The idea is to have the same environment on a couple of computers (desktop, laptop, netbook), including personal files and settings, which also eliminates the hassle with bookmarks, cookies and saved passwords in Firefox/Iceweasel.

So I would like to setup all machines the same way with the same apps, then have a local fileserver as master-storage for /home, and sync each machine's /home against the fileserver either on boot, or whenever needed. Just mounting /home over NFS is a no-go, because the laptop and netbook do not stay in LAN all the time, and the fileserver should serve as backup as well.

I am still not sure which way to go, unison seems to be the only definite solution that came up on my research.

Does anybody suggest any different approach than just using unison for that task?

Indeed, Clonezilla is the way to go. I don't know in what size that gzipped dd image would have resulted (it just didn't want to work out, most likely due to a problem with my fileserver), but Clonezille made a complete backup of that 160 GB drive which now only takes about 16.7 GB disk space.

Thanks a lot for all your suggestions, guys. The link provided by oupsemma pointed exactly to the info I was looking for:

dd if=/dev/sdX bs=64k conv=noerror,sync | pv | gzip -c -9 | ssh user@server dd of=sdX.img.gz

It's already running as I write this.

BTW the mini laptop in question is an Acer Aspire One 753 with 160GB hdd and preinstalled Win7. The hdd layout shows 3 partitions:

   Device Boot      Start         End      Blocks   Id  System
/dev/sda1               1        1658    13312000   27  Unknown
/dev/sda2   *        1658        1671      102400    7  HPFS/NTFS
/dev/sda3            1671       19458   142874456    7  HPFS/NTFS

Backing up each partition separately would have been pretty unpractical, I really only want that backup just in case I sell/give that machine to anybody else, and that one insists in using the included Win7, so I can restore to factory default with an untouched Win7 OEM installation.

Way too complicated, I have no intention in even booting into Win7 once, I just want to save that Win7 OEM-preinstall as is in its virgin state.

I thought of booting into a CB live environment, sshfs-mount my fileserver, pipe the dd output through a compression program to store the compressed image directly onto the fileserver. Does anyone here have any experience with such a procedure? Is this even possible that way? Which compression program does the best job in such a case?

I just got a new mini laptop, which has a 160 GB HDD with Win7 preinstalled, but no install-media enclosed. Before I install CB on that machine, I want to backup that Win7 installation just in case.

What's the best way to backup the whole harddisc content, without wasting 160 GB on another harddrive?

Just using 'dd' for making a 1:1 image does not seem to be the best way, although it doesn't seem to get any easier, or can an image made with dd be compressed?

6

(13 replies, posted in Help & Support (Stable))

I just checked, if the option is enabled in XFCE power manager, then I opened a watch -n1 'grep MHz /proc/cpuinfo' in a terminal, and unplugged power from my laptop, but the CPU frequency just sticked with the maximum value at all times (it's an IBM ThinkPad T42p btw).

7

(13 replies, posted in Help & Support (Stable))

If neither cpufreqd nor cpufrequtils is installed, how does XFCE power manager perform CPU frequency control?

Having that option checked, I could see no effect, CPU freq just stayed on maximum.

You are right, your ACPI script approach is simple and easy, but I was wondering as XFCE power manager is installed, why it doesn't actually do CPU frequency control, although that option is activated.

Yes, I use full hdd encryption.

9

(13 replies, posted in Help & Support (Stable))

hardran3 wrote:

XFCE power manager is changing the CPU governor. Right click your battery icon and select preferences. Go to the Extended tab and make sure CPU frequency control is unchecked. That should fix it. cpufreqd is not in the standard install and is unneeded. Everything needed is in the kernel.

In the standard CB Openbox installation, cpufreq is not installed, and although CPU frequency control is checked in XFCE power manager, it does not do anything. So either something else is missing, or the cpufreq package is nevertheless needed for it to work. Setting the governor or specific CPU frequency settings by hand surely is not the best solution.

Using that method to boot without plymouth doesn't show that error on tty1.

This error message can be seen after each boot if I switch from the GDM-login-screen back to tty1. The exact line is:

(process:253): GLib-WARNING **: getpwuid_r(): failed due to unknown user id (0)

Doesn't look like it has been mentioned here in CB forums before, although a net-search shows some threads on ubuntuforums, which suggest it may have something to do with plymouth.

Any idea?

I just hate it, when strange error messages are shown, but everything seems to be working normally.

I like the idea of CB Openbox Edition, but I desire a completely different software selection. Since I want to perform installations not on one, but on a couple of machines (laptops, netbooks and desktops), it would be quite an effort to first install CB as it comes on the normal install media, and then to exchange all the relevant programs and system configurations over and over again.

Is there a simple way to create a customized CB install media, which then already incorporates the desired changes?

I know that CB is built using the Debian Live Project, but the info on live.debian.net is quite overwhelming, and I just don't have the time to learn how to build my own distribution.

Any idea?

13

(108 replies, posted in CrunchBang Talk)

What difference does it make to install Iceweasel 4 from http://mozilla.debian.net/ instead of just extracting the Firefox archive to /opt (which is what I just did)?

Firefox directly from Mozilla is already version 4.0.1, whereas I guess Iceweasel from http://mozilla.debian.net/ is still 4.0.0, and I really think that such an important tool as a webbrowser always should be used in the latest version (if I understood right, Firefox installed from the Mozilla archive simply can be updated to the latest version by starting as root using the built-in update feature).

Thanks for the hint, but as I already have found out, AltGr does not seem to be usable as "Super" replacement.

I tried it with 0x6c and ISO_Level3_Shift as identified by xev, and of course I restarted OB after changing rc.xml, but it just does not work.

No idea, what else would do. Using the right Strg would be my next best bet, since left and right Strg are identified separately as Control_L and Control_R, but both react with the generic version "C", so this will either be complicated or not even possible at all.

I really didn't think I'll ever miss those stupid M$ keys. Any other option, or do I have to live without those "Super" shortcuts on my ThinkPad altogether?

I just installed CB on an old IBM ThinkPad T42p, which has no "Windows" keys, and now I am curious what replacement I should use for the "Super" shortcut keys.

The only available keys are

  • Shift (left)

  • Strg (left)

  • Alt

  • AltGr

  • Strg (right)

  • Shift (right)

Shift surely does not serve the purpose, and I am unsure about the rest, as I don't want to compromise shortcuts already taken otherwise.

So what's the best way to still be able to use the 10 "Super" key shortcuts defined in the default CB installation, and what does the replacement entry in rc.xml look like?

16

(15 replies, posted in Help & Support (Stable))

It has nothing to do with hinting, I always have full hinting + RGB-subpixel-hinting enabled on all my boxes, as well as dpi on system default.

The dpi setting only influences the size of fonts, not their shape or rendering. I just checked with xdpyinfo, Lenny@17" runs at 95x96 dpi, #!@24" at 94x94 dpi, which is what the xserver determines from the monitor's DDC info, and both screens are showing correct sizes.

Fonts on my #! setup are not too bad, at least not like on Debian Lenny without the cleartype patches, but I can see a considerable difference, for example if I look at the letter "w".

Strangely the effect is not consistent. If I type "www" in the address-bar or search-engine field of firefox, it looks worse, than a "w" showing up in the forum text in the firefox browser window, or if I type "www" in gedit. But independently of the app I type a "w" in, it just does not look the same as on my Debian Lenny box with the cleartype patches, where all font-rendering looks just awesome, whereas on my #! setup I'd describe it as a little crappy.

I know, that most people do not really recognize that effect, especially if you are not used to anything else, and if you do not have two screens side by side to compare, but I do, and I have to say, there really is a huge difference. The font-rendering on my Debian Lenny setup is so much more pleasing to the eye, I really don't want to miss that.

So if it's not a font thing, and has nothing to do with hinting, I guess libcairo + libfreetype + libxft on Ubuntu do not have the full cleartype patchset applied (which may be a license issue?). That way it comes down to the following order:

(1) libcairo + libfreetype + libxft patched with the cleartype patchset
(2) stock Ubuntu (definitely has patched font-rendering, but obviously not the full cleartype patchset)
(3) stock libcairo + libfreetype + libxft without patches (as supplied by the Debian repos)

So either I have to start fiddling around with patching the Ubuntu sources of libcairo + libfreetype + libxft myself, or I try to get to a #! setup on Debian Lenny (which most likely would be a better idea anyway, because I just feel more comfortable with Debian Lenny, than with Ubuntu).

17

(15 replies, posted in Help & Support (Stable))

I haven't used Ubuntu myself for a while, but Debian Lenny with Xfce and cleartype-patched libcairo + libfreetype + libxft.

Using a stock #! installation on my 24" TFT monitor, I can clearly see, that font-rendering is not nearly as perfect, as it is on my Debian Lenny box, running side-by-side on a 17" TFT.

Menu -> Preferences -> User Interface Settings already shows full hinting + RGB-subpixel-hinting, so either something else is wrong or missing (like some fonts?), or Ubuntu is not using the full potential of patches for libcairo + libfreetype + libxft, and my eyes are just too accustomed to my Debian Lenny setup.

Anybody else seeing any differences between #! and normal Ubuntu front-rendering?

Please do not omit the lite edition! I'm still waiting for that one.

Of course there was something on the old site I didn't see wink

Philip wrote:

Apart from the front page...

So the frontpage is just a handmade php?

BTW It's all running on a VPS? Can you give some info on that (CPU, RAM, diskspace,...) one? It seems to be pretty powerful, because everything is loading damn fast here.

20

(81 replies, posted in Tips, Tricks & Scripts)

Isn't there an easier way to prevent the installation of nautilus altogether, by creating a fake nautilus package, that just sets a link to thunar? Isn't something similar done with the fake mono package (mononono), to prevent any mono stuff to be installed?

Philip, can you please explain, which components you have used to setup the #! linux website?

I mean, which software is used for the main website, blog, forums and wiki? It all fits together so nicely, I would like to use a similar setup for my personal website.

Other Ubuntu-based distros have the whole bunch of Ubuntu just with customized config and/or artwork + additional tools.

As #! is set on top a minimal Ubuntu installation, why not using Debian Lenny as a base instead?

I have some installations of Ubuntu, Debian Lenny and sidux here, and I have to say, that Debian Lenny and sidux work a lot better than Ubuntu. Both are faster, no printing problems (for some reason I could not get printing to our printserver to work correctly with Ubuntu), easier to establish a cleaner setup, and most likely more stable.

Just a thought, but it #! on Debian would seem to make more sense than #! on Ubuntu...

23

(5 replies, posted in Help & Support (Stable))

Tomorrow I'll most likely do the installation with German localization, using the Ubuntu alternate cd and a "command-line" installation, then the #! install-script, and report back how it went.

Still running the livecd here, and before I try to do a harddisc installation, I'd like to know, how #! handles localization, so locale, language-packs, dictionaries, keymap and such.

Till now I only found the info about "setxkbmap" in autostart.sh, but shouldn't it be the cleaner solution, to set the locale correctly?

I am wondering, if this problem can be solved, by installing the base system using a "command-line" installation of Ubuntu, and then getting to #! using the script. Does that make any difference concerning localization?

I'll have to do the harddisc installation that way anyway, because I want a fully encrypted installation on my laptop, and that's only possible using the alternate installation cd.

25

(76 replies, posted in WM/DE Talk)

I think, hakimaki meant an autoupdating application menu, like it is in Xfce, which looks for all the *.desktop files, each time it is accessed. In Xfce, each time you install or uninstall an app, it automatically appears (or vanishes, if not present any more) in the right category in the menu, which is quite nice and comfortable, because you do not need to manually edit the menu, if you fiddle around with new programs.

Is that kind of autoupdating menu possible with Openbox?

That functionality is pretty common, I remember having it as well, when playing around with FVWM.