache Update Phase                   332/332
Image State Update Phase                         2/2

A clone of openindiana exists and has been updated and activated.
On the next boot the Boot Environment openindiana-1 will be mounted on '/'.
Reboot when ready to switch to this updated BE.

If you want, you can install the mc package (Midnight Commander), after checking if it’s in the repository:

# pkg search mc
INDEX      ACTION VALUE                   PACKAGE
basename   dir    usr/share/doc/mc        pkg:/file/mc@4.7.3-0.151
basename   dir    usr/share/mc            pkg:/file/mc@4.7.3-0.151
basename   file   usr/bin/mc              pkg:/file/mc@4.7.3-0.151
pkg.fmri   set    openindiana.org/file/mc pkg:/file/mc@4.7.3-0.151

root@openindiana:/home/anton# pkg install mc
Creating Plan |
Packages to install:     2
Create boot environment:    No
DOWNLOAD                                  PKGS       FILES    XFER (MB)
runtime/python-24                          0/2    294/2271     1.1/10.6
file/mc                                    1/2   2119/2271     9.3/10.6
Completed                                  2/2   2271/2271    10.6/10.6

PHASE                                        ACTIONS
Install Phase                              2457/2457

PHASE                                          ITEMS
Package State Update Phase                       2/2
Image State Update Phase                         2/2

The Midnight Commander (mc) file manager

For a package list, run pkg list; for more verbose output, run pkg info. Searching for, and attempting to install Apache yields the message “No updates necessary for this image”, which means it was already installed in the initial deployment. To check if a package is installed, use the following command:

root@openindiana:/home/anton# pkg list | grep apache
web/server/apache-22                          2.2.16-0.151    installed  -----

Let’s try to install the nginx Web server, which isn’t in the OpenIndiana repository, by downloading the source and compiling it. Create a src folder, set your HTTP proxy if required, and use wget to download the tarball. However, no GCC compiler is installed. What’s the compiler package name?

root@openindiana:/home/anton#  pkg search gcc | grep set
description  set  GCC Runtime libraries   pkg:/system/library/gcc-3-runtime@3.4.3-0.151
pkg.description set  GCC Runtime - Shared libraries used by gcc and other gnu components pkg:/system/library/gcc-3-runtime@3.4.3-0.151
pkg.summary  set  GCC Runtime libraries   pkg:/system/library/gcc-3-runtime@3.4.3-0.151
description  set   gcc - The GNU C compiler  pkg:/developer/gcc-3@3.4.3-0.151
pkg.description set   GCC Runtime - Shared libraries used by gcc and other gnu components pkg:/system/library/gcc-3-runtime@3.4.3-0.151
pkg.summary  set  gcc - The GNU C compiler   pkg:/developer/gcc-3@3.4.3-0.151

The package we need is called “pkg:/developer/gcc-3 @ 3.4.3-0.151″, so we can install it with the following command:

root@openindiana:/home/anton# pkg install pkg:/developer/gcc-3@3.4.3-0.151

Next, compile nginx with the thoroughly familiar ./configure –prefix=/opt/nginx-1.1.0 and make recipe. We can omit the output as unnecessary.

As a hosting platform, OpenIndiana could be surprisingly good. Besides Web servers, you can also find PHP, Python, Ruby, Java and several other packages in the repository, which can be configured or installed with minimum effort. However, some packages need extra care, like PostgreSQL. Let’s see what happens when we install it:

root@openindiana:/home/anton# pkg install pkg:/service/database/postgres-84@8.4.4-0.151
               Packages to install:     2
           Create boot environment:     No
               Services to restart:     1
DOWNLOAD                                PKGS       FILES       XFER (MB)
Completed                               2/2        211/211     9.6/9.6

PHASE                                   ACTIONS
Install Phase                           1/352 The Boot Environment openindiana-2 failed to be updated.
A snapshot was taken before the failed attempt and is mounted here /tmp/tmpfWSs0u. Use 'beadm
unmount openindiana-3' and then 'beadm activate openindiana-3' if you wish to boot to this BE.

pkg: The requested operation cannot be completed due to invalid package metadata.  Details follow:

The action 'dir group=postgres mode=0755 owner=postgres path=var/postgres' in package
'pkg://openindiana.org/service/database/postgres-84@8.4.4,5.11-0.151:20110523T151027Z' has invalid
attribute(s):
  '//var/postgres' cannot be installed; 'postgres' is an unknown or invalid user.

To install PostgreSQL, you need to have created both a user and group named “postgres”. Otherwise, the package simply won’t be able to assign the right permissions to the newly created /var/postgres.

Note the interesting point: the package manager always lets you not just remove the package, but even roll it back, in case it was installed improperly. Because of the underlying ZFS, the package is originally installed into a temporary location, and only after successful installation is the temporary location integrated with the main filesystem, known also as BE (Boot Environment). In the example above, the package has been installed into the temporary location “openindiana-3″.

root@openindiana:/home/anton# beadm list
BE            Active Mountpoint Space Policy Created
openindiana   -      -          5.92M static 2011-07-01 16:11
openindiana-1 -      -          448M  static 2011-08-05 12:10
openindiana-2 NR     /          2.39G static 2011-08-05 12:20
openindiana-3 -      -          56.0K static 2011-08-05 12:45

You can force the integration process, so an unsuccessful package installation would be joined with the main filesystem with something like: beadm activate openindiana-3. However, it’s highly recommended to have proper installation without errors. Create the “postgres” user and group, and then reinstall PostgreSQL.

Service control

It’s time to learn how to control services. The svcs command shows a list of all services. The svcadm command lets you start/stop the desired service. The svccfg and svcprop commands are intended for fine-tuning.

Here’s how to first check service status (disabled == not running), and then start a service:

root@openindiana:/home/anton# svcs apache22
STATE          STIME    FMRI
disabled       Aug_08   svc:/network/http:apache22

root@openindiana:/home/anton# svcadm enable apache22
root@openindiana:/home/anton# svcs apache22
STATE          STIME    FMRI
online         10:02:37 svc:/network/http:apache22

Filesystem organisation

Yes, I’m talking about the structure of ZFS.

root@openindiana:/home/anton# df -h
Filesystem            Size  Used Avail Use% Mounted on
rpool/ROOT/openindiana-2
15G  3.0G   12G  21% /
swap                  2.4G  368K  2.4G   1% /etc/svc/volatile
/usr/lib/libc/libc_hwcap1.so.1
15G  3.0G   12G  21% /lib/libc.so.1
swap                  2.4G   48K  2.4G   1% /tmp
swap                  2.4G   72K  2.4G   1% /var/run
rpool/export           12G   32K   12G   1% /export
rpool/export/home      12G   32K   12G   1% /export/home
rpool/export/home/anton
16G  4.4G   12G  29% /export/home/anton
rpool                  12G   46K   12G   1% /rpool
/dev/dsk/c2t0d0s2     549M  549M     0 100% /media/OpenIndiana_Text_X86
/export/home/anton     16G  4.4G   12G  29% /home/anton

You may recall that we allocated 22 GB of space to OI. However, the root system occupies just 15 GB (the first line). Where is the rest? Note that there are three swap partitions used, mounted at /var/tmp, /tmp, and purely for system usage as /etc/svc/volatile. These swaps “ate” ~7 GB! The rest of the space is dynamically reallocated for the home directory, mounted and seen as /export.

Please keep in mind that ZFS is extremely powerful, in terms of the used features. Briefly, what are these? First, you can see the ZFS sections that are on duty:

root@openindiana:/home/anton# zfs list
NAME                       USED  AVAIL  REFER  MOUNTPOINT
rpool                     10.4G  11.1G    46K  /rpool
rpool/ROOT                3.96G  11.1G    31K  legacy
rpool/ROOT/openindiana    5.92M  11.1G  1.70G  /
rpool/ROOT/openindiana-1   448M  11.1G  1.66G  /
rpool/ROOT/openindiana-2  3.51G  11.1G  2.92G  /
rpool/ROOT/openindiana-3    56K  11.1G  1.90G  /
rpool/ROOT/openindiana-4    91K  11.1G  1.90G  /
rpool/ROOT/openindiana-5    55K  11.1G  1.90G  /
rpool/dump                1.00G  11.1G  1.00G  -
rpool/export              4.36G  11.1G    32K  /export
rpool/export/home         4.36G  11.1G    32K  /export/home
rpool/export/home/anton   4.36G  11.1G  4.36G  /export/home/anton
rpool/swap                1.06G  12.1G   132M  -

Check the history of the changes made to the filesystem:

root@openindiana:/home/anton# zpool history
History for 'rpool':
2011-07-01.16:11:03 zpool create -f rpool c1d0s0
2011-07-01.16:11:04 zfs set org.openindiana.caiman:install=busy rpool
2011-07-01.16:11:07 zfs create -b 4096 -V 1024m rpool/swap
2011-07-01.16:11:10 zfs create -b 131072 -V 1024m rpool/dump
2011-07-01.16:11:32 zfs set mountpoint=/a/export rpool/export
2011-07-01.16:11:33 zfs set mountpoint=/a/export/home rpool/export/home
2011-07-01.16:11:37 zfs set mountpoint=/a/export/home/anton rpool/export/home/anton
2011-07-01.16:22:50 zpool set bootfs=rpool/ROOT/openindiana rpool
2011-07-01.16:24:40 zfs set org.openindiana.caiman:install=ready rpool
2011-07-01.16:24:41 zfs set mountpoint=/export/home/anton rpool/export/home/anton
2011-07-01.16:24:45 zfs set mountpoint=/export/home rpool/export/home
2011-07-01.16:24:49 zfs set mountpoint=/export rpool/export

View the current load on ZFS:

root@openindiana:/home/anton# zpool iostat
capacity     operations    bandwidth
pool        alloc   free   read  write   read  write
----------  -----  -----  -----  -----  -----  -----
rpool       9.45G  12.4G      0      0  1.39K  3.27K
----------  -----  -----  -----  -----  -----  -----

Check the integrity of the filesystem:

root@openindiana:/home/anton# zdb -c rpool

Traversing all blocks to verify metadata checksums and verify nothing leaked ...

        No leaks (block sum matches space maps exactly)

        bp count:          280189
        bp logical:    10321356800      avg:  36837
        bp physical:   10087483904      avg:  36002     compression:   1.02
        bp allocated:  10148575232      avg:  36220     compression:   1.02
        bp deduped:             0    ref>1:      0   deduplication:   1.00
        SPA allocated: 10148575232     used: 43.21%

You can prepare filesystem snapshots (or backups). Decide exactly what parts of the filesystem you want to copy (there’s no reason to snapshot the entire filesystem, especially if the critical parts are stored only in /export/home/anton).

# zfs snapshot -r rpool/export/home/anton@snap4

Later, you will be able to recover it with:

# zfs rollback rpool/export/home/anton@snap4

If a rollback isn’t advisable, but you want to retrieve lost data, you can mount this snapshot at any directory, and copy the required data out:

# zfs clone -o mountpoint=/tmp/snapshot/ rpool/export/home/anton@snap4 rpool/export/home/anton_bak

The possible operations with ZFS, as well as ZFS itself are very promising. It combines backup/restore features, data compression on the fly, LVM manager, etc.

It looks like the Illumos Foundation, responsible for current and future OpenIndiana development, has every chance of bringing, not a rival, but rather, a competitive technology into the Linux world.

In a subsequent article in this series, we will discuss the virtualisation, disk and zone management aspects of OpenIndiana.

It is quite well-known that the two behemoths among free software operating systems, the ones that both represent and impact GNU/Linux, are Red Hat’s Fedora and Ubuntu from Canonical. Ubuntu’s latest version, 11.04, also known as the Natty Narwhal, was released on April 28, and a review of this version appeared in the June. Fedora followed on May 24 with version 15, codenamed Lovelock. Now, when you have two such similar releases at practically the same time, it creates rather a “Big Choice”, and it might even create a “Big Dilemma”. So, how do the two stack up against each other?

Delving into the core

Examining the core first, we find that the foundations of both are almost the same. The kernel is at version 2.6.38 in Natty, with Lovelock’s just a few point releases ahead. Kernel version 2.6.38 brings a lot of desktop-oriented improvements, resulting in much better desktop performance than was previously the norm.

Both distributions have the xorg-server at 1.10.1, udev version 167 and freetype version 2.4.4. However, while Ubuntu has already been using GRUB 2 at version 1.99rc1, Fedora staunchly continues with GRUB 1 at version 0.97. Whichever version you use, this aspect does not create much of a hassle — just the method of configuring GRUB differs in the two versions. Another area of difference is GCC. Lovelock includes version 4.6.0 of the compiler collection, while Natty includes 4.5.2 — so developers who want that wee bit more, may be happier with Lovelock.

With the preliminaries over, let’s get around to the more visible aspects and look at the giants from the outside!

The desktop environments

We will not compare KDE on the two distros. The first reason is that, both of them include version 4.6.2 of KDE; second, KDE is not the default environment of either distribution; and third, KDE 4.6 is not as radically different from 4.5 as the default environments of Lovelock and Natty are from their previous counterparts.

Lovelock and GNOME 3

Lovelock is the first major distribution to include GNOME 3 as the default desktop environment (see Figure 1) and GNOME 3 is… well, there’s a lot to say about GNOME 3. Even if you are not biased, you have to admit that GNOME 3 makes almost everything harder than it was previously.

Figure 1: GNOME 3 desktop

A major issue with software developers is their desire to reinvent everything from scratch. This is not essentially a bad thing, and GNOME up to its major version 2, GNOME was definitely plagued by legacy code such as CORBA. However, the problem with the developers and maintainers of GNOME is that they prioritise simplicity to the extent of bordering on the simplistic. This is exactly the case with GNOME 3.

The default dual-panel setup of GNOME has been replaced by a single panel at the top, and a translucent, hidden notification area at the bottom. The top panel displays the clock in the centre. Clicking on the clock opens up an enhanced version of the standard calendar, which allows managing your events from the calendar itself (Figure 2).

Figure 2: GNOME 3 calendar

The top-right portion of the panel contains the usual system management icons, including the Network Manager, the interface of which has changed somewhat in Lovelock. The networking menu displays a simplified version of the available networks, each with a representation of a toggle switch to the right (Figure 3).

Figure 3: GNOME 3 network menu

Simply clicking a network’s entry turns it on or off. The network settings and systems settings windows have also changed (Figures 4 and 5, respectively); the latter was named Control Centre in previous versions of GNOME. The system settings window appears similar to the KDE one, but some important features have been removed from the user interface. One is that you can no longer customise the desktop theme using the built-in tools as easily as you could in previous versions. In fact, the default tools don’t have any option to change the theme at all! Another UI change is that now there are no buttons present to minimise or maximise windows. To do that, either right-click the title bar and select the appropriate option, or use the keyboard to achieve the same effect—both much more cumbersome than the established norm.

Figure 4: GNOME 3 network settings

Figure 5: GNOME 3 system settings

The top panel’s default menu has also been replaced by an Activities entry. You can either click on it, or point your mouse to the top-left corner of the screen. This opens up a screen with bookmarked applications on the left and those that are currently running in the middle, along with an Applications tab that lists all installed apps. However, the convenience of having the applications listed in groups (Accessories, Internet, etc.) is gone. The right side of the Activities screen shows the currently available virtual desktops. There is always one more than the number you are using, so you can keep creating as many as you like.

GNOME 3 is new software, and, just as was the case with KDE 4, this first version still has issues that need to be sorted out. For example, Figure 6 shows the Activities screen along with an error notification at the bottom. This particular error is not fatal, but it keeps recurring, and diminishes one’s confidence in the software. Notifications in GNOME 3 are unobtrusive, and can be pulled up at a later time, as shown in Figure 7. The Activities screen also allows you to search for applications and files (Figure 8).

Figure 6: Activities — multitasking — error

Figure 7: GNOME 3 ABRT

Figure 8: GNOME 3 search

Another deviation from the norm is the user menu in the top-right corner of the screen (Figure 9). The restart and shutdown options are now gone, and only ‘Log Out’ and ‘Suspend’ remain. To restart or power off the computer, either press and hold the Alt key while the menu is open, which replaces the ‘Suspend’ item with an option to power off or restart — or log out and then proceed to shutdown from the login screen.

Figure 9: GNOME 3 user system menu

If you have more than one window open in GNOME 3, to switch between them, you either press Alt+Tab or open the Activities screen. The panel no longer has a task-bar to list open windows.

So it seems that GNOME 3 will take some time getting used to. Let’s see how Natty fares.

Natty and Unity

The developers at Canonical have not adopted GNOME 3 for Natty, although they will for its successor, the Oneiric Ocelot. Ubuntu already had a netbook interface, which was somewhat similar to Natty’s new interface, Unity — Canonical wanted to create a uniform look-and-feel across its desktop and netbook versions.

Also, the folks at Canonical were not entirely in favour of all the proposed changes in GNOME 3. Whatever be the reason, the decision to not use GNOME 3 in Natty has resulted in a much more stable environment than the one in Lovelock. So, what is Unity and what does it bring to the table?

Unity is just a shell running on top of GNOME. In Natty, therefore, you can select the Ubuntu Classic session from the login screen, if you want the default GNOME 2 interface instead of Unity. Figure 10 shows a customised Unity desktop on Natty.

Figure 10: The Unity desktop shell in Natty

Like GNOME 3, Unity also uses a single top panel, but there is no notification area at the bottom. Notifications appear in the top-right portion of the screen, as in previous versions of Ubuntu. Additionally, the top panel now serves as an integrated menu bar for almost all applications, similar to the way the Mac OSX UI works. This provides maximum vertical screen space (Figure 11). However, the absence of a taskbar even here means that to switch between open windows, you either press Alt+Tab or use the launcher.

Figure 11: A maximised window in Unity merges nicely with the top panel

The launcher is a bar on the left side of the Unity screen that holds bookmarked applications and some lenses. Lenses are screens that show searchable lists of your applications, files, results on the Web, etc. However, unlike the GNOME 3 Applications list, you can view the categorised lists of your installed applications. Also, the applications lens shows random selections of the apps available for installation from Natty’s configured software repositories, making it a useful place to discover new apps.

When you open a window, the launcher hides itself if the window needs its space, just as in the case of a maximised window. To get the launcher back, either hold your pointer at the left edge of the screen, or press and hold the Super (Windows) key on the keyboard. The latter action also causes the launcher to display numbers and letters on top of its icons. You can press a number or letter to open the corresponding window/application.

Also, right-clicking an icon on the launcher opens a context menu with useful actions. You can remove any application from the launcher, and you can also drag items in the launcher to re-position them. All of this is much more than what the default GNOME 3 panel or Activities screen does.

In the top-left corner of the screen is the icon for the Dash, a feature similar to the applications lens, which allows you to search for or browse installed applications. The shortcut key for the Dash is the OS (Windows) logo key.

As for the rest, since Unity runs atop GNOME, the basic interface is that provided by GNOME itself.

The outer reaches

We now come to the outermost aspect of our giants — the environment they provide for actually getting work (or play) done on the computer. The most famous component of this environment is probably the office suite. Both Lovelock and Natty include version 3.3.2 of LibreOffice, the new fork of the popular OpenOffice.org suite of applications.

LibreOffice is owned and operated by the community, so there is no concern of it being monopolised by any single entity. Also, the new suite has the crucial support of several big companies. It will be best if OpenOffice.org can be merged into LibreOffice, thus liberating it from Oracle, but as of now, most major GNU/Linux distributions are switching over to LibreOffice.

Apart from this, the rest of the software is also pretty standard and up to date. What is not available in the live or installation discs can, of course, be installed from the repositories. The installers are, as usual, radically different in the two distros. The venerable Synaptic Package Manager is available in Natty, just as it should be.

However, the new software manager, Ubuntu Software Centre, has seen some improvements. It now allows for purchasing software directly through its interface (although a bug prevented me from actually going through with a purchase at one go), and it also now shows libraries and other non-GUI software in search results, on demand, apart from providing some recommendations based on your searches and installations.

The standard PackageKit installer in Lovelock, on the other hand, seems pretty bare-bones in comparison. It allows simple configuration of the software sources, but is nowhere near Synaptic in power. However, YUM, the installation back-end on Lovelock, scores an extremely important point over the Debian Apt in Natty — it allows delta packages to be downloaded. A delta is just the code that is different in a new version of a package that is already installed. Downloading deltas instead of entire packages saves huge amounts of time and bandwidth for data transfers during upgrades.

However, yet another factor negatively affecting Lovelock is that Fedora does not have integrated repositories for proprietary software. It depends on the external RPM Fusion repositories, and they may not always follow the latest Fedora release on time. For example, RPM Fusion did not have the proprietary drivers for the Broadcom BCM4312 Wi-Fi chip for the current Lovelock kernel even several days after the Lovelock release (they are available now, though there is no guarantee that they will be available on time in case of a kernel upgrade).

Ubuntu, on the other hand, has always had integrated proprietary software repositories. They are not supported by Canonical, of course, but they do track the latest Ubuntu release correctly and Natty is no exception.

The verdict

Instead of beating about the bush, it is better to admit that, as of now, Natty is much more usable than Lovelock. However, there are some points that must still be taken into consideration. First of all, the future of GNOME is GNOME 3. Even the next version of Ubuntu will have GNOME 3 as its base for Unity. So, whether you like it or not, if you want to stay with GNOME, you will have to consider GNOME 3 sometime or the other.

Thus, you may want to delve into it right away, and report whatever bugs and missing features you encounter, so that the software matures more quickly; after all, free software is dependent not on corporations, but on users. And besides, KDE version 4.6 also has some serious bugs (such as PolicyKit randomly crashing at the end of a session), which have not yet been addressed by the developers.

Also, Fedora is known for its leadership and innovation in the GNU/Linux field. If you want to stay absolutely current on the Free Software front, Fedora is the way to go. Lovelock, for example, has a dynamically configurable firewall, which means that you do not have to stop and restart the firewall to effect some changes to its rules.

You also should keep in mind that if you choose Fedora, you will not just start up to date, but you will also stay up to date. For example, a Fedora release will, as long as it is supported, keep getting the latest versions of all its software, including the kernel and desktop environment, which is not the case with Ubuntu.

The final decision is, of course, up to you. Both Lovelock and Natty are critical releases that pave the way for the future of the GNU/Linux desktop. Whether you are comfortable with handling the latest stuff, with the occasional trip-up, or want the relative safety of a traditional environment, the choice is yours.

Whichever you choose, you will be a participant in one of the most important revolutions in the free software world, instead of being a mere spectator.

In Natty, Unity is the default desktop atop GNOME; it requires a 3D graphics accelerator, as it uses Compiz Fusion heavily — however, you can switch to GNOME Classic from the login menu, if you don’t have the right hardware. Ubuntu 11.10 will be shipped with a 2D version of Unity, for systems with average graphics cards.

Natty replaces Ubuntu Netbook Edition for all PCs and netbooks. Available in the usual variants — Kubuntu (KDE), Xubuntu (XFCE), Edubuntu and more — the software defaults have also changed. LibreOffice 3.3.2 is now the default office suite; Firefox 4 is the default browser; and Banshee replaces the Rhythmbox music player.

The Ubiquity installer has ‘graduated’

The Ubiquity installer is getting much smarter and understandable with every incremental release. People new to Linux (who fear messing up their existing OS while doing a dual-boot installation), and those who don’t understand what swap space is, or how much they need of it, will like Ubiquity.

Ubiquity installer

This installer is quite impressive; it guides you at every step, letting you know what’s happening, what you might want to do, and how it can be done. It detects whether you are installing on a system with an existing Windows installation, or upgrading from an earlier Ubuntu install, etc. It also has an expert partitioning option for experienced Linux users.

Once you enter the required choices, the installer begins copying files in the background, while you fill in additional information like the time zone, user details and more. The migration assistant, too, works flawlessly, and migrates your documents, pictures, user settings and so on without any hassle. You can also choose to install third-party software like Flash, MP3 codecs, Java, etc.

Install third-party stuff right from the installer

Installation is not much speedier. Boot time from a live USB was less than a minute on a Core2Duo laptop, and two minutes on my netbook.

The Unity bar is here to stay

Like it or hate it, the Unity bar is clearly here to stay. It’s had a major face-lift since 10.10, though it still sits atop GDE. It’s not much different once you get past the layout and some basic controls. The new Unity interface is very innovative and neat, and saves screen real estate. It has a dock-style bar on the left with common applications, and a standard panel across the top. It auto-hides when you maximise windows, and comes back when you hover the mouse in the upper-right corner. The Unity bar scrolls icons when you hover the mouse pointer over the icons stacked like a deck of cards in the lower-right corner.

Click app icons to start them. A caret sign appears to the left of the icons for apps that are already running; a caret on the right of the icon shows the currently active app. Shift-clicking an icon launches a new instance of an already-running app; tiny lights show multiple instances. You can drag-and-drop to rearrange icons, and remove an icon by dragging it off the bar. To quit an app, right-click and select Quit from the menu.

The Ubuntu button (top-left corner of the screen) gives you an iPad-like home screen, with high-resolution icons for your favourite destinations. The Workspace Switcher feature, accessed via a sidebar icon, defaults to four virtual desktops.

Workspace switcher

Mac-like app menu bar

The Ubuntu 10.10 netbook edition missed the very important unified menu bar. Yes, Natty has made some Apple-like changes — the top bar serves as a universal app menu bar, displaying the application’s menu bar (when they have one) here, instead of in the app window. Natty thus provides users the maximum possible vertical screen space.

Universal menu bar

Unity’s keyboard navigation

Another addition with Unity is navigation for keyboard shortcuts, most of which use the Windows key. Holding down the Windows key shows numbers for each dock item; launch one by pressing the corresponding number. Tap the Windows key for a quick-search box from the upper-left corner, which lets you launch applications, search files and folders, and access system settings and utilities. For a wallpaper with a complete keyboard shortcuts listing, visit askubuntu.com.

Quick search bar

Updated software defaults

The Linux kernel is now the much more stable version 2.6.38.2, so users get improved overall system performance. The open source Banshee 2.0 Music Player, which replaces Rhythmbox, was originally named Sonance, until 2005. It uses the Helix and GStreamer platforms for handling media formats. It supports bookmarks, the Amazon MP3 store, video support, audio books support for libraries, metadata fix-up, integration with the sound menu, and a few more features.

Natty uses LibreOffice 3.3.2, developed by The Document Foundation, which is a fork of the well-known OpenOffice.org. With the same features as OpenOffice.org, it will be easy for users to switch to LibreOffice. Firefox 4 débuts as the default browser in 11.04. With an improved user interface, “door-hanger” notifications, a redesigned extension manager and Firefox Sync, users now have a modern browser.

The updated Software Centre, Ubuntu’s version of an app store, is already well-liked by users since the release of Ubuntu 10.10. The Ubuntu One control panel now supports selective file synchronising, and the synchronisation itself is improved. The launcher icon also displays file sync progress. In the visual improvements department, scrollbars are now “zero-size”, the icons much better, and the overall feel one of an iOS device.

Software Centre

Things to do after installing

First, of course, install all recommended updates using the Update Manager and restart the machine.
For a pleasant Unity experience, hardware drivers must be updated, since Unity demands 3D acceleration. If Unity isn’t working, run the Additional Drivers program and activate the driver. If you have an nVidia graphics card, then you have to install it using a PPA, as follows:

sudo add-apt-repository ppa:ubuntu-x-swat/x-updates
sudo apt-get update
sudo apt-get install nvidia-current nvidia-current-modaliases nvidia-settings

When you restart your system, you should be greeted with the new Unity interface.

Install Ubuntu Tweak, which enables customising Ubuntu the way you like it, from the Software Centre, or through a PPA, as follows:

sudo add-apt-repository ppa:tualatrix/ppa
sudo apt-get update
sudo apt-get install ubuntu-tweak

Basic applications you should install

• GIMP
• VLC
• Chromium browser
• Multiget, a very nice and easy direct-downloads manager for Ubuntu
• Bleachbit, a system cleaner that cleans out all useless packages, freeing space on the root filesystem for better performance.

If there is something you want that is not available through Software Centre, you can still download it from the Web in .deb format, and open it with /usr/bin/gdebi-gtk (you must have gdebi and gdebi-core installed). A package installer window appears; hit “Install”, and it’s done.

I enjoyed the new improvements a lot. I suggest you take the plunge and see what’s available for you in Ubuntu Linux 11.04 “Natty Narwhal”.

My first acquaintance with this distribution was quite unusual. The thing that caught my eye was an artistic graphical design of BarPanel — the desktop panel, which is written in Python, and can be used with any Linux distribution. Being curious, I tried to figure out what other products are made in a similar manner, and by a similar developer.

As it turned out, there were other software made to be used by Igelle DSV: Melas, a build and packing tool; the Ige Package Manager, and Igelle Connector. This distribution has been declared as a general-purpose operating system for desktop computers, laptops, netbooks and servers. I’ve spent some time and explored it with my work laptop, as well as with a desktop PC, and even with my MIPS netbook. It works almost flawlessly.

Okay, let’s proceed to a real experience. Let’s take the ISO image from Igelle’s website, burn it to a CD-ROM and try to boot any x86 machine.

First look

The initial bootstrapping is as usual — first the Linux kernel with an initramfs image loads into memory, and then the hardware is initialised. After a minute, you’ll get a working desktop based on LXDE. Visually, it looks like a contemporary distribution (Figure 1). However, let’s dig a little bit deeper, and figure out how it really works.

Figure 1: First look at Igelle DSV at LiveCD mode

Figure 2: The distribution can be installed from LiveCD mode in a minute

I’ve installed Igelle DSV into a virtual machine first, and allocated 1 GB of virtual hard disk space for it, just to see how it works. Don’t be surprised that the root partition appears so small. In fact, this is quite a large space, and Igelle can live with it quite happily, despite it being a full installation:

$ df -h
Filesystem            Size  Used Avail Use% Mounted on
rootfs               1008M  620M  338M  65% /
udev                  506M  1.1M  505M   1% /dev
none                 1008M  620M  338M  65% /
/dev/sda1            1008M  620M  338M  65% /igelleroot
/var/run              506M   40K  506M   1% /var/run
/var/lock             506M     0  506M   0% /var/lock
/dev/shm              506M   88K  506M   1% /dev/shm
none                 1008M  620M  338M  65% /var/spool/cups

The df -h command shows that the /dev/sda1 partition is mounted at /igelleroot, where the GRUB loader stores its components. There is another subdirectory that hides here — igelle/1.0.0. This contains a big SQUASHFS image file, together with the Linux kernel for this Igelle release, and an initramfs disk as well. This concept isn’t quite new, and it is used among many popular Live CDs, like Knoppix.

anton@DSV_100:~$ ls -l /igelleroot/
total 24
drwxr-xr-x 3 root root  4096 Apr 15 08:49 boot/
drwxr-xr-x 4 root root  4096 Apr 15 08:57 igelle/
drwx------ 2 root root 16384 Apr 15 08:49 lost+found/

anton@DSV_100:~$ ls -l /igelleroot/igelle/1.0.0/
total 600888
-rw-r--r--  1 root root  14751413 Feb 18  2010 initramfs
drwxr-xr-x 13 root root      4096 Apr 15 10:16 overlay/
-rwx------  1 root root 598695936 Feb 18  2010 rootfs*
-rw-r--r--  1 root root   1848784 Feb 18  2010 vmlinuz

It’s possible to have multiple versions of Igelle DSV stored on disk. Once the root filesystem (the file itself is named rootfs) is mounted in read-only mode, an additional overlay subdirectory is being mounted as an AUFS device, where all the changes to / will be written. Igelle’s root filesystem is thus actually a mix of read-only SQUASHFS images, and the AUFS overlay directory:

rootfs on / type rootfs (rw)
udev on /dev type tmpfs (rw,relatime,mode=755)
/sys on /sys type sysfs (rw,relatime)
/proc on /proc type proc (rw,relatime)
none on / type aufs (rw,relatime,si=8da72793)
/dev/sda1 on /igelleroot type ext4 (rw,relatime,barrier=1,data=ordered)
/var/run on /var/run type tmpfs (rw,nosuid,nodev,noexec,relatime,mode=755)
/var/lock on /var/lock type tmpfs (rw,nosuid,nodev,noexec,relatime)
/dev/shm on /dev/shm type tmpfs (rw,relatime)
/dev/pts on /dev/pts type devpts (rw,relatime,gid=5,mode=620)
none on /proc/bus/usb type usbfs (rw,relatime)
none on /var/spool/cups type aufs (rw,relatime,si=8da72793)
gvfs-fuse-daemon on /home/anton/.gvfs type fuse.gvfs-fuse-daemon (rw,nosuid,nodev,relatime,user_id=1000,group_id=1000)

Why is this approach (using an overlay and hybrid storage) so good? First, you’re left with an intact and consistent root filesystem. All changes made in /usr, /var, /etc, are actually stored separately. Remove the overlay, and the system returns to its original state, with default settings.

Second, you can simultaneously have multiple versions of Igelle distributions on the same disk partition, because settings for each release are stored in individual folders. For example, version 1.0.0 is located at the absolute location /igelleroot/igelle/1.0.0, while the next version will have its settings stored in /igelleroot/igelle/2.0.0. Moreover, you can place multiple Igelle versions along with any preinstalled Linux distro — there’s no need to remove it, or repartition. Just don’t forget to add a line to the GRUB loader configuration, /igelleroot/boot/grub/grub.cfg.

Figure 3: Igelle’s architecture

Configuration

The first thing every novice tries to do with a new environment is to set it up in the most comfortable way. Of course, the widget that catches your eye is a translucent panel at the bottom of the desktop — a la Apple Macintosh. This is an Esther-Panel — a part of the Igelle project (Figure 4).

Figure 4: The Esther-panel application resembles Apple’s Macintosh quick- launch

In order to add a new entry to Esther-Panel (a.k.a. Quick-Launch), just run LXTerminal (Igelle –> Applications –> Command line) and make a symbolic link from the necessary GNOME application profile to the ~/.local/share/favorite-applications directory:

$ ln -s /usr/share/applications/brasero.desktop ~/.local/share/favorite-applications/

The panel must be restarted after this, with something like killall esther-panel. After that, the new application is added to the panel.

By default, all programs are displayed in English. However, sometimes you may need national language support. To have localisation applied to the desktop (for example, the French language), simply run the following commands:

# mkdir /usr/lib/locale
# localedef -ci fr_FR -f UTF-8 fr

When you launch the calculator application with the LANG environment variable set to “fr” (LANG=fr galculator), you’ll see its menus in French. However, to make the full desktop appear in French, by default, you should add the following line into ~/.config/esther-session.config:

esther-session-control setenv LANG fr

All GNOME applications used by the system expect to have localisation files for each language. If you correctly configured your language, but an application is still displayed in English, then this means that your particular application has no translation file yet; this job still needs to be done by the translation team.

What about input in a local language? That’s easy — place the following command into a startup script (for example, .bashrc in your home directory), and you can now switch between US and Arabic keyboards by pressing both Shift keys:

setxkbmap -option grp:switch,grp:shift_toggle,grp_led:scroll us,ar

Self-contained packages

What is a self-contained package, and why is it needed? Imagine that you have a package called MyEditor. It keeps all files (localisation files, dynamic libraries, icons and so forth) in one specific location — say, /opt/MyEditor. Is it comfortable? Very! Besides, you have an opportunity to install and uninstall this package from the system in the blink of an eye, just like the mount/umount commands, but in this case, using the sjdctl command.

By default, the Igelle system has bleeding-edge software only, like the latest X11 libraries (and those that developers require), GNOME and QT, and the GCC compiler — that’s why when you run Igelle, you can compile any application from source. For example, in Live CD mode, I was able to compile MPlayer, rdesktop and vncviewer, from scratch. This definitely helped me a lot when I faced an emergency situation.

To view a full list of packages that are preinstalled in the rootfs, use: Igelle –> About this system –> Packages.

The CD image’s contents are limited by size (650 MB) and unfortunately, some much-required applications like Kino, AbiWord or VLC weren’t included. However, you can find them in a special repository. All you need is to download files with the sjapp extension, and install it. Alternatively, you can download a specification file only (with an sjs extension), which describes how the package should be built, and create a binary package by yourself.

A specific package that can’t be found even in this repo has to be built manually. In one particular case, I preferred using a file manager like Midnight Commander. I decided to compile the package for it first, and place it in the /opt/mc directory (exactly where an mc binary package would be installed). In order to compile the binary package, you might need to be familiar with the meta-compiler syntax, and how to correctly edit a package description. Let’s look at the mc-4.7.1.sjs file:

mc: sjapp {
        setself "url" "http://www.midnight-commander.org/downloads/35"
        setself "version" "4.7.1"
configure() {
      ./configure --with-screen=ncurses --prefix=/opt/mc --datarootdir=/opt/mc/share --localedir=/opt/mc/locale --libexecdir=/opt/mc/libexec --sysconfdir=/opt/mc/etc/
}
compile() {
      make
}

install() {
                rm -rf "`getself staging`"
                mkdir -p "`getself staging`/`getself prefix`"
                make DESTDIR="`getself staging`" install

                mkdir -p "`getself staging`/`getself prefix`/share/applications";
                cat > "`getself staging`/`getself prefix`/share/applications/mc.desktop" <<EOF
                        [Desktop Entry]
                        Name=Midnight Commander
                        Exec=lxterminal -e /opt/mc/bin/mc
                        Icon=/usr/share/icons/gnome/32x32/devices/gtk-floppy.png
                        Terminal=true
                        Type=Application
                EOF
}
}

Here’s a brief comment on the functions and variables used. At the beginning, you need to declare the metapackage name (mc: sjapp {... here comes the body ... }). Then declare a version for this new package (setself version 4.7.1) and the location of the source code — (the statement setself url). After that, define the functions configure(), compile(), and install(), which as the names imply, will be transferred to the make program.

After editing the .sjs file, you need to run a meta-compiler named sjappc (the command is sjappc mc-4.7.1.sjs). After a while, the sjappc meta-compiler generates a binary package for Midnight Commander, invoking the GCC compiler to compile the application. The binary package is, in fact, a SQUASHFS file, which you can safely install into the system by issuing the following command:

# sjdctl install-app mc.sjapp

You can now run Midnight Commander directly, through LXTerminal, specifying the full path to the binary, /opt/mc/bin/mc. Or you can click on the icon with the floppy disk, titled Midnight Commander.

Figure 5: Midnight Commander compiled and deployed within Igelle DSV

A more complicated SJS description file would be for the Opera Web browser. Since this browser is not an open source product, you need to remove the compile() and configure() functions. On the other hand, the install() function has now become longer:

opera : sjapp {
        setself "url" "http://snapshot.opera.com/unix/labs-6177/opera-10.50-6177.linux.i386.tar.gz"
        setself "version" "10.50"

        install() {
                rm -rf "`getself staging`"
                mkdir -p "`getself staging`/`getself prefix`"

                cp -R "`getself workdir`"/usr/* "`getself staging`/`getself prefix`"

                mkdir -p "`getself staging`/`getself prefix`/bin";
                cat "`getself workdir`/opera" | \
                        sed s@^OPERA_BINARYDIR=usr/lib/opera@OPERA_BINARYDIR=/opt/opera/lib/opera@g | \
                        sed s@OPERA_DIR=\"\$PWD/usr/share/opera\"@OPERA_DIR=/opt/opera/share/opera@g > \
                        "`getself staging`/`getself prefix`/bin/opera"
                cp "`getself staging`/`getself prefix`/bin/opera" /home/ab
                chmod a+x "`getself staging`/`getself prefix`/bin/opera";

                mkdir -p "`getself staging`/`getself prefix`/share/applications";
                cat > "`getself staging`/`getself prefix`/share/applications/opera.desktop" <<EOF
                        [Desktop Entry]
                        Name=Opera 10.50
                        Exec="`getself prefix`/bin/opera"
                        Icon=/opt/opera/share/icons/hicolor/32x32/apps/opera.png
                        Terminal=false
                        Type=Application
                        Categories=Application;Game;
                EOF
        }
}

After a meta-compilation process, mount the package, and then Opera is ready to be used!

A little more patience is required to build the Chromium browser, because it needs three to four more auxiliary libraries that are used only by Chromium. However, nothing is impossible! See Figure 6 for more details.

Figure 6: Open-source browser Chromium runs smoothly in Igelle DSV

Future versions

Time does not stand still, and Igelle will develop further. New features and capabilities will be added, so it can run on multiple hardware platforms — not just x86, but also ARM and MIPS laptops. In particular, there was recently a logical bifurcation of the Igelle project into several key areas: mobile systems, servers/desktop systems, and embedded appliances — named Igelle Dash, Igelle Spark and Igelle Pulse, respectively.

There was also an interior redesign of the system’s architecture. Now, if you decide to create a fully portable Igelle application or package, it will be tied to the special library libigelle. This library alone will later decide exactly what resources are needed for this application, and dynamically reroute the necessary requests to the required underlying libraries (see Figure 7).

Figure 7: Future Igelle’s architecture

What seems interesting though is a Wine suite integrated into the Igelle Pulse distribution, the primary focus of which is x86 desktops and servers. Having the Wine emulating layer allows users to run many Windows applications smoothly. Very often, many Windows programs still have no equivalent in the Linux world; so this feature seems especially attractive to users who’ve migrated.

Most newbie Linux users expect several simple things from a general distribution: preinstalled office/network/collaboration applications to start with; a flexible and configurable environment; the presence of a pool of prepackaged applications; as well as an ability to boot the system in a live mode. At the same time, to keep the distribution truly Linux, it must be LSB-compatible, and support multiple hardware platforms: x86, ARM, MIPS, PowerPC and others.

The Igelle distribution was created in an effort to offer users just such a flexible working environment. With a vanilla GNU/Linux kernel and a stack of GNU/OSS applications, it’s generally capable of running on all major hardware platforms that the current Linux kernel supports. Those applications that aren’t yet in a pool of prepackaged applications can be easily compiled with GCC, packaged with sjdctl, and mounted or unmounted, on the fly. Since Igelle provides a built-in GCC environment, every user, from the novice to the highly experienced, can easily and quickly transform their installation to suit their preferences.

Igelle’s overlay mechanism lets you have a single distributed, basically packaged and unmodified ISO file, with transparently mounted software “upgrades”. Igelle can be installed on a freshly reformatted media, or a single ISO file can be placed on an existing filesystem. The overlay concept allows for two booting modes: the normal and “live” mode. The latter helps restore systems from an inconsistent/damaged state. This concept also allows users to access/keep personal data on already existing partitions — again, the maximum flexibility.

In short, Igelle really delivers you the flexibility to tailor your system as you please!

GNOME, an easy-to-use desktop environment, will teach you how simple a Linux OS really is, with its new GNOME Shell. GNOME 3 can also be used without using GNOME Shell as the interface. I tried the GNOME 3 preview in my openSUSE 11.4 system, and based this review on that experience (so there might be slight differences in other distros).

Appearance

In GNOME Shell, the top panel has one button on the left, called Activities. This replaces the old Application, System and Places menus. The top right corner has indication applets, similar to previous versions, but with the addition of social networking applets using Empathy. This is a concept already implemented in Ubuntu, called the Ubuntu Messaging Menu. There is an entry for System Settings in the top right corner menu, where you have the Shutdown and Restart options. The clock is placed in the centre of the top panel and clicking on it will display the calendar, along with an attractive event scheduler. If you choose to open the calendar, you are redirected to Evolution.

Clicking the Activities button displays two tabs: Windows and Applications. The window switcher displays all the open windows in an efficient way, giving users a preview of all of them. When the Activities icon is clicked, you also get an application launcher dock on the left, where you can pin your most-used apps by right-clicking the app in its active state, and choosing the Add to Favourites option.

The Applications tab is where you can find your desired apps. It initially displays all the apps, but the category list is located on the right. There is a search bar included above the category list, where you can type the name of the app or file that you are looking for. However, the search bar failed to work in the GNOME 3 preview. The shell displays error messages and status messages in a pop-up screen, which does not interfere with the application that you are currently using.

In the bottom right corner, you will get a gnome-settings-daemon icon that is used to change the screen resolution, which I believe is a feature of openSUSE. The app that is currently active has its icon placed adjacent to the Activities button.

Figure 1: GNOME 3 desktop after system startup

Hardware integration

Deeper integration with the hardware is a key feature of GNOME 3, as mentioned by its developers. When tested on my laptop, it automatically added the Suspend option, but failed to start the battery indicator applet. Compositing and other graphical effects worked out of the box on my laptop — if 3D capabilities of the graphics card is not supported by the underlying OS, GNOME 3 has a fallback mode that looks like a stripped-down version of GNOME 2.x.

When a device is plugged in, GNOME 3 immediately detects it and displays a status message at the bottom. The System Settings option in the menu can be used to configure the devices. GNOME 3 displays the configuration icon for devices in the System Settings only for devices that are detected. It does make full use of the hardware, but there is always room for improvement.

Social networking from the desktop

Social networking is built on Empathy, which uses the Telepathy framework. You can add your Google, Facebook and other chat accounts by entering the details in Empathy. The My Account option on the top right corner menu is only to add user details, and not chat accounts. In case you are using some other application and your friend types a chat message, Empathy will not interfere with the active application, but displays the message in a dialogue box that rises from the bottom of the display, and allows you to reply without minimising or closing the active app.

Unity vs GNOME Shell

Unity, which was developed in a very short time, is preferred by most people for use on their laptops and netbooks. GNOME Shell provides an interface suitable for both laptops and desktops. As GNOME 3 development started two years back, it ensures stability.

Unity pushes the title bar and the menu bar of an application’s window to the top panel, which provides more space to work on small screens (netbooks). GNOME 3, too, ought to be able to detect the screen size of the system and introduce this feature in later versions.

Trying GNOME 3

The GNOME 3 preview can be tried in openSUSE 11.4. After installing openSUSE, you just have to update the system to try the GNOME 3 preview (or install GNOME Shell from the package manager). To try the actual GNOME 3 in openSUSE, you need to upgrade the OS to its Tumbleweed version, which is a rolling release of the OS, as follows:

zypper ar --refresh http://download.opensuse.org/repositories/openSUSE:/Tumbleweed/standard/ Tumbleweed
zypper dup --from Tumbleweed

To try GNOME 3 in Ubuntu 10.10 run the following command:

sudo add-apt-repository ppa:ricotz/testing && sudo apt-get update
sudo apt-get install gnome-shell

Next, hit Alt+F2, and run the following set of commands:

gnome-shell --replace

To install GNOME 3 in Natty run the following set of commands:

sudo apt-add-repository ppa:gnome3-team/gnome3
sudo apt-get update
sudo apt-get install gnome-desktop3

As far as other distros, I also tried GNOME 3 in Debian, but the display crashed and I was left in a shell environment. For stability, Debian might not currently support GNOME 3 adequately. openSUSE 11.4 provides a GNOME 3 preview, but still has lots of glitches in the integration. Fedora 15 beta ships with GNOME 3, and gets the credit for the best integration of GNOME 3 with GNOME Shell.

Experiment with GNOME 3, and improve it. Spread the word on the use of GNOME 3-based distros to others. The questions I asked in this article’s teaser can only be answered by those who have tried GNOME 3. Do check out the new GNOME 3 website, which has a lot more information on it.

Let’s take the money out of this. If money were no issue, you’d buy a subscription for Red Hat Enterprise Linux. You’d get a copy of RHEL 5.3 and some technical support, and you’d be entitled to upgrades to RHEL.

Bring money back into the equation and you’ll think again. A basic subscription for RHEL with two server sockets will set you back about Rs 23,380 per year. That’s a basic subscription, meant for SMEs and it’s expensive. The most expensive subscription is about Rs 1,67,140 per year.

Now what if I were to tell you that you can get a copy of RHEL, with community support, for use on unlimited servers for an unlimited amount of time — all for free! Yes, it’s called CentOS, and it’s supposed to be the best free Enterprise Linux out there.

This is what Wikipedia has to say: “Red Hat Enterprise Linux is a commercial product, available only through a paid subscription service that provides access to software updates and varying levels of technical support. The product is largely composed of software packages distributed under open source licenses, and the source code for those packages is made public by Red Hat.”

It’s this source code that makes Community Enterprise Operating System possible. CentOS builds this source code in such a way that the final product is an almost exact clone of RHEL. It’s binary compatible with RHEL, to the extent that updates for RHEL can be used in CentOS and vice versa. If you’re running CentOS, you’re running RHEL.

There are a few key differences between RHEL and CentOS. Red Hat uses proprietary software for systems management, including the infamous up2date, which is used to download system updates from the Red Hat Update Network. CentOS replaces all the proprietary parts of the OS with open source components. Up2date is replaced with YUM (YellowDog Updater, Modified), the same tool used in Fedora, and CentOS uses YUM repositories for updates and software installation.

So what features does CentOS offer?

1. It’s free of cost, and has community support. It’s binary compatible with Red Hat Enterprise Linux.
2. It’s rock solid — suitable for use in servers, clusters and workstations, as well as desktops and laptops.
3. It has integrated virtualisation support with a Xen kernel.
4. Included are tools for clustering and storage clustering.
5. The ability to run as just another desktop OS with all standard desktop features.

So let’s get started with the review.

Test bed

A copy of Sun xVM VirtualBox 3.0.2 runs on Fedora 11’s PAE kernel on an Intel Core2Duo E7200 processor (2.53 GHz and without VT). The VM has 1 GB of RAM.

First impressions

Before version 5.3, I’d used a customised version of CentOS 5.1. The first thing that struck me, from the installer itself, was the new Artwork stack. The CentOS developers have outdone themselves in this sphere, where it’s normally accepted that all servers should look ugly. Speaking of installation, it’s the same old Anaconda at version 11.1.2.168, with all the usual options. Check the installation box for details.

The desktop

In CentOS, you can choose between KDE 3.5.4 and GNOME 2.16.0. Though I prefer neither, for stability and familiarity’s sake, I chose GNOME for the review. Visually, nothing has changed; it’s the same old icons, the same old pointers and the same old GNOME. Something that could have been done was to add more shortcuts to administration tools. The new wallpaper, however, is refreshing.

The kernel version stands at 2.6.18-125.2.1.el5. Throughout a major version of CentOS and RHEL (such as 5.x), the versions of software cannot change. So, because CentOS 5.0 came out with KDE 3.5.4 (and kernel 2.6.18), all minor versions of CentOS 5 have to carry those versions of software. No cosmetic fixes are provided. Updates are necessary bug fixes or security fixes only. Therefore, OpenOffice stands at version 2.3, The GIMP at version 2.2, GCC at version 4.1.2, etc. So laugh out loud.

The server within

Jokes apart, let’s come in to what CentOS actually is — a server operating system. Right from installation, CentOS is optimised for server functions. The ‘Recommended Partitioning Scheme’ involves creating a separate /boot partition, and puts all the other partitions into an LVM. Pretty cool, because ‘Dynamic Storage Scalability’ (whatever that may be) is a big thing in servers.

At the Web stable, Apache version 2.2.3-22 heads the stack, with PHP at version 5.1.6. I have had bad experiences with that PHP version; it would not let me run RoundCube Webmail on it. MySQL and PostgreSQL are at versions 5.0.45 and 8.1.11. There you have it, your LAMP or LAPP stack. Personally, I’ve got nothing against this except the old PHP. PHP 5.2.10 and 5.3 SRPMs (Source RPMs) for CentOS are available at http://www.clfsrpm.net/php/. You’ll need to build them. You’ll also need to have the EPEL repositories added.

When it comes to FTP, VSFTPD 2.0.5 just doesn’t cut it. It isn’t as popular as ProFTPD, and people say it’s not as easy to configure. I say, ditch it and install ProFTPD, which is available in the RPMForge repos.

Note: Any CentOS user should have at least the RPMFusion, RPMForge and EPEL repos added. Lots of software is available in these repos that not only make our lives easier, but are also necessary for some tasks.

As for a LDAP server, CentOS includes CentOS Directory Server version 8.1.0. CentOS-DS is just a rebranded copy of Fedora Directory Server, so anyone with experience in FDS or Red Hat DS would have absolutely no trouble using CentOS-DS.

Now, let’s come to virtualisation. CentOS includes Xen and QEMU. I was surprised to see something like KVM absent. Looks like KVM is in the Extras repository. I didn’t install it, because my CPU doesn’t support VT.

Virtualisation is provided by QEMU and Xen. Though management is a completely different matter. It is provided by some tool called Virt-Manager. Accessible from Applications -> System Tools -> Virtual Machine Manager, this tool has somewhat revolutionised virtualisation on Linux. It first made its mainstream appearance in Fedora 10. This tool follows a connection-based architecture. You open up this tool and connect to one or several machines and choose your Hypervisor of choice. It currently supports Xen, KVM and QEMU, so you’re covered unless you are a VirtualBox fanatic like me.

Once you are connected to a machine, you create a new VM on it, using a wizard. Because it can use QEMU as a backend, in fully-virtualised mode, you can have i686, x86_64, ppc, sparc, mips and mipsel as architectures. Of course, in para-virtualised mode, you can only use your host arch. All other options are standard to what you would expect from VirtualBox. It’s a very powerful tool. I say stick with it.

The other software are pretty typical. For DNS, a CHROOTed BIND 9.3.4 is installed. A ‘Legacy Servers’ group includes servers for Telnet, BootParamD, XinetD, RARPD, RSH, RUsers, RWho, Talk and TFTP. The Network Servers group includes, amongst other things, OpenLDAP, Amanda-server, DHCP, DHCPv6, Quagga, Privoxy, etc. There are GUI tools to manage servers, available from System -> Administration -> Server Settings. These GUI tools make life for sysadmins a lot easier. However, they do cut out some functionality, leaving no option but to manually edit files to change advanced settings.

Last but not the least, at the e-mail section, Dovecot is used as the default IMAP and POP3 daemon, but even though Postfix is included, Sendmail is still used as the default MTA. It’s just a matter of removing Sendmail and installing Postfix, though.

Issues while running natively

My PC has an nVidia nForce Networking Controller. CentOS 5.3 refuses to work with it. To resolve this issue, manually download the latest kmod-forcedeth package from the updates repo and install it, then run modprobe forcedeth. Instant fix.

Another issue is that since NTFS is broken in the kernel, it is not provided and you have to install ntfs-3g from RPMForge. Even after installing it, drives are not recognised in the GNOME and KDE virtual filesystems and have to be mounted and browsed, manually.

We have better-looking and more expensive enterprise Linux options with more up-to-date software, but nothing beats CentOS. It’s free, and it’s so very modular that you can use it as your home OS with all multimedia codecs and 3D effects, and without any server software. It’s so stable that only a power cut can bring it down. CentOS is just not something that can be laughed at. It might well be the world’s best freebie out there.

Naturally, the next step was to head over to fedoraproject.org and download the ISO. As is the norm with any other distro, Fedora 11 (codenamed Leonardo Leonidas) comes as an installable DVD, or live CDs of various spins, including GNOME and KDE. I chose GNOME (for a change), because features such as automatic codec and mime installation are reportedly GNOME-only for this release. Talk about treating KDE as your step child ;-)

The test systems

An HP 550 laptop with the following specs:

• Intel Core 2 Duo T5470 (1.6 GHz, 800 MHz FSB, 2 MB L2 cache)
• 1 GB 667 MHz DDR2 SDRAM
• 160 GB 5400 rpm SATA
• Mobile Intel GMA X3100 graphics

An assembled AMD-based desktop with the following specs:

• AMD Athlon X2 5600+ Dual Core (2.8 GHz, 2000 MHz HyperTransport, 1 MB L2 Cache)
• 2 GB 800 MHz DDR2 SDRAM
• 250 GB 7200 rpm SATA + 160GB 7200 rpm PATA
• NVDIA 7050 PV (integrated graphics inside an ASUS M2N-VM DVI MoBo)

Test drive on a live CD

Since the kernel modesetting (KMS) works on Intel (as well as ATI) graphics cards, but not on NVIDIA by default, I booted the HP laptop first. KMS is the backbone of Plymouth, the replacement for RHGB for boot splash, which helps in a flicker-free X server initialisation. While booting the live CD, I noticed the new splash screen—the Fedora infinity logo. You see the magic of KMS when GDM loads after the initial boot—the transition effect is, in fact, smooth. You understand the point that Fedora developers are trying to make when they say: “The start-up experience needs to be flicker-free, seamless and shiny.”

On the GDM screen, I also noticed what I believe is the new option for logging in using a finger impression, provided you have a fingerprint reader. I don’t, so I couldn’t test this one.

After logging into the desktop, you’re greeted by the excellent Fedora wallpaper—I loved it for the first time since the project’s inception. It has that professional touch that was always missing. Apart from this, things look pretty much the same visually. It’s the same GNOME 2.6 that we already know of from the Ubuntu and Mandriva Spring releases. The icon set, dubbed Fedora, is the plain old customised theme that made its debut with Fedora 7. Whatever happened to the Echo theme?

A change that I noticed is the new volume control tool. This one gets rid of the separate GNOME ALSA mixer and the one offered by PulseAudio, and integrates everything at one place. It makes sense to control all volumes from one location, but the application lacks the amount of mixers present in the ALSA mixer.

What’s quite discouraging to see is the lack of applications on the live CD. OpenOffice.org is missing, and so is a general digital camera/photo manager application like F-Spot. Well, of course, these can be installed from the online repository, but what is it that fills up the 687MB CD that leaves no room for OOo (which is present by default on all other live CD distros)? The Fedora team says it chooses to support a more complete set of languages that takes up the space as compared to others supporting only English (US). Good enough reasoning, but I guess if you choose to use another localisation setting, it’s a smaller download compared to OOo’s size, which most people will anyway need. The provided AbiWord as a replacement word processor is no solution at all.

Installation notes

Well, time to install the distro. The live installer, I gotta admit, is snappy. The only downside: you don’t have a say in the filesystem you’d like to use for your root. Fedora defaults to ext4. We’ve all heard a lot about data loss in this new FS when a computer is not properly shut down or if the OS crashes—a problem also common with XFS. Ext3, that way, has always been a safe bet. The kernel developers have obviously taken note of the complaints and patches to cure this have been included in version 2.6.30. Fedora, however, ships with 2.6.29, but the developers have reportedly backported these patches to its kernel.

I’d say, good job there. However, I have serious issues with giving the user no choice apart from ext4 for his root FS. The reason being, as I’ve been told, live CDs don’t have the ability to change filesystems. Essentially, they transfer a single pre-formatted image to the disk and write a boot loader on the system at the end of the process. You can, however, choose your own FS if you’re using the installable DVD. Well, other live systems—viz., Ubuntu, Mandriva, openSUSE—surely let you select your own FS even when you install from live CDs.

Anyway, ext4 it is. The problem again is, it’ll force you to create a separate /boot partition of ext3 type, because GRUB still doesn’t have support for ext4. Anyway, enough with the complaints, the installer manages to copy the whole system within 5 to 6 minutes, which I gotta admit is amongt the fastest.

Live CD installer reports ext4 is the only option for root partition

LFY, on the other hand, chose to bundle the DVD for being a more complete package. I got access to the DVD a bit later, which prompted me for fresh install on my assembled desktop PC. Anaconda behaves pretty much the same way, with all the options remaining the same. I still have issues with the partition screen where, by default, it’s set to erase all your partitions. You must select to manually set the partitions from the drop down, unless you’re using a blank hard disk.

While the installation using the live CD was faster compared to the other distros, the installable DVD took a lot of time installing packages. It took more than half-an-hour with the default choice of packages—the only thing that I had added to it were the packages under the virtualisation section. What I’ve noticed with Anaconda over the last few releases is that it starts off installing the packages at good speed, and then gradually slows down irrespective of the package size that it’s installing.

Back to the desktop

Booting up the brand new system, I notice the flicker-free X server initialisation missing because of being on a system with an NVIDIA chipset. The good thing is Fedora now uses the free Nouveau drivers for NVIDIA instead of NV—we all knew how unreliable NV was, right? I remember Fedora 10 live CD won’t even start the X server; F11, on the other hand, even lets me use the maximum resolution of 1280×1028 using Nouveau.

Oh yes, the DVD installation gave me a more complete GNOME desktop, including OOo and F-Spot. Talking about OOo, its icons are still the ones that appeared with the release of Red Hat 8 back in 2002, and they are ugly.

Anyway, time to see what else is there on the DVD. Alas, the PackageKit/Yum interface doesn’t include the DVD as a source by default and expects you to download the packages even if they are available on the DVD. Talk about waste of bandwidth and time. What’s worse, the GUI doesn’t give you an option to add it. Heck, even a simple option to add a proxy is missing.

The only option it lets you have control over is checking boxes to enable/disable the repositories. I understand this is adequate if you’re only interested in adding third-party repos like RPM Fusion. You can simply head over to the website and install the RPMs for free and non-free repos, and it will automagically configure and enable the repos in your sources. But what if you’re behind a proxy or want to add the DVD as a source? Well, then you’ve got to manually edit some files.

Here’s how you add the DVD as a source—as root, create a file called /etc/yum.repos.d/fedora-dvd.repo and enter the following text in it:

[fedora-dvd]
name=Fedora 11 DVD
baseurl=file:///media/Fedora%2011%20i386%20DVD/
enabled=1
gpgcheck=0

That’s it! Now the package manager is intelligent enough to not download the packages that are already on the DVD.

And here’s how you set up Yum/PackageKit to use a proxy—append the following text in the /etc/yum.conf file:

#The proxy server - proxy server:port number
proxy=http://MyProxyURL:PortNumber
#The proxy user name and password
proxy_username=MyUserName
proxy_password=MyPassword

Of course, fill in your appropriate proxy URL/IP, port number, user name and password details above.

Fedora desktop with the Echo icon theme

So, the first thing I did was install the Echo icon theme. Although, it’s not a finished product, it’s still gorgeous! Next thing was to test how the auto mime installation works. I headed over to my music folder and double clicked on an mp3 file. It automatically prompted me to press ‘Search’ to look for appropriate codecs, and returned gstreamer-ugly-plugins soon. Somehow, I wasn’t so lucky with AVI files.

Media player automatically looks for missing codecs

Still on the subject of PackageKit and Yum, I think it’s rather slow in figuring out dependencies or even searching packages. I’m sure those from the apt-get wonderland would be disappointed here. Heck, I even find Mandy’s urpmi and openSUSE’s zypper faster.
To end the discussion on package managers, the 11th edition of Fedora also officially supports delta RPMs, thanks to the Presto Yum plug-in, although this is not enabled by default. Install it by executing: yum install yum-presto.

Installing Exaile using PackageKit

You must be aware delta RPMs are a great way to save on your bandwidth traffic. This is because when updates are made available for any package, the package manager only needs to download what has changed in the new release since the last time (aka the diff patch). However, please note, it will only work if the said repository supports delta RPMs.

Coming back to applications, we all know the needs of users vary significantly. However, Fedora, just like other distros, has done a pretty decent job of selecting the defaults. And if you need more, you have the DVD and the online repos.

Traceback (most recent call last):
File "/usr/share/system-config-display/xconf.py", line 376, in
dialog = xConfigDialog.XConfigDialog(hardware_state, xconfig, rhpxl.videocard.VideoCardInfo())
File "/usr/share/system-config-display/xConfigDialog.py", line 641, in __init__
if len(self.xconfig.layout[0].adjacencies) > 1:
IndexError: index out-of-bounds

One thing I’d like to mention here is the decision to include Firefox 3.5. Well, it’s still a beta release, but nonetheless, it’s working like a charm on my installations. The new version debuts with the private browsing feature—you can access it from Tools—>Start Private Browsing. The immediate result is, Firefox saves and closes all the currently open tabs, and pulls up a new empty window with the following info: “In a Private Browsing session, Firefox won’t keep any browser history, search history, download history, web form history, cookies, or temporary internet files. However, files you download and bookmarks you make will be kept.” In addition, there’s more fine-grained control over clearing your browser history—you can choose to delete the history for an hour, two hours, four hours, today’s, or everything. Additionally included is Thunderbird 3.0 beta, which finally comes with an outbox :-)

Overall

Although the overall experience with Fedora looks to be quite disappointing, it’s also a pleasure running it—where else would you be able to check on the latest features that are awaiting you? Fedora 11 still proves to be among the most innovative GNU/Linux distros. The only thing missing is proper usability checks on essential features/options people come to expect from their desktop OS. Then again, we’ve too many distros that cater to the usability aspect to make GNU/Linux acceptable as a desktop OS by the next generation of users.

Critics often say Fedora is the test bed for Red Hat Enterprise Linux. I’d like to differ and point out that Fedora is, in fact, the test bed for most, if not all, GNU/Linux distros.

Ratings

• Installer: 3/5
• Ease of use: 3/5
• Features: 4/5
• Stability: 4/5
• Community: 4.5/5
• Overall: 3.5/5

Resources

• Fedora 11 Feature List
• Kernel Modesetting
• Nouveau Modesetting
• Graphical Boot Sequence
• RPM Fusion

Update: The headline of this article wrongly quoted the Fedora 11 release name as Leonardo previously. The incorrect release name has also been published in the July 2009 issue of the magazine. The reviewer regrets the error.