Jump to content

Hirogen2

Members
  • Posts

    52
  • Joined

  • Last visited

Everything posted by Hirogen2

  1. Another solution (which ALWAYS works) is to insert another HD into the system, and stuff any BOOT ISO onto it.
  2. mknod /dev/ippp0 c 45 128 chmod 600 /dev/ippp0
  3. Somewhere in the deep deep depths of the K control center you can set the icons' sizes. I think it's under the "Desktop" category or so...
  4. If teh "asm" directories would also be split depending on containing hardware, this would affect all i387 FPUs, not only x86 and x86_64 :-)
  5. Hirogen2

    how to

    When at the bootloader, add rw init=/bin/bash then edit /etc/shadow when the prompt appears and change something like root:TrxvqrcGlSKFo:11723:0:99999:0::: into root::11723:0:99999:0::: then umount -a and restart the system (this time without the rw init=/bin/bash). Since this can pose a serious threat in rooms open for the public, one can secure LILO with the 'restricted' parameter (see lilo.conf(5)). 'restricted' requires a password if options have been passed to the interactive boot loader.
  6. ps aufwwx | grep smb See if the first field (of the parent-most smbd) says 'root'. Childs of the initial smbd are allowed to run as something else than root.
  7. A VirtualHost directive allows (to the most part) anything that can appear in the default (/etc/httpd/httpd.conf) section, like DocumentRoot, etc. even SSL contexts.
  8. When you have the sources of the currently running kernel, you don't need any reboot.
  9. It usually works when the kernel version is the same. NVidia of course stops working when the libGL files are not of the same version as the kernel module. (As it says in the docs, as I experienced myself)
  10. Do a live update, but be sure not to fux0r anything vital up, because otherwise you're stuck. (Includes: Kernel, Glibc, Internet connection programs like ppp, if applies). It works the following way: In whatever install frontend your distro uses, choose the new '10.0' tree as installation source, then go into the package selection dialogs and choose all packages for reinstall. That's at least the way it works with SuSE. Of course, there is also a way for the real pedantic ones. It's less comfortable but provides maximum safeguard: Download the necessary RPMs (or whatever appropriate) and install them by hand: rpm -Uhv xxx.rpm If it requires additional packages, add their filenames together with xxx on the command line. That way, you can only fsck up one part at a time. (Well, not always). Because some of the new packages require other packages or dropped dependencies, one might need to use the --nodeps flag (rpm), but don't wonder if afterwards an application doesnot work.
  11. mount /mountpoint -o remount might not actually do what you think 'remount' is :-) I have had several "problems" with floppy drives too, either it was: - ide cable turned wrong, - missed a pin (ouch to the rest), - power supply was wrong in, - mobo side of the ide cable was wrong... all sorts of stuff.
  12. Most X applications take the X server's DPI value to pick a font size. Unfortunately, at 1024x768 resolution and like 2048x1024 desktop size, this means that all (well, most) applications will double their font sizes. This is due to the DPI value calculated from the desktop size rather than the resolution. A way to bypass this is to pass -dpi 96 to the X server when starting. (96 might not actually be the real DPI, but Windows (grr!) uses 96 for "good readable fonts, not to small not to large", and so one should use -dpi 96.)
  13. Likewise here and hell I dunno which idiot had that idea for any ix86/Linux would be perfectly fine.
  14. Not to forget to mention that China is also planning on using something different than Windows, preferably Linux.
  15. Option maybe not, but module? If so (find /lib/modules/`uname -r` -iname '*via*') you may need to recompile the kernel to build the via drivers. If they are installed, edit /etc/modules.conf....
  16. I presume that your distribution's init scripts take the HD down on purpose. But it looks like they coded it bad-way so that it also shuts them down (i.e. with hdparm -y /dev/hda) when in reboot (init 6) besides poweroff (init 0) mode. I am not 100% sure about the usefulness of shutting them off, but IMO it's obvious that letting the HDs spin down themselves using an IDE command is better than hitting the magic button (which is the same as a BIOS shutoff), because giving the IDE command gives the HD the possibility to clean everything up, i.e. write buffers, SMART info, etc. Look into /etc/init.d/halt and look for hdparm or anything else (possibly comments) that say 'will powerdown the HD'
  17. Is smbd running as something else than root?
  18. If you cannot boot from CDROM, I suggest hijacking rescue floppies (walk over to SuSE) and use them to get into the system. Still using devfs? It's marked obsolete.
  19. I'd also like to point to Xiph MGM ( http://xiph.org/mgm/ ) but the pages seem to be unavailable...
  20. Basically, it is in everything before 2.4.27 and 2.6.7
  21. The release number (number after the dash) indicates changes to the RPM package as itself, not necessarily to the pristine source inside it.
  22. You usually do not need all steps for a normal kernel build. That really depends. If the module (read: choice you made in xconfig) has parts in other directories of the kernel which are compiled-in, you will get a 'unknown symbol reference' or similar when doing make modules_install. Only if one knows that a module just compiles a module and not "a module and something in the kernel image", s/he should do make modules, otherwise it's best to run bzImage as well. If you kept the kernel tree by not running mrproper and clean, the compile process would be way faster because everything but your new choice is already compiled. Patching the kernel is about the same as choosing something in menuconfig/xconfig. Only that patching usually requires a kernel remake even if you have not changed anything within xconfig. (Except if the patch changes docs, which arenot needed for compilation)
  23. for img in `ls *.jpg` do convert -sample 640x480 $img web-$img done Why bother with ls, when you can do it directly? for img in *.jpg; do convert -sample 640x480 "$img" "web-$img"; done;
  24. That's true indeed. The rule "twice as swap as RAM" is way obsolete and was only meaningful when we had machines with 16MB RAM being a standard. back2topic: Sometimes it's amazing... I run two fat gcc's, one VMware (requesting 96MB plus some overhead) and all can be kept in RAM (256mb). But whenever you want to play games such as Unreal Tournament with lots of add-ons (stock UT doesnot need much) or even Americas Army, the UT or AA (which even is UT2K3 based) takes way up to 500 MB virtual size according to 'ps', and then you're lucky when you got some swap.
×
×
  • Create New...