0.7.1 Is Finally Here!
#1
Posted 27 January 2010 - 06:06 AM
Well its taken me weeks of on-and-off working on this version to finally get it here to you guys. Pretty much all of which was taken up trying to fix two blocker problems. The first one being that somehow, something changed after 0.7.0 in the Gentoo booting system which caused /etc/init.d/checkroot to try and mount root as read/write during the boot process. This is of course impossible given that root is the disk and it is mounted via a loop device. Shortly after this error came up another one showed up in the form of an error while trying to mount the SquashFS image to the loop device, stating a file could not be found. After a few days of digging around it turned out that rather than looking straight for the block device at /dev/loop0 as was always done, now mount was trying to search for a symlink /dev/loop/0 which would then point to /dev/loop0 Adding the proper directory and symlink solved that issue. And then we were back to checkroot. After seeking advice on Gentoo's forums and IRC, and toying around trying to remove checkroot, someone suggested making root read-only on /etc/fstab. Duh, not sure why I didnt think of that. Gentoo-wiki's LiveCD guide didnt indicate it so I didnt even test it.
A final third quick bug came from (likely) an update to SLiM which reads a parameter in it's config file differently. It was causing X to start, but not Fluxbox. A quick Google search found the bad line and a simple fix was all that was needed.
Despite all this, I feel I still got a lot done for this release. (yes, this is where the good news starts)
Its near midnight and Im in bed now so I'll keep this brief:
- Forever, I've been building the SquashFS image of / which included the contents of /etc/, /home/, /mnt/, /var/, and /root. Since all of these directories are mounted via tmpfs as read/write and have tarballs extracted into them, keeping the contents of these directories as part of the SquashFS image was only wasting space. Deleting the contents of these directories in the image freed up roughly 6mb.
- The cryptic and annoying boot message in 0.7.0 asking you if you wish to test the initramfs and then requiring the user to enter a key, has been fixed. Now, the message more properly displays " press 'b' for initramfs shell" and is on a one second timer. If the user doesnt press 'b' within a second, the boot process continues as per normal. Pressing 'b' will run 'busybox sh' which will load a busybox shell. So rather than previously in 0.7.0 having to run commands such as 'busybox ls' or 'busybox vi' you can now run 'ls' or 'vi'. Simple.
- A small fix has been applied to Conky. As it turns out, displaying battery status for laptops was being displayed incorrectly. Changing a few lines in /etc/conky/conkymaker.sh fixed this. Laptop users rejoice! (you better.. *shakes fist*)
- Similar to the issue noted in item 1, the initramfs also had redundant binaries and libraries. With busybox present, there was no need to have the extra binaries for 'tar', 'mount', etc. Since these commands can be run in busybox. So they now are, and this results in removing the extra binaries, and any unneeded libraries. This shrunk the initramfs by about 200kb. Not much, but its still doing things the right way, and thats what counts.
- During the first few weeks of trying to solve the root issues with 0.7.1 I was getting quite annoyed with having to wait for the entire compile process to take place when I made significant changes to the ISO that couldnt be done otherwise. So I began writing an UPDATE script. The script is similar in nature to the MAKE_NEW and MAKE_c_NEW scripts. It will take a previous version of FLY and update it to release a newer version. It will copy the old version to a new directory, update the ebuilds, recompile the kernel, and remake the ISO. If no packages need updating, and only the kernel needs to be compiled, the process takes about an hour. Far less time than the 10-11 hours required for a full remake. The process takes longer the more packages need updating though. At the release of a new minor version (ie from 0.7.x to 0.8.0) the UPDATE script wont be used, just so we can start with a clean slate.
- Whatever issue caused failures last year with kernels above 2.6.25 concerning the 'cdroot' variable in Grub seems to have disappeared. So it has been re-added as part of the kernel parameters. This gets rid of the annoying warning about the kernel modules being read-only during boot up.
- The installer has been tweaked a little bit. Previously, if you installed FLY, when you'd boot you'd get a warning about v86d, and then you'd have to wait a while before anything showed up on screen. It could lead you to believe FLY had crashed. This is due to an error with the contents of /dev A fix for this is displayed on /etc/issue (the login message in CLI) but since FLY now boots to the GUI login manager SLiM, the problem - and the solution - would be camoflaged from the user. Now the /dev fix is applied during the installation. So you get a cleaner install, and when booting, you'll see the boot process shortly after the v86d error message. Im still working on trying to fix that.
- Again due to the errors encountered on this version I poured through the kernel looking for anything which related. While I didnt find anything I did take the oppertunity to add support for a few filesystems: JFS, XFS, ReiserFS, HFS, and HFS+
I think thats all I had on my mind to say about this version. I didnt test it extensively, I wanted to finally get something out to you guys. So please, if you ever had to try one version of FLY, make this the one. Get your friends to try it. Get your mom to try it. Even uncle Lewis. I dont want to be the only one working on this!!
For now, you can grab the torrent file in the 'current version' section of FLY's website http://linuxfly.net direct downloads will be available in the morning, Im uploading now and it wont be done for another hour or so, and I plan on being in never never land by then.
#5
Posted 04 February 2010 - 07:56 PM
netinfinity, on 27 January 2010 - 05:27 PM, said:
I'm experiencing poor performance with this version ran in a VM with 512 Mbytes RAM using VirtualBox with Backtrack 4 as Host on my laptop with Intel Pentium dual core 2.0Ghz cpu.
Any thoughts on this anyone?
#6
Posted 04 February 2010 - 08:25 PM
Anonymous User, on 04 February 2010 - 07:56 PM, said:
Any thoughts on this anyone?
Yeah, don't use VirtualBox. Look I've found myself saying this a lot lately, yet some people (*cough*baphomet*cough*) can't seem to get over it. If you're hardware doesn't support virutalization (VT-x/AMD-v) then your going to suffer a huge performance hit (which equates down to a waste of hardware).
I'd say switch to KVM which is kernel accelerated qemu (you won't benefit from kernel acceleration if you dont have vt-x/amd-v). However all in all it's a more light weight and robust VM suite.
<fanboy moment>
Also, Linus Torvalds recognizes KVM as the de facto VM suite for Linux
</fanboy momen>
VM's are primarily meant for server and thin station deployments, never should the be used for heavy desktop work. At least in the technologies current state that is. Perhaps when we see greater advancements in paravirtualization (KVM and xen aren't terrible at para-virt atm) and parallel systems become increasingly cheaper, then maybe my stance on this will change. However at this current point in time, desktop VM usage is nothing but a shitty hack.
#7
Posted 04 February 2010 - 09:26 PM
One thing I have noticed though is there seems to be a delay between SLiM starting up and displaying the login screen.
#9
Posted 04 February 2010 - 11:53 PM
Good point, I thought of that too. There is an option for that but I'm only able to give it 1 core atm. I may need to do some digging around. I believe there is a module that I need to load in order to do this effectively.
Thetan,
I should have clarified before... My hardware more than meets the requirements for smooth running VM's at least in my experience with various combonations of hosts and guests of the following operating systems.
Sabayon
Ubuntu
Arch
Backtrack
DVL
Windows XP/Vista/7
I've been able to run up to 3 VMs at once before a tolerable lag occurs.
Pilot,
Now that you mention it I think I've had best results while running VMware Workstation. Unfortunately my pirated copy is for win32 So I guess I'll dig around for documentation on running Vbox through Ubuntu as BT4 is Ubuntu based.
#11
Posted 05 February 2010 - 05:22 PM
EDIT:
FYI, running vmware workstation from a Windows host is like x10 performance wise. Unfortunetly I ran into a bit of trouble running the install script. It got to the part where it starts to download files and it just hangs there. I imagine that my school was filtering whatever port is used and I'm going to try again from home tonight.
This post has been edited by Anonymous User: 10 February 2010 - 05:24 PM
#13
Posted 11 February 2010 - 06:19 PM
#17
Posted 12 February 2010 - 01:34 AM
Anonymous User, on 11 February 2010 - 07:45 PM, said:
Lol, ok so i was planning to do this today at work (it's mostly dead today) so i come in to find that the only boxes around here that have hardware support for virtualization are being used by clients or either in production server use -_-, i'm sooooo bringing my laptop to work tomorrow just to do this.
#20
Posted 17 February 2010 - 12:03 AM
Also just in case someone is going to ask, yes I am sure that I am connected to the internet.
EDIT:
Just realized this might be pertinent. I'm trying to install to a 12G Fixed size VM using VMware Workstation on a Windows 7/32bit host. For operating System type I chose Other Linux 2.6.*
This post has been edited by Anonymous User: 17 February 2010 - 12:05 AM

Sign In
Register
Help

MultiQuote