While that already is a nice feature, there's an aspect that makes this particularly exciting: with this in place we can close a major gap between Linux OSes on a bare-metal systems and Linux OSes as container guests: with this scheme in place the container manager can look at a disk image, and use the meta data we added to mount the file systems it contains to the right places before transitioning into the container. This is important since Linux containers generally cannot manage their own block devices and rely on the container manager to set up all file systems correctly. Or to say this with different words: this will soon enable us to deploy and boot OS images created with generic installers like Anaconda without any change in container managers such as nspawn and libvirt-lxc. The disk images Anaconda generates will become truly portable between containerized and non-containerized setups!
To make this reality, in the 211 release systemd-nspawn also gained support for dissecting disk images and mounting according to this specification. With the new "-i" switch you can now point it to a disk image file (or real block device), and it will make sense of its GPT partition table, and mount things properly and boot into it. In fact this is actually valuable as a recovery tool too! If you boot from some emergency medium, and now would like to get access to your installed system to make some changes, it's really simple now, simply run "systemd-nspawn -i /dev/sda" pointing to the right block device, and nspawn will get you a shell prompt, with everything properly mounted. Or add a "-b" to the command line, and it will try to boot up the thing in its entirety inside a cotnainer (which will usually work just fine, until something requires hardware device to be around, which is not virtualized for containers, i.e. gdm will eventually fail, you get no graphical session – but you do get a text login).
also updated his one-stop Fedora disk image generator script (http://people.freedesktop.org/~kay/installer/) to follow the specification. Just invoke "installer.sh foobar" and you get a disk image called "foobar", which you then simply can boot into with "systemd-nspawn -i foobar -b". Or you can do "installer.sh /dev/sdb" to create an UEFI bootable disk image on your USB stick. And then, you can use "systemd-nspawn -i /dev/sdb -b" to play around withit even without reboot! Yay!
What's missing until this appears on real systems? Well, first of all, Anaconda currently doesn't use the GPT partition type GUIDs the specificion suggests to use. We hope we can convince the Anaconda guys to make use of this (currently, they use the Windows (!) GPT partition types, which is really wrong, which hopefully makes this an easy sell). It would be great if other Linux distributions would start doing this too.
And then, we'd also need support for this in libvirt-lxc, I am already talking to about this. With this in place we can move disk images between bare-metal UEFI systems, nspawn and libvirt-lxc freely.
But even with that in place, there's one thing missing: ideally we'd like to make sure that the same images can be booted up on bare-metal, container managers and ... KVM. However, the spec is based on GPT and thus requires firmware with GPT support, which is only really available on UEFI systems. Now, it's relatively easy to make kvm use a UEFI firmware, in which case this all works fine. However, the license on the UEFI FAT driver is not free, so it will not appear prepackaged in the distributions in a way that just works. :-( The broken license is something to blame Intel for. I wished they'd work more with the Open Source community to make UEFI available for virtualization too!
You might wonder how these GPT disk images following this specification relate to Docker images? Well, both are packaging formats in a way, however, Docker is for packaging individual apps (or combinations of them), while with these GPT disk images we strictly focus on packaging full OS images.
Oh, and also note that we are not really introducing a new disk image definition here. We just build on GPT and by defining a couple of new GPT partition type IDs we open it up to be suitable for mounting within containers too.
Anyway, that's all for now!
(archers its in community)
once installed ..
espeak "text you want read out"
espeak -f /path/to/txt/file you want read out.
im lovin it.
notify-send -i ~/.icons/arch48.png "Archlinux i686 ONLINE" "Welcome Back Commander" && espeak "Archlinux i 6 8 6 ONLINE. Welcome Back Commander"
English folk may require subtitles
needless to say their marriage didn,t last.
- Making it work ◔_◔Steel Erector/Scaffolder/Banksman, present
- Glasgow City Centre
- Glasgow college of Food & Technology
Rangers clinch Scottish League 1 title after 3-0 win over Airdrieonians
Lee McCulloch's hat-trick saw Ally McCoist's side secure the championship.
Flappy Bird Is Still Available for Download, If You Know Where to Look
Flappy Bird, we hardly knew ye. But all is not lost dear iOS and Android users, as it just so happens that Flappy is still alive and well on
How to delete old Android Device from my Google account? - Google Produc...
WM Mak, 11/12/11 4:45 AM. I also want to delete my old Android Samsung Galaxy S2 device. Just upgraded to the new Samsung Galaxy Note Androi
Nexus 7 (2012, 2013) And Nexus 10 Android 4.4 KitKat Update Rolling Out ...
In a pair of exciting tweets (and a Google+ post), the Android team has announced that the WiFi Nexus 7 (both 2012 and 2013 models) will beg
Linux Mint Creator Responds to "Vulnerability" Statements Made by Ubuntu...
Clement Lefebvre, the creator of Linux Mint and one of the developers, has commented on the statements made by one of the Ubuntu devs, Olive
Ubuntu Developers Say That Linux Mint Is a Vulnerable System
Ubuntu developers have stated in the official mailing list of the distribution that Linux Mint is a vulnerable system and people shouldn't d
The Quest to Build a Truly Free Version of Android | Wired Enterprise | ...
You've probably heard that Android is free and open source. But that's not entirely true, and the team behind Replicant wants to change that
Manjaro Linux: No Assembly Required ~ Linux Advocates
Linux Advocate Dietrich Schmitz takes Manjaro 0.8.7 for a spin. Verdict: He likes it. Read why.