Re: [gentoo-user] Error message in Xorg.log for intel xorg driver
On Fri, Jan 16, 2009 at 1:58 AM, Paul Hartman paul.hartman+gen...@gmail.com wrote: On Fri, Jan 16, 2009 at 1:36 AM, Dale rdalek1...@gmail.com wrote: Paul Hartman wrote: On Fri, Jan 16, 2009 at 12:59 AM, Dale rdalek1...@gmail.com wrote: Wolfgang Liebich wrote: On Thu, Jan 15, 2009 at 09:35:11AM -0600, Paul Hartman wrote: On Thu, Jan 15, 2009 at 2:58 AM, Wolfgang Liebich wolfgang.lieb...@siemens.com wrote: Furthermore yesterday I had a total lockup when I came to work at the morning --- could not login at kdm, kdm would ignore all keyboard input etc. I had to do a hard restart with the Magic SysRQ key (remount ro, hard reboot). Do you have evdev installed? Without it, you probably won't have any keyboard or mouse. Recent xorg made dramatic changes to the way hardware is detected/configured by using HAL and evdev. xorg.conf is basically unused now when it comes to configuring hardware. I don't even have keyboard or mouse, or video modelines or anything like that in mine. Search the list archives or the gentoo web forums, there are many many people who had the same issues (assuming it's the cause of yours). Evdev is installed, but I configured the kbd driver (I have a MS Natural Keyboard, btw --- what's the best driver for that keyboard?). I still have an xorg.conf (and I'm not very inclined to change it as long as it works :-). Furthermore -- after the reboot everything worked again as before. It seems to have been some fluke, but I want to know where it comes from. TIA, Wolfgang Someone else like me. I still have my xorg.conf and want to keep it too. I don't have evdev installed but from the way it sounds, me and you may have to change in the future, maybe near future. I'm sort of wondering what pulls in evdev anyway? I got a fully running KDE and this is my new install. Nothing pulled it in here. I may be missing a USE flag or something. Let's hope this works for a while longer yet. ;-) Dale :-) :-) You need evdev in your INPUT_DEVICES variable (mine lives in make.conf). In my case I have: INPUT_DEVICES=keyboard mouse joystick evdev and portage automagically built those packages. So if evdev failed for some reason, it would fall back to the keyboard and mouse drivers you think? That I would be willing to try if that is the case. Dale :-) :-) I don't know, for me it simply works as intended so... maybe I'll try to remove the keyboard and mouse and see what happens :) but in my case my xorg.conf is virtually empty aside from some fonts and nvidia card options. My display and input devices just work without being specified in xorg.conf with drivers, modelines or any of that stuff. I changed monitors yesterday and simply killed X and it restarted in the optimal resolution for the new monitor. I've plugged different mouse/keyboard and it just works automatically. The HAL policies in /etc/hal/fdi/policy contain the same exact settings as xorg.conf only formatted a little differently... you can give device-specific custom settings if you need and I think everything you have done in xorg.conf can be done the new way. I should say they CAN contain the same exact settings. It is up to you to put them there :)
Re: [gentoo-user] Error message in Xorg.log for intel xorg driver
Paul Hartman wrote: On Fri, Jan 16, 2009 at 1:58 AM, Paul Hartman paul.hartman+gen...@gmail.com wrote: On Fri, Jan 16, 2009 at 1:36 AM, Dale rdalek1...@gmail.com wrote: Paul Hartman wrote: On Fri, Jan 16, 2009 at 12:59 AM, Dale rdalek1...@gmail.com wrote: Wolfgang Liebich wrote: On Thu, Jan 15, 2009 at 09:35:11AM -0600, Paul Hartman wrote: On Thu, Jan 15, 2009 at 2:58 AM, Wolfgang Liebich wolfgang.lieb...@siemens.com wrote: Furthermore yesterday I had a total lockup when I came to work at the morning --- could not login at kdm, kdm would ignore all keyboard input etc. I had to do a hard restart with the Magic SysRQ key (remount ro, hard reboot). Do you have evdev installed? Without it, you probably won't have any keyboard or mouse. Recent xorg made dramatic changes to the way hardware is detected/configured by using HAL and evdev. xorg.conf is basically unused now when it comes to configuring hardware. I don't even have keyboard or mouse, or video modelines or anything like that in mine. Search the list archives or the gentoo web forums, there are many many people who had the same issues (assuming it's the cause of yours). Evdev is installed, but I configured the kbd driver (I have a MS Natural Keyboard, btw --- what's the best driver for that keyboard?). I still have an xorg.conf (and I'm not very inclined to change it as long as it works :-). Furthermore -- after the reboot everything worked again as before. It seems to have been some fluke, but I want to know where it comes from. TIA, Wolfgang Someone else like me. I still have my xorg.conf and want to keep it too. I don't have evdev installed but from the way it sounds, me and you may have to change in the future, maybe near future. I'm sort of wondering what pulls in evdev anyway? I got a fully running KDE and this is my new install. Nothing pulled it in here. I may be missing a USE flag or something. Let's hope this works for a while longer yet. ;-) Dale :-) :-) You need evdev in your INPUT_DEVICES variable (mine lives in make.conf). In my case I have: INPUT_DEVICES=keyboard mouse joystick evdev and portage automagically built those packages. So if evdev failed for some reason, it would fall back to the keyboard and mouse drivers you think? That I would be willing to try if that is the case. Dale :-) :-) I don't know, for me it simply works as intended so... maybe I'll try to remove the keyboard and mouse and see what happens :) but in my case my xorg.conf is virtually empty aside from some fonts and nvidia card options. My display and input devices just work without being specified in xorg.conf with drivers, modelines or any of that stuff. I changed monitors yesterday and simply killed X and it restarted in the optimal resolution for the new monitor. I've plugged different mouse/keyboard and it just works automatically. The HAL policies in /etc/hal/fdi/policy contain the same exact settings as xorg.conf only formatted a little differently... you can give device-specific custom settings if you need and I think everything you have done in xorg.conf can be done the new way. I should say they CAN contain the same exact settings. It is up to you to put them there :) I'm curious about this now. I run my monitor at 1280x1024 but it can run 1600x something. Thing is, everything is so small, I can't really see anything. Even the mouse pointer is really small, about the size of a pencil lead. If I know where it is I can find it otherwise I have to push to a corner, then find it and go from there. I need new glasses but can't afford it right now. If I can still run at 1280x1024, this may be worth trying out. I would rather try it while the old way still works rather than wait until it doesn't and run into . . . issues. Dale :-) :-)
Re: [gentoo-user] e2fsprogs blocking question
On Thu, 15 Jan 2009 23:59:32 -0200, Alejandro wrote: That's right, but now the new portage is stable so it is handled on stable systems. The block was handled automatically when it first appeared on ~arch systems. Which version of portage do this? I am on amd64 stable and have the problem a couple of week ago, and i don/t remember any portage update,,, 2.1.6 AFAIK, which contains some of the new 2.2 features. I've not used it myself because I always use testing versions of portage, even on stable boxes. -- Neil Bothwick Secret hacker rule #11: hackers read manuals. signature.asc Description: PGP signature
Re: [gentoo-user] e2fsprogs blocking question
Alejandro wrote: 2009/1/15 Neil Bothwick n...@digimed.co.uk mailto:n...@digimed.co.uk On Fri, 16 Jan 2009 00:00:07 +0100, Geralt wrote: You don't need to remove anything, just let portage handle the block for you. Blocks marked with a b (instead of a B) can be handled by recent portage releases. are you sure that his works in this case? This blocking bug was some time before the new Portage went stable and back then you had to resolve it by hand. That's right, but now the new portage is stable so it is handled on stable systems. The block was handled automatically when it first appeared on ~arch systems. -- Neil Bothwick Is it true that cannibals don't eat clowns because they taste funny? Which version of portage do this? I am on amd64 stable and have the problem a couple of week ago, and i don/t remember any portage update,,, I know portage-2.2_rc20 works well. I have not had any trouble on mine and you may want to give it a shot. It is still keyworded I think but it does handle the blocks very well. Your choice on whether to install or not. Dale :-) :-)
Re: [gentoo-user] rebuilding dependent packages
On Thu, 15 Jan 2009 18:37:20 -0500, Michael P. Soulier wrote: The libavcodec library went from version 51 to 52, which broke transcode. The --deep argument did not find the dependency there and rebuild transcode. On my FreeBSD server, portupgrade has the -r and -R arguments to force rebuilds of dependent and reverse-dependent packages. Is there a way to have emerge do the same? emerge @preserved-libs does that with portage 2.2. Emerge even tells you to run it, and hangs on to the old versions until you do so, so your system is never broken. Revdep-rebuild is good for fixing things after they are broken, but the new portage approach of not breaking them is much nicer :) -- Neil Bothwick If at first you don't succeed you'll get lot's of advice. signature.asc Description: PGP signature
Re: [gentoo-user] Error message in Xorg.log for intel xorg driver
2009/1/16 Dale rdalek1...@gmail.com: So if evdev failed for some reason, it would fall back to the keyboard and mouse drivers you think? That I would be willing to try if that is the case. I don't think there is a fallback option. You can either use the new or the old way. Somebody correct me if I am wrong. New way: INPUT_DEVICES=evdev in /etc/make.conf build xorg-server with USE=hal configure input devices in an appropiate fdi file under /etc/hal/fdi/policy/ instead of xorg.conf Old way: INPUT_DEVICES=kbd mouse in /etc/make.conf build xorg-server with USE=-hal configure input devices in xorg.conf -- Regards, Daniel
Re: [gentoo-user] Error message in Xorg.log for intel xorg driver
On Fri, Jan 16, 2009 at 2:20 AM, Dale rdalek1...@gmail.com wrote: Paul Hartman wrote: On Fri, Jan 16, 2009 at 1:58 AM, Paul Hartman paul.hartman+gen...@gmail.com wrote: On Fri, Jan 16, 2009 at 1:36 AM, Dale rdalek1...@gmail.com wrote: Paul Hartman wrote: On Fri, Jan 16, 2009 at 12:59 AM, Dale rdalek1...@gmail.com wrote: Wolfgang Liebich wrote: On Thu, Jan 15, 2009 at 09:35:11AM -0600, Paul Hartman wrote: On Thu, Jan 15, 2009 at 2:58 AM, Wolfgang Liebich wolfgang.lieb...@siemens.com wrote: Furthermore yesterday I had a total lockup when I came to work at the morning --- could not login at kdm, kdm would ignore all keyboard input etc. I had to do a hard restart with the Magic SysRQ key (remount ro, hard reboot). Do you have evdev installed? Without it, you probably won't have any keyboard or mouse. Recent xorg made dramatic changes to the way hardware is detected/configured by using HAL and evdev. xorg.conf is basically unused now when it comes to configuring hardware. I don't even have keyboard or mouse, or video modelines or anything like that in mine. Search the list archives or the gentoo web forums, there are many many people who had the same issues (assuming it's the cause of yours). Evdev is installed, but I configured the kbd driver (I have a MS Natural Keyboard, btw --- what's the best driver for that keyboard?). I still have an xorg.conf (and I'm not very inclined to change it as long as it works :-). Furthermore -- after the reboot everything worked again as before. It seems to have been some fluke, but I want to know where it comes from. TIA, Wolfgang Someone else like me. I still have my xorg.conf and want to keep it too. I don't have evdev installed but from the way it sounds, me and you may have to change in the future, maybe near future. I'm sort of wondering what pulls in evdev anyway? I got a fully running KDE and this is my new install. Nothing pulled it in here. I may be missing a USE flag or something. Let's hope this works for a while longer yet. ;-) Dale :-) :-) You need evdev in your INPUT_DEVICES variable (mine lives in make.conf). In my case I have: INPUT_DEVICES=keyboard mouse joystick evdev and portage automagically built those packages. So if evdev failed for some reason, it would fall back to the keyboard and mouse drivers you think? That I would be willing to try if that is the case. Dale :-) :-) I don't know, for me it simply works as intended so... maybe I'll try to remove the keyboard and mouse and see what happens :) but in my case my xorg.conf is virtually empty aside from some fonts and nvidia card options. My display and input devices just work without being specified in xorg.conf with drivers, modelines or any of that stuff. I changed monitors yesterday and simply killed X and it restarted in the optimal resolution for the new monitor. I've plugged different mouse/keyboard and it just works automatically. The HAL policies in /etc/hal/fdi/policy contain the same exact settings as xorg.conf only formatted a little differently... you can give device-specific custom settings if you need and I think everything you have done in xorg.conf can be done the new way. I should say they CAN contain the same exact settings. It is up to you to put them there :) I'm curious about this now. I run my monitor at 1280x1024 but it can run 1600x something. Thing is, everything is so small, I can't really see anything. Even the mouse pointer is really small, about the size of a pencil lead. If I know where it is I can find it otherwise I have to push to a corner, then find it and go from there. I need new glasses but can't afford it right now. If I can still run at 1280x1024, this may be worth trying out. I would rather try it while the old way still works rather than wait until it doesn't and run into . . . issues. Surely you can, the exact instructions depend on your video drivers and desktop environment. I'm using KDE 3.5 and it gives me an exhaustive list of screen resolutions I can choose to use as the default. As far as the size of things, I have a 2042x1152 monitor and my mouse cursor is normal sized, not as small as yours sounds. (no, I'm not bragging about who has the bigger mouse cursor :P) As far as everything being too small at the higher resolution, it sounds like your DPI setting may not be correct. If it's set, things should be roughly the same physical size on any monitor in any resolution, a 12-point font on a 15inch monitor running 1024x768 should be the same physical size as a 12-point font on a 20-inch monitor running 1600x1200. Window decorations etc should be the same size on any system if the WM/DE respects DPI. 1024x768 vs 1280x960 vs 1600x1200 should look the same from afar, the higher resolutions will just look better because they are better. :) A simple test to see if your system DPI is correct is to create a blank text
[gentoo-user] A circular dependency problem with notification-daemon and libnotify...
-BEGIN PGP SIGNED MESSAGE- Hash: SHA512 Hello, I was wondering if anyone else was having this problem. I am running an AMD64 arch, and when I try to emerge notification-daemon, it will not compile because libnotify is not present. If I try to emerge libnotify, it tries to merge notification-daemon first, and I get the same problem. When I run: emerge --keep-going libnotify, it will not compile because it depends on notification-daemon. More information to follow (I have to reboot to Gentoo, capture what I need and send it to this OS). Regards, Chris PS: I was wondering if removing the 'gstreamer' USE flag from notification-daemon might fix the problem. -BEGIN PGP SIGNATURE- iQIcBAEBCgAGBQJJcGD+AAoJEIAhA8M9p9DAM+kQAIPlsCHGK/bHyNuV8YFIQVQO QjSLugTaq3Udz2YYoYHAqNzRVmziMyy/57cHzDTT3XkS3geFeStIzZHRmtaxwm2f 6IXfGkINUD8XASC+zjMznph6LosVy5jK4aXDtdqmNNeo1N9xJrhmrw1O88py/hnW ygDStQD9JOTQi7bEbOK0PPdQnJff4STAJH8M0zIzTeFhbpkCJF6wgm6emChbhT/3 JZ+azQ3fxOcQMzpzVXvKfByi2gToZCFUMr9a29R366s1vHiV7FRX+mWjZ5aFTTC9 Yo6xBCva0xdEPnqL555hq+R6EpufjuOjLKQNOa1WaOYvns50M3zmjqL1bhVB9bFI K+m7PkTAo7Il7f025ips7H+1yKeATnL/12J6PV/Suj2iW3tJXDsHBDP7HJ4OmxmJ XP+GmWoJNbKO0m2j1pPf0T9twSTxdmZ76Se9dxAms3VBraHez5ONoqxrUdYQnDqn J4t80x/ya8GjsiIIlyWhiQmiF3re24ld47ZsJkJQH9Oq5oCaDu8jljgKWjmeppYh JOkHsddXzEMJgqm7bEWusMB0aO0JxcbE38B3Nez86jquDuzRI2YGHzAGIr9mUWvo JuNVVRSA3Km9DsPoRcF6sRgIYZagmXcRxteQVGZJ8pqoSOA4mezwnN700INUvww3 9i91ZEl9s9vIXvmU8JkT =l4Yu -END PGP SIGNATURE-
Re: [gentoo-user] A circular dependency problem with notification-daemon and libnotify...
On Fri, 16 Jan 2009 05:27:12 -0500, Chris Walters wrote: I was wondering if anyone else was having this problem. I am running an AMD64 arch, and when I try to emerge notification-daemon, it will not compile because libnotify is not present. If I try to emerge libnotify, it tries to merge notification-daemon first, and I get the same problem. PS: I was wondering if removing the 'gstreamer' USE flag from notification-daemon might fix the problem. The only amd64 ebuild of notification-daemon, 0.3.7, does not have a gstreamer USE flag. That's only present in the 0.4.0 ebuild, which is ~amd64. Are you trying to mix stable and testing packages? -- Neil Bothwick Bother, said Pooh, as the vice squad took his GIFS signature.asc Description: PGP signature
[gentoo-user] ebuild dependency logic - please explain it to me
Hi, I'm still struggling with the logic of dependencies in ebuild files e.g. kde-base/kdelibs-3.5.10-r2.ebuild contains the line !avahi? ( !bindist? ( net-misc/mDNSResponder !kde-misc/kdnssd-avahi ) ) This causes a dependency loop since I do have 'avahi' installed which blocks net-misc/mDNSResponder but other packages need avahi. Why does kdelibs-3.5.10-r2 try to pull in net-misc/mDNSResponder This is with portage-2.2_rc22 . Many thanks for a hint, Helmut. -- Helmut Jarausch Lehrstuhl fuer Numerische Mathematik RWTH - Aachen University D 52056 Aachen, Germany
Re: [gentoo-user] ebuild dependency logic - please explain it to me
On Friday 16 January 2009 13:49:04 Helmut Jarausch wrote: Hi, I'm still struggling with the logic of dependencies in ebuild files e.g. kde-base/kdelibs-3.5.10-r2.ebuild contains the line !avahi? ( !bindist? ( net-misc/mDNSResponder !kde-misc/kdnssd-avahi ) ) This causes a dependency loop since I do have 'avahi' installed which blocks net-misc/mDNSResponder but other packages need avahi. It may be installed, but is it in your USE? The output of eix -e kdelibs would be useful here Why does kdelibs-3.5.10-r2 try to pull in net-misc/mDNSResponder This is with portage-2.2_rc22 . Many thanks for a hint, Helmut. -- alan dot mckinnon at gmail dot com
Re: [gentoo-user] e2fsprogs blocking question
2009/1/16 Dale rdalek1...@gmail.com Alejandro wrote: 2009/1/15 Neil Bothwick n...@digimed.co.uk mailto:n...@digimed.co.uk On Fri, 16 Jan 2009 00:00:07 +0100, Geralt wrote: You don't need to remove anything, just let portage handle the block for you. Blocks marked with a b (instead of a B) can be handled by recent portage releases. are you sure that his works in this case? This blocking bug was some time before the new Portage went stable and back then you had to resolve it by hand. That's right, but now the new portage is stable so it is handled on stable systems. The block was handled automatically when it first appeared on ~arch systems. -- Neil Bothwick Is it true that cannibals don't eat clowns because they taste funny? Which version of portage do this? I am on amd64 stable and have the problem a couple of week ago, and i don/t remember any portage update,,, I know portage-2.2_rc20 works well. I have not had any trouble on mine and you may want to give it a shot. It is still keyworded I think but it does handle the blocks very well. Your choice on whether to install or not. Dale :-) :-) Thanks for th einfo! I will give a try... Cheers!
Re: [gentoo-user] ebuild dependency logic - please explain it to me
On 16 Jan, Alan McKinnon wrote: On Friday 16 January 2009 13:49:04 Helmut Jarausch wrote: Hi, I'm still struggling with the logic of dependencies in ebuild files e.g. kde-base/kdelibs-3.5.10-r2.ebuild contains the line !avahi? ( !bindist? ( net-misc/mDNSResponder !kde-misc/kdnssd-avahi ) ) This causes a dependency loop since I do have 'avahi' installed which blocks net-misc/mDNSResponder but other packages need avahi. It may be installed, but is it in your USE? The output of eix -e kdelibs would be useful here It's a bit strange: I have the avahi use-flag in /etc/make.conf equery uses kde-base/kdelibs shows the avahi use-flag but eix -e kde-base/kdelibs doesn't. Now, unmerging kde-base/kdelibs:3.5 and reemerging it again (from source - not from a package) seems to work but takes some time on my slow machine. Thanks for the hint though it's still a bit strange. Helmut. -- Helmut Jarausch Lehrstuhl fuer Numerische Mathematik RWTH - Aachen University D 52056 Aachen, Germany
Re: [gentoo-user] ebuild dependency logic - please explain it to me
2009/1/16 Helmut Jarausch jarau...@igpm.rwth-aachen.de: Hi, I'm still struggling with the logic of dependencies in ebuild files e.g. kde-base/kdelibs-3.5.10-r2.ebuild contains the line !avahi? ( !bindist? ( net-misc/mDNSResponder !kde-misc/kdnssd-avahi ) ) This causes a dependency loop since I do have 'avahi' installed which blocks net-misc/mDNSResponder but other packages need avahi. Why does kdelibs-3.5.10-r2 try to pull in net-misc/mDNSResponder Because you have the avahi flag disabled for kdelibs? -- Regards, Daniel
Re: [gentoo-user] ebuild dependency logic - please explain it to me
On Friday 16 January 2009 14:22:27 Helmut Jarausch wrote: On 16 Jan, Alan McKinnon wrote: On Friday 16 January 2009 13:49:04 Helmut Jarausch wrote: Hi, I'm still struggling with the logic of dependencies in ebuild files e.g. kde-base/kdelibs-3.5.10-r2.ebuild contains the line !avahi? ( !bindist? ( net-misc/mDNSResponder !kde-misc/kdnssd-avahi ) ) This causes a dependency loop since I do have 'avahi' installed which blocks net-misc/mDNSResponder but other packages need avahi. It may be installed, but is it in your USE? The output of eix -e kdelibs would be useful here It's a bit strange: I have the avahi use-flag in /etc/make.conf OK equery uses kde-base/kdelibs shows the avahi use-flag I find equery uses to give silly output. It is telling you that the ebuild looks at the avahi USE flag (i.e. it's in IUSE), not that it is actually active for that package. but eix -e kde-base/kdelibs doesn't. eix gives the output you actually want - it tells you which flags are listed, and is you have selected them or not for the package. Most likely is that you have 'kdelibs -avahi' in package.use Try these: grep -r kdelibs /etc/portage/* grep -r avahi /etc/portage/* to see better what is going on -- alan dot mckinnon at gmail dot com
Re: [gentoo-user] A circular dependency problem with notification-daemon and libnotify...
-BEGIN PGP SIGNED MESSAGE- Hash: SHA512 Neil Bothwick wrote: The only amd64 ebuild of notification-daemon, 0.3.7, does not have a gstreamer USE flag. That's only present in the 0.4.0 ebuild, which is ~amd64. Are you trying to mix stable and testing packages? No, at this point, I am using a full testing setup, since I want to pull in kde-4 and its dependencies. Everyone keeps saying how good kde-4 is, yet this circular dependency is killing my ability to pull it in. Regards, Chris -BEGIN PGP SIGNATURE- iQIcBAEBCgAGBQJJcImzAAoJEIAhA8M9p9DAtCMQAMlFjiGsKa8LIA6ljdioTM2c esdkG7jDC1stVjRj8yCiPvpzRR0k7WZL4G1XX6WjdkhbUHTZy+8PIZ4i4suq60nx qqpse94JODbgNPzK7bcyc+OTNWsyRjT7Xvm46JDt4kszadvLP39Ty5r57ZzcBfjv J+QSUa4z7rflV8qhF1QvY0qT461Hdo1B0jSqlxSaj3wB6kd4fSKASKV/p/ceVdFE E/fxx++E6RyC4WHettMteOlYtGsMGonnvbQkaKhMtOLlp45W0SHf3pOFqQ6FVYnn 1Nsfgz7FImkTvFrFwMhqd0BWNQsYNATeJETuib25BeP/9MvgX8OB9X5M45vU/lwb Y6HWCu0yODpsh8ZMGg1cyRWrFdvnTkU9P1POCZqEYD4CQCYQ8QVEic2hywLr4O4K I+NqKgYZXP6GSriNK9Cebw8zPCDQ3FMPFOocIeLiN1PMkaSdsXXWfpV/4iRTcQ6c F9vkP8VXEVu+lZiUgW6KBwlp8OBj6NOnVSDCUJZATIkEz/z1dFDSmbDXt3e8OSzV tXQygnBEdtpgxKU5kg9NJwTT/YDscUWt3dimAEcayYRHYf9wLzI9QutEis88i1S3 akJEHIKKYHj9HYsvtKpRdCVsqB86cwWVlOCtAI3vYCCgNNNqpoZJzbTWnoMbahxZ WpwE/jCK7S0SBX6Ey9Jk =hPzb -END PGP SIGNATURE-
Re: [gentoo-user] Bash Server Sockets
On Thu, 2009-01-15 at 21:53 -0800, Hilco Wijbenga wrote: Hi all, In Bash /dev/tcp/host/port can be used to write to a TCP socket. This works nicely so I was very curious whether it would work the other way too: is it possible to have a Bash script listen on a particular port as if it were a server? I couldn't find anything in the Bash manual about it. Google does find a few examples but they all use nc. But that's cheating! ;-) Is it possible with just Bash, no extra tools? (If yes, please enlighten me as to how, obviously I could not get it to work.) ... and some would even say using bash to begin with is cheating.
[gentoo-user] /bin/loadkeys not found after emerge -DuN world
Hi, I've completed my updates of this older Mac Mini. The machine boots and runs fine but there's a small problem in the boot console: * Loading key mappings * /bin/loadkeys not found ** ERROR: cannot start consolefont as keymaps could not start None of this occurred prior to the emerge -DuN world. I tried using equery belongs /bin/loadkeys but it didn't find anything, There seem to be a number of bugs that discuss this which are a bit beyond my understanding of what to do, like this one: http://bugs.gentoo.org/show_bug.cgi?id=232072 I guess it's caused by some issue in baselayout and may be fixed in a newer version. I can wait for the new version to become stable if this won't cause any big problems but I cannot evaluate the severity of this on my own. Comments? Can I leave it alone and not worry about problems? Thanks, Mark
Re: [gentoo-user] /bin/loadkeys not found after emerge -DuN world
On Fri, 2009-01-16 at 07:24 -0800, Mark Knecht wrote: Hi, I've completed my updates of this older Mac Mini. The machine boots and runs fine but there's a small problem in the boot console: * Loading key mappings * /bin/loadkeys not found ** ERROR: cannot start consolefont as keymaps could not start None of this occurred prior to the emerge -DuN world. I tried using equery belongs /bin/loadkeys but it didn't find anything, There seem to be a number of bugs that discuss this which are a bit beyond my understanding of what to do, like this one: My loadkeys is in /usr/bin http://bugs.gentoo.org/show_bug.cgi?id=232072 I guess it's caused by some issue in baselayout and may be fixed in a newer version. I can wait for the new version to become stable if this won't cause any big problems but I cannot evaluate the severity of this on my own. Comments? Can I leave it alone and not worry about problems? A word of advice, when a bugzilla is resolved as DUPLICATE you should (almost always) immediately click on the bug that it's a duplicate of because that's where all the action is taking place: http://bugs.gentoo.org/show_bug.cgi?id=215496 HTH, -a
Re: [gentoo-user] /bin/loadkeys not found after emerge -DuN world
On Fri, Jan 16, 2009 at 7:32 AM, Albert Hopkins mar...@letterboxes.org wrote:
On Fri, 2009-01-16 at 07:24 -0800, Mark Knecht wrote:
Hi,
I've completed my updates of this older Mac Mini. The machine boots
and runs fine but there's a small problem in the boot console:
* Loading key mappings
* /bin/loadkeys not found
** ERROR: cannot start consolefont as keymaps could not start
None of this occurred prior to the emerge -DuN world.
I tried using equery belongs /bin/loadkeys but it didn't find
anything, There seem to be a number of bugs that discuss this which
are a bit beyond my understanding of what to do, like this one:
My loadkeys is in /usr/bin
Interesting difference. I'm setting up this Power PC based MacMini to
be a secondary MythTV backend server. My x86 backend server agrees
with you:
x86
Sector9 ~ # slocate loadkeys
/usr/share/man/man1/loadkeys.1.bz2
/bin/loadkeys
Sector9 ~ #
ppc
MacMini ~ # slocate loadkeys
/usr/bin/loadkeys
/usr/share/man/man1/loadkeys.1.bz2
MacMini ~ #
That file seems to be supplied by kbd:
MacMini ~ # equery belongs /usr/bin/loadkeys
[ Searching for file(s) /usr/bin/loadkeys in *... ]
sys-apps/kbd-1.14.1-r1 (/usr/bin/loadkeys)
MacMini ~ #
However the keymaps initscript comes from baselayout?
MacMini ~ # equery belongs /etc/init.d/keymaps
[ Searching for file(s) /etc/init.d/keymaps in *... ]
sys-apps/baselayout-1.12.11.1 (/etc/init.d/keymaps)
MacMini ~ #
It seems that my current mythbackend machine has an older kbd package:
Sector9 ~ # emerge -pv baselayout kbd
These are the packages that would be merged, in order:
Calculating dependencies... done!
[ebuild R ] sys-apps/baselayout-1.12.11.1 USE=unicode -bootstrap
-build -static 0 kB
[ebuild R ] sys-apps/kbd-1.13-r1 USE=nls 0 kB
Total: 2 packages (2 reinstalls), Size of downloads: 0 kB
Sector9 ~ #
http://bugs.gentoo.org/show_bug.cgi?id=232072
I guess it's caused by some issue in baselayout and may be fixed in
a newer version. I can wait for the new version to become stable if
this won't cause any big problems but I cannot evaluate the severity
of this on my own.
Comments? Can I leave it alone and not worry about problems?
A word of advice, when a bugzilla is resolved as DUPLICATE you should
(almost always) immediately click on the bug that it's a duplicate of
because that's where all the action is taking place:
http://bugs.gentoo.org/show_bug.cgi?id=215496
Actually I should/could have posted that but it seemed to me it didn't
say much of anything. I thought the one I posted had more info.
Anyway, I see the older version is in portage so I suppose I should
jsut go back to that for now?
MacMini ~ # eix -I kbd
[I] sys-apps/kbd
Available versions: 1.12-r8 1.13-r1 1.14.1-r1 ~1.15 {nls}
Installed versions: 1.14.1-r1(14:18:42 01/15/09)(nls)
Homepage:http://freshmeat.net/projects/kbd/
Description: Keyboard and console utilities
MacMini ~ #
Thanks,
Mark
Re: [gentoo-user] How do I change MSS separately from MTU?
On Wednesday 14 January 2009, Mick wrote: On Monday 12 January 2009, Walter Dnes wrote: The only suggestion I've found via Google is iptables mangle. Does it manage to change MSS without changing MTU? If so, what is the invocation in the mangle table? It would probably be something like: iptables --insert OUTPUT --jump TCPMSS --protocol tcp --set-mss 1408 Oops! I just checked the manual: === TCPMSS This target allows to alter the MSS value of TCP SYN packets, to control the maximum size for that connection (usually limiting it to your outgoing interface's MTU minus 40). Of course, it can only be used in conjunction with -p tcp. It is only valid in the *mangle* table. === Then the rule can be set as follows: === iptables -t mangle -A FORWARD -p tcp --tcp-flags SYN,RST SYN \ -j TCPMSS --set-mss 1408 === If you have forwarding disabled on your box I would try the OUTPUT chain instead of FORWARD and see what this gets you. I think you can also set the advertised (by your machine) MSS for a network using ip route: ip route add 192.168.1.0/24 dev eth0 advmss 1408 PS. I am not sure if the above will break your connection because of dropped packets, or how it will interact with the MTU set at 1492. In my case I have just set my MTU at 1492 to cater for the PPP authentication on my ISP's ADSL network. I leave the MSS to be at what the kernel wants it to be - typically MSS = MTU - 40. Hope this helps. -- Regards, Mick signature.asc Description: This is a digitally signed message part.
[gentoo-user] Re: uvesafb - does it require use of initramfs/initrd?
Paul Hartman paul.hartman+gen...@gmail.com writes: I'm ashamed to admit I made the most basic mistake. I compiled uvesafb as a module. Oops! Compiled it as Y instead of M and now I have a pair of Tux sitting atop my kernel boot screen and no more 80x25 horror. :) Is there some difference in uvesafb and vesafb? I've always just ignored the uvesafb choice and used plain vesafb. I just assumed from the name of it and the menuconfig help on it that it was something only usable in `userspace'. I took that to mean after bootup.. something you'd do from the command line. Anyone here that can explain what the difference is.
Re: [gentoo-user] /bin/loadkeys not found after emerge -DuN world
On Fri, 2009-01-16 at 07:56 -0800, Mark Knecht wrote: [...] A word of advice, when a bugzilla is resolved as DUPLICATE you should (almost always) immediately click on the bug that it's a duplicate of because that's where all the action is taking place: http://bugs.gentoo.org/show_bug.cgi?id=215496 Actually I should/could have posted that but it seemed to me it didn't say much of anything. I thought the one I posted had more info. If I'm not mistaken, comments #4 and #5 (the last two comments) pertain to you.
Re: [gentoo-user] /bin/loadkeys not found after emerge -DuN world
On Fri, Jan 16, 2009 at 10:36 AM, Albert Hopkins mar...@letterboxes.org wrote: On Fri, 2009-01-16 at 07:56 -0800, Mark Knecht wrote: [...] A word of advice, when a bugzilla is resolved as DUPLICATE you should (almost always) immediately click on the bug that it's a duplicate of because that's where all the action is taking place: http://bugs.gentoo.org/show_bug.cgi?id=215496 Actually I should/could have posted that but it seemed to me it didn't say much of anything. I thought the one I posted had more info. If I'm not mistaken, comments #4 and #5 (the last two comments) pertain to you. Well, yes, but they don't say anything (to me) as best I read. Or maybe I don't understand how the folks that work in those areas talk these days. Basically it seems it's broken for ppc, the folks in that thread understand it's broken, and then what? Those comments were two months ago. I'm lost. I went back to kbd-1.13-r1and the problem is gone but then another problem appears where it cannot map character code 0 to some other value. I've lost that message for now as I start MythTV automatically and cannot get back in the console becuase of X messages but it's in the bug system somewhere. I found it earlier this morning. Anyway, I think the two sides (x86/amd64 ppc) aren't consistently handled and that these folks know it but haven't fixed it for some reason. I only ran into it because of an update to a machine that's been gathering dust. - Mark
[gentoo-user] Build failure hwinfo both stable and testing
I tried to install hwinfo today and failed with the latest
~hwinfo-14.19. I couldn't make anything usefull of the failure
message so backed off to the stable version 13.28. I got the same
failure so wondering if anyone can make sense of the (partial)
output below.
Does it mean this file is missing (i10_v86.o):
make[2]: *** [i10_v86.o] Error 1
Or do the earlier error message indicate something wrong in the code?
Or something completely differnet?
These pkgs were installed today... they were pulled in by
hwinfo:
Fri Jan 16 12:16:31 2009 dev-libs/libusb-0.1.12-r4
Fri Jan 16 12:17:22 2009 sys-fs/sysfsutils-2.1.0
Fri Jan 16 12:17:48 2009 dev-util/gperf-3.0.3
Fri Jan 16 12:18:12 2009 sys-apps/eject-2.1.5-r1
Fri Jan 16 12:18:26 2009 sys-apps/dmidecode-2.10
Fri Jan 16 12:18:35 2009 app-arch/rpm2targz-9.0.0.3g
Fri Jan 16 12:19:25 2009 x11-terms/xterm-239
Fri Jan 16 12:20:04 2009 sys-fs/device-mapper-1.02.28
Fri Jan 16 12:20:38 2009 sys-apps/usbutils-0.73
Fri Jan 16 12:21:32 2009 sys-fs/cryptsetup-1.0.6-r2
Fri Jan 16 12:22:03 2009 x11-apps/xinit-1.0.8-r3
Fri Jan 16 12:32:20 2009 x11-base/xorg-server-1.5.3-r1
Fri Jan 16 12:32:53 2009 x11-drivers/xf86-input-keyboard-1.3.2
Fri Jan 16 12:33:25 2009 x11-drivers/xf86-input-evdev-2.1.1
Fri Jan 16 12:34:00 2009 x11-drivers/xf86-video-vmware-10.16.5
Fri Jan 16 12:34:38 2009 x11-drivers/xf86-input-mouse-1.4.0
Fri Jan 16 12:35:20 2009 dev-python/pyxf86config-0.3.34-r2
Fri Jan 16 12:37:58 2009 sys-apps/hal-0.5.11-r6
Fri Jan 16 12:38:14 2009 app-misc/hal-info-20080508
emerge hwinfo output
[...]
i10_v86.c: In function 'setup_vm86':
i10_v86.c:104: error: 'VIF_MASK' undeclared (first use in this function)
i10_v86.c:104: error: (Each undeclared identifier is reported only once
i10_v86.c:104: error: for each function it appears in.)
i10_v86.c:104: error: 'VIP_MASK' undeclared (first use in this function)
i10_v86.c: In function 'run_bios_int':
i10_v86.c:474: error: 'VIF_MASK' undeclared (first use in this function)
i10_v86.c:475: error: 'IF_MASK' undeclared (first use in this function)
i10_v86.c:486: error: 'TF_MASK' undeclared (first use in this function)
i10_v86.c:486: error: 'NT_MASK' undeclared (first use in this function)
make[2]: *** [i10_v86.o] Error 1
make[2]: Leaving directory
`/var/tmp/portage/sys-apps/hwinfo-13.28/work/hwinfo-13.28/src/int10'
make[1]: *** [subdirs] Error 2
make[1]: Leaving directory
`/var/tmp/portage/sys-apps/hwinfo-13.28/work/hwinfo-13.28/src'
make: *** [subdirs] Error 2
*
* ERROR: sys-apps/hwinfo-13.28 failed.
* Call stack:
* ebuild.sh, line 49: Called src_compile
* environment, line 2092: Called die
* The specific snippet of code:
* emake -j1 EXTRA_FLAGS=${CFLAGS} || die emake failed
* The die message:
* emake failed
Re: [gentoo-user] /bin/loadkeys not found after emerge -DuN world
On Fri, 2009-01-16 at 10:59 -0800, Mark Knecht wrote: [...] If I'm not mistaken, comments #4 and #5 (the last two comments) pertain to you. Well, yes, but they don't say anything (to me) as best I read. Or maybe I don't understand how the folks that work in those areas talk these days. I believe what comment #4 is saying is that while, sys-apps/baselayout-1.12.12 fixes the issue that version is not in ppc stable. I believe what vapier is saying, albeit crudely and tersely, in comment #5 is that he did is job (which was to fix the bug) and if you need the fix in stable then you need to file a ppc stabilization bug. Presumably vapier doesn't handle ppc stabilization bugs so he's passing the buck. Basically it seems it's broken for ppc, the folks in that thread understand it's broken, and then what? Those comments were two months ago. If the author of comment #4 hasn't already done so, I would recommend that you file a ppc stabilization bug for baselayout-1.12.12. I'm lost. HTH I went back to kbd-1.13-r1and the problem is gone but then another problem appears where it cannot map character code 0 to some other value. I've lost that message for now as I start MythTV automatically and cannot get back in the console becuase of X messages but it's in the bug system somewhere. I found it earlier this morning. Ok... but that's a separate issue... Anyway, I think the two sides (x86/amd64 ppc) aren't consistently handled and that these folks know it but haven't fixed it for some reason. I only ran into it because of an update to a machine that's been gathering dust. Not wanting to go in to a long discussion (too late for that I guess) there is basically the regular dev folks, who presumably vapier belongs to, and then there are the arch teams (the above is 3 sides, actually; not two) who handle testing and stabilization on a particular platform. These are not always the same people and, as often the case in the Gentoo development world, one side doesn't necessarily know/care what the other side is doing. Even between the arch teams there is little consistency i.e. the x86 arch team doesn't care what's happening in the ppc world. Vapier handles the software, but he doesn't necessarily mess with the ACCEPT_KEYWORDS variable. That's why he's requesting you open another bug for that. -a
Re: [gentoo-user] Re: uvesafb - does it require use of initramfs/initrd?
On Fri, Jan 16, 2009 at 12:33 PM, rea...@newsguy.com wrote: Paul Hartman paul.hartman+gen...@gmail.com writes: I'm ashamed to admit I made the most basic mistake. I compiled uvesafb as a module. Oops! Compiled it as Y instead of M and now I have a pair of Tux sitting atop my kernel boot screen and no more 80x25 horror. :) Is there some difference in uvesafb and vesafb? I've always just ignored the uvesafb choice and used plain vesafb. I just assumed from the name of it and the menuconfig help on it that it was something only usable in `userspace'. I took that to mean after bootup.. something you'd do from the command line. Anyone here that can explain what the difference is. According to the website: uvesafb is a generic framebuffer driver for Linux systems and the direct successor of vesafb-tng. Its main features are: * works on non-x86 systems, * the Video BIOS code is run in userspace by a helper application, * can be compiled as a module, * adjustable refresh rates with VBE 3.0-compliant graphic cards. It also enumerates all of the supported modes when you cat /sys/class/graphics/fb0/modes which is handy... no need for vga=0x382 or whatever. They are nice human-readable modes lik 1024x768-60 or whatever. You can also disable the framebuffer entirely or change modes from the commandline once the system is up and running (maybe vesafb lets you do that too, I'm not sure). Now I just need to find a good consolefont that doesn't look squished in 16:9 aspect ratio. Right now I'm using ter-112n (from terminus-fonts) and it's pretty good but still a little too wide for my taste. Paul
Re: [gentoo-user] Build failure hwinfo both stable and testing
On Fri, Jan 16, 2009 at 1:04 PM, rea...@newsguy.com wrote:
I tried to install hwinfo today and failed with the latest
~hwinfo-14.19. I couldn't make anything usefull of the failure
message so backed off to the stable version 13.28. I got the same
failure so wondering if anyone can make sense of the (partial)
output below.
Does it mean this file is missing (i10_v86.o):
make[2]: *** [i10_v86.o] Error 1
Or do the earlier error message indicate something wrong in the code?
Or something completely differnet?
These pkgs were installed today... they were pulled in by
hwinfo:
Fri Jan 16 12:16:31 2009 dev-libs/libusb-0.1.12-r4
Fri Jan 16 12:17:22 2009 sys-fs/sysfsutils-2.1.0
Fri Jan 16 12:17:48 2009 dev-util/gperf-3.0.3
Fri Jan 16 12:18:12 2009 sys-apps/eject-2.1.5-r1
Fri Jan 16 12:18:26 2009 sys-apps/dmidecode-2.10
Fri Jan 16 12:18:35 2009 app-arch/rpm2targz-9.0.0.3g
Fri Jan 16 12:19:25 2009 x11-terms/xterm-239
Fri Jan 16 12:20:04 2009 sys-fs/device-mapper-1.02.28
Fri Jan 16 12:20:38 2009 sys-apps/usbutils-0.73
Fri Jan 16 12:21:32 2009 sys-fs/cryptsetup-1.0.6-r2
Fri Jan 16 12:22:03 2009 x11-apps/xinit-1.0.8-r3
Fri Jan 16 12:32:20 2009 x11-base/xorg-server-1.5.3-r1
Fri Jan 16 12:32:53 2009 x11-drivers/xf86-input-keyboard-1.3.2
Fri Jan 16 12:33:25 2009 x11-drivers/xf86-input-evdev-2.1.1
Fri Jan 16 12:34:00 2009 x11-drivers/xf86-video-vmware-10.16.5
Fri Jan 16 12:34:38 2009 x11-drivers/xf86-input-mouse-1.4.0
Fri Jan 16 12:35:20 2009 dev-python/pyxf86config-0.3.34-r2
Fri Jan 16 12:37:58 2009 sys-apps/hal-0.5.11-r6
Fri Jan 16 12:38:14 2009 app-misc/hal-info-20080508
emerge hwinfo output
[...]
i10_v86.c: In function 'setup_vm86':
i10_v86.c:104: error: 'VIF_MASK' undeclared (first use in this function)
i10_v86.c:104: error: (Each undeclared identifier is reported only once
i10_v86.c:104: error: for each function it appears in.)
i10_v86.c:104: error: 'VIP_MASK' undeclared (first use in this function)
i10_v86.c: In function 'run_bios_int':
i10_v86.c:474: error: 'VIF_MASK' undeclared (first use in this function)
i10_v86.c:475: error: 'IF_MASK' undeclared (first use in this function)
i10_v86.c:486: error: 'TF_MASK' undeclared (first use in this function)
i10_v86.c:486: error: 'NT_MASK' undeclared (first use in this function)
make[2]: *** [i10_v86.o] Error 1
make[2]: Leaving directory
`/var/tmp/portage/sys-apps/hwinfo-13.28/work/hwinfo-13.28/src/int10'
make[1]: *** [subdirs] Error 2
make[1]: Leaving directory
`/var/tmp/portage/sys-apps/hwinfo-13.28/work/hwinfo-13.28/src'
make: *** [subdirs] Error 2
*
* ERROR: sys-apps/hwinfo-13.28 failed.
* Call stack:
* ebuild.sh, line 49: Called src_compile
* environment, line 2092: Called die
* The specific snippet of code:
* emake -j1 EXTRA_FLAGS=${CFLAGS} || die emake failed
* The die message:
* emake failed
I just emerged sys-apps/hwinfo-14.19 and it worked for me. I'm on
~amd64, default/linux/amd64/2008.0/desktop, gcc-4.3.2,
glibc-2.9_p20081201-r1, 2.6.28-gentoo x86_64.
[gentoo-user] Append string on Kernel builds
In the first section during a `makeconfig' session, there is a line (the second one) that says: Local Version - append to kernel release I like to use that and put `-$MYHOST' as string. I wondered if there is any way to set a numericly incrementing string. Maybe some trick syntax that can go in that spot?
Re: [gentoo-user] Append string on Kernel builds
Am Freitag, 16. Januar 2009 21:43:56 schrieb rea...@newsguy.com: In the first section during a `makeconfig' session, there is a line (the second one) that says: Local Version - append to kernel release I like to use that and put `-$MYHOST' as string. I wondered if there is any way to set a numericly incrementing string. Maybe some trick syntax that can go in that spot? The build system does that automatically as long as you don't make mrproper, see uname -a output. HTH... Dirk signature.asc Description: This is a digitally signed message part.
Re: [gentoo-user] Append string on Kernel builds
On Friday January 16 2009 18:58:55 Dirk Heinrichs wrote: The build system does that automatically as long as you don't make mrproper, You can backup .version
[gentoo-user] Re: Append string on Kernel builds
Dirk Heinrichs dirk.heinri...@online.de writes: I like to use that and put `-$MYHOST' as string. I wondered if there is any way to set a numericly incrementing string. Maybe some trick syntax that can go in that spot? The build system does that automatically as long as you don't make mrproper, see uname -a output. No.. its not the same as what I'm talking about. When you set the item in menuconfig: General Setup/Local Version [...] The string you set there is appended to that actual build product like vmlinuz-2.6.26-gentoo-$HOST The vmlinuz that gets sent to /boot/ when you say `make install' is named that way, along with the config-XXX and System-XXX that is moved there. Keeps things kind of tidy in /boot/ if you are mucking around with several kernels. What I asked was if there is some tricky syntax I could use on that kernel setting that would do: linux-2.6.26-gentoo-$HOST-N Where N is an incremented number every time I build the kernel without running `mrproper'.
[gentoo-user] Re: madwifi Stuck beacon causes mpd to skip
Whenever I get the following message in dmesg: wifi0: ath_bstuck_tasklet: Stuck beacon; resetting (beacon miss count: 11) the music playing on mpd skips. Does anyone know more about this? - Grant I think this is fixed by downgrading from madwifi-ng-svn- to madwifi-ng-svn-3876. Both are from the berkano overlay. - Grant Had stuck beacon problems there too, had to downgrade to madwifi-ng and disable SMP in the kernel since they aren't compatible when madwifi-ng is in master mode. - Grant
[gentoo-user] Disable remote login for certain user
One of the users set up on my router is for whoever is sitting in front of the router and wants to log in. For that reason, the password needs to be simple and I'd like to prevent that user from being able to log in if they aren't in front of the system since the password is simple and should be easy to hack. Should I do that via an ssh config setting, in shorewall, or somewhere else? - Grant
Re: [gentoo-user] Restricting Firefox website access
That sounds good, how can I do that? iptables module owner handles that stuff, just man iptables if you'll have any trouble. iptables -A OUTPUT -m owner --uid-owner someuser -m tcp --dport http -j REJECT I brought this to the shorewall list for config advice, but I was told: a) NO PACKET FILTERING FIREWALL (which includes Shorewall) has any notion of domains. So filterinG by domain is a non-starter. b) When referring to packet filters, filtering by user id (e.g., root) can only be done for connections originating from the firewall. See man shoreall-rules and read about the USER/GROUP column. Here was my original request: I'd like to restrict the websites one of the computers on my network can access in Firefox. It only needs to access 2 different domain names and I don't want it to be able to access any others. I can restrict it at the router if necessary because the router is a Gentoo system. I think this leaves a squid proxy setup as my only option? - Grant
Re: [gentoo-user] Restricting Firefox website access
On Fri, 16 Jan 2009 21:34:59 -0800 Grant emailgr...@gmail.com wrote: I think this leaves a squid proxy setup as my only option? Sorry, I haven't noticed the fact that there are machines behind the firewall that need to be restricted, and aforementioned rule certainly won't do that. Squid setup should certainly be a solid solution to the problem. It should also save quite a lot of traffic and speed up browsing via common cache. You can actually disable nat on the firewall if there are no specific software requiments that can't work with http proxy, which are quite rare, with the exception of games and p2p software. And since you're using gentoo you can also pass rsync traffic through a proxy. Rsync (as well as wget and lots of other tools) will use proxy automatically if RSYNC_PROXY (http_proxy/ftp_proxy for other apps, lower- and uppercase) env var is set. For squid to pass rsync traffic you'll need to specify rsync ports in squid.conf, like this: acl SSL_ports port 873 # rsync acl Safe_ports port 873 # rsync -- Mike Kazantsev // fraggod.net signature.asc Description: PGP signature
Re: [gentoo-user] Disable remote login for certain user
On Fri, 16 Jan 2009 21:28:07 -0800 Grant emailgr...@gmail.com wrote: Should I do that via an ssh config setting, in shorewall, or somewhere else? I believe the right way would be to add 'account required pam_access.so' line to /etc/pam.d/system-auth and define login restrictions in /etc/securety/access.conf (it's also quite well documented). That way you'll block ssh/ftp/mail etc logins for that account, which should also be prone to brutforce attacks because of weak password. The catch is, of course, that you should have pam on your system ;) -- Mike Kazantsev // fraggod.net signature.asc Description: PGP signature
