[gentoo-user] Re: Gentoo for many servers
Alex Schuster wrote: Alan McKinnon wrote: clusterssh will let you log into many machines at once and run emerge -avuND world everywhere This is way cool. I just started using it on eight Fedora servers I am administrating. Nice, now this is an improvement over my 'for $h in $HOSTS; do ssh $h yum install foo; done' approach. You could have a look at app-admin/puppet [1][2] which supposedly takes car of these things. [...] Now I am thinking about a Gentoo installation instead. Pros: - Continuous updates, no downtime for upgrading, only when I decide to install a new kernel. This is really really cool. I fear the upgrade from Fedora 10 to 12 which has to be done soon. - Some improvement in speed. Those machines do A LOT of numbercrunching, which jobs often lasting for days, so even small improvements would be nice. - Easier debugging. When things do not work, I think it's easier to dig into the problem. No fancy, but sometimes buggy GUIs hiding basic functionality. These two things would probably be your best selling points for your idea. - Heck, Gentoo is _cooler_ than typical distributions. And emerging with distcc on about 8*4 cores would be fun :) Being 'cool' doesn't count, at least last time I looked. - I am probably the only one who can administrate them. That is a huge disadvantage. Cons: - If something will not work with this not so common (meta)distribution, people will say always trouble with your Gentoo Schmentoo, it works fine in Fedora. Fedora is more mainstream, if something does not work there, then it's okay for the people to accept it. - I fear that big packages like Matlab are made for and tested on the typical distributions, and may have problems with the not-so-common Gentoo. I think someone here just had such a problem with Mathematica (which we do currently not use). [...] If you're using commercial software which is only supported by Redhat, Novell, etc. then you should think twice about replacing it. But I'm guessing that those packages don't have to be installed on every machine. So, I'd suggest that you use Gentoo on those boxes where you'd have the biggest advantage using it and no or minimal disadvantages. - I am probably the only one who can administrate them. I think Gentoo is easier to maintain in the long run, but only when you take the time to learn it. With Fedora, you do not need much more than the 'yum install' command. There is no need to read complicated X.org upgrade guides and such. [...] Please do your colleagues and successors a favor and document your whole setup really good. Regards, Andi [1] http://reductivelabs.com/products/puppet/ [2] http://log.onthebrink.de/2008/05/using-puppet-on-gentoo.html
Re: [gentoo-user] simple firewall
Hi, gigli wrote: Hi I wonder if there is any easy firewall for gentoo. I tried ubuntu for a while and used their ufw, which was very simple. My needs: Block incoming traffic except for sshd and https (and sometimes bittorrent) and allow my lan to connect to my samba share, mythtv and mysql when i use openvpn or allways, which would be easyist. My box is usually protected by pfsense. net-firewall/firehol is a fairly light-weight iptables rule generator. You just have to specify which services to allow and in some cases protocol and portnumber for services unknown to firehol. Regards, Andi
Re: [gentoo-user] perfect IDE
Hi, Andrei Hanganu wrote: helo group, i've been trying the past 2-3 years to find the most usable and nice ide for c/c++ code writing. I've been through vim/vim + plugins/emacs + different modes/anjuta/kdevelop/codeblocks/eclipse/netbeans ... every single one of them has at least one drawback. I'm thinking the more I get to know Vim and the available plugins, the more it becomes like an IDE to me. I guess the same is true for Emacs. My advice would be to take on of those or any other open IDE and learn and extend them to the point that it's perfect for you. Now for your feature requirements list I'm going to concentrate on Vim and Emacs as those two are the ones I know. In short words, i am looking for an ide that can do this: - syntax highlighting - concurrent editing of multiple files (splitting) - tabs or buffer list - file browser - regex search/replace Both Vim and Emacs can do these basic features. Vim even provides a mechanism for saving and restoring editing sessions. - autocomplete (on the fly, not on demand, and maybe smart? - identify structures/classes ) Haven't tried it yet, but for Vim word_complete.vim[1] seems to be what you're looking for. You should also have a look at Omnicompletion. As Emacs has hooks for nearly everything it should be doable with it as well. - project manager Don't know about that but it would be nice to have simpler project specific settings for Emacs/Vim. - symbol list/browser current editing buffer That's pretty much ctags/etags, maybe cscope. - flexible build options that include scons, not just makefile You can put the following in ~/.vimrc: autocmd BufEnter ~/path/to/project/* set makeprg=scons - code folding (with detection of blocks) Vim does it[2]; Emacs seems to have some kind of FoldingMode according to Google. - lightweight/ergonomic interface (i dislike space being occupied by the bar that displays the line numbers, with a padding of 10px for example) Both of them are very customisable in this regard. i don't desire gdb or valgrind integration, but would be a + Emacs features gdb integration and there's Clewn[3] for GVim. As for me, I'm rather using a separate screen[4] window in the same session. Regards, Andi [1] http://www.vim.org/scripts/script.php?script_id=73 [2] http://www.linux.com/articles/114138 [3] http://clewn.sourceforge.net/ [4] http://www.gnu.org/software/screen/
Re: [gentoo-user] kernel linux-2.6.27-gentoo-r7 won't load network!
Alan McKinnon wrote: On Wednesday 07 January 2009 00:37:31 Denis wrote: I'm doubtful that using oldconfig would make a driver not work because I have done it this way for a while, going between various 2.6.x versions, like 2.6.21 to 2.6.24... And there was never an issue. Maybe the E1000 driver somehow got messed up in this particular version of the kernel. Now, that e1000e driver someone mentioned - how is it different from the e1000 driver, and does the kernel float two versions, or it depends on which kernel it is? e1000e breaks the hardware e1000 does not break the hardware Or maybe it's the other way round Anyway, two different drivers for the same hardware, the broken one was removed from the sources till it gets fixed. You can always build both drivers in a case like this but obviously you can only use one of them at a time Actually, there was a bug with Ftrace overwriting the flash ROM of a NIC (which shouldn't have been mapped RW anyway) with e1000e in certain cases. IIRC this was only present in the 2.6.27 release candidates and got fixed by mapping it read-only for the final release. 2.6.28 then had the fix for the Ftrace bug. Regards, Andi
Re: [gentoo-user] Awesome vs Xmonad
Hi, Man Shankar wrote: Hello, I want to try out the tiling window managers. I would want to know the experiences of the users about awesome and xmonad. Primarily i would like to know which of those two tiling WMs has worked for you guys. The hurdles you encountered and the gains you got thereof. Currently i am a happy e16 user, but the fact that the tiling WMs manage the windows makes me attracted to them. Please comment. I switched from e16 to xmonad last summer and haven't regret it so far. One important thing though is to get used to the tiling paradigm, i.e. letting the wm do all the resize and positioning work. I suggest you try it some time and see if it fits you. Personally I started using it only on my home pc while I kept e16 on the laptop for work until I couldn't resist a complete switch to it anymore. I've recently also started using awesome in a few virtual machines, mainly due to the large size of the xmonad dependencies (GHC takes up quite some space). From my point of view they both look fairly the same with awesome having a few more features (tagging, widgets). It also helps to regard the configuration file (xmonad and = awesome-3.0) as the main program, e.g. my xmonad.hs looks a bit like a Haskell program where different modules get imported and the main window manager module loaded at the end. You can do quite a lot with those two. Aside from that, the main difference between them are the programming languages they're written in because you have to use it for the configuration file. Awesome uses Lua which is a simple but powerful imperative scripting language and xmonad uses Haskell, an advanced functional programming language which many consider as rather hard to learn. Personally, I didn't know anything about Haskell before using xmonad and I have to admit that I had a few very hard times with it when I wanted to do some advanced (or even simple) configuration changes. But once you wind your head around the functional paradigm (and all those operators and monads) you can do a lot with it. Have a look at the xmonad config archive[1] for some examples. If you're going to use awesome I'd recommend having a look at x11-misc/dmenu as I didn't see any default integration of it in the awesome config (though I might have missed it). Regards, Andi [1] http://www.haskell.org/haskellwiki/Xmonad/Config_archive
Re: [gentoo-user] xf86-video-ati-6.8.0-r1 problems
Mick wrote: On Friday 18 July 2008, Andreas Niederl wrote: [...] I'm just guessing here, but maybe your kernel is missing DRM support for your graphics chip. What's the output of grep -i radeon /boot/config-$(uname -r) ? # grep -i radeon /boot/config-$(uname -r) CONFIG_FB_RADEON=y CONFIG_FB_RADEON_I2C=y CONFIG_FB_RADEON_BACKLIGHT=y # CONFIG_FB_RADEON_DEBUG is not set CONFIG_DRM and CONFIG_DRM_RADEON are needed for direct rendering. The respective options are found in Device Drivers - Character Devices. Regards, Andi -- gentoo-user@lists.gentoo.org mailing list
Re: [gentoo-user] xf86-video-ati-6.8.0-r1 problems
Andreas Niederl wrote: Mick wrote: [...] # grep -i radeon /boot/config-$(uname -r) CONFIG_FB_RADEON=y CONFIG_FB_RADEON_I2C=y CONFIG_FB_RADEON_BACKLIGHT=y # CONFIG_FB_RADEON_DEBUG is not set CONFIG_DRM and CONFIG_DRM_RADEON are needed for direct rendering. The respective options are found in Device Drivers - Character Devices. Actually, they moved to Device Drivers - Graphics Support - Direct Rendering Manager in recent kernels. Regards, Andi -- gentoo-user@lists.gentoo.org mailing list
Re: [gentoo-user] xf86-video-ati-6.8.0-r1 problems
Mick wrote: On Thursday 17 July 2008, Andrew Tchernoivanov wrote: # modprobe -v radeon_drv FATAL: Module radeon_drv not found. That's strange. But try one more thing - copy this radeon_drv from /usr/lib/xorg/modules/drivers to /lib/modules/kernel version/kernel/drivers/video Because according to modprobe's man page, by default it will look driver in /lib/modules/... So try to copy it and write here, what will modprobe say First I linked it: # ls -la /lib/modules/2.6.24-gentoo-r8/kernel/drivers/video/ total 8 drwxr-xr-x 3 root root 144 Jul 17 22:21 . drwxr-xr-x 14 root root 344 May 14 18:34 .. drwxr-xr-x 2 root root 72 May 14 18:34 backlight -rw-r--r-- 1 root root 4291 May 14 18:34 output.ko lrwxrwxrwx 1 root root 43 Jul 17 22:21 radeon.so - /usr/lib/xorg/modules/drivers/radeon_drv.so but since that did not work, I copied it: # ls -la /lib/modules/2.6.24-gentoo-r8/kernel/drivers/video/ total 448 drwxr-xr-x 3 root root176 Jul 17 22:27 . drwxr-xr-x 14 root root344 May 14 18:34 .. drwxr-xr-x 2 root root 72 May 14 18:34 backlight -rw-r--r-- 1 root root 4291 May 14 18:34 output.ko lrwxrwxrwx 1 root root 43 Jul 17 22:21 radeon.so - /usr/lib/xorg/modules/drivers/radeon_drv.so -rwxr-xr-x 1 root root 448808 Jul 17 07:09 radeon_drv.so Modprobing either of the two fails with the same error message. That's because this is not a kernel module. drmOpenDevice: node name is /dev/dri/card0 drmOpenDevice: open result is -1, (No such device or address) drmOpenDevice: open result is -1, (No such device or address) drmOpenDevice: Open failed [drm] failed to load kernel module radeon (EE) RADEON(0): [dri] RADEONDRIGetVersion failed to open the DRM [dri] Disabling DRI. I'm just guessing here, but maybe your kernel is missing DRM support for your graphics chip. What's the output of grep -i radeon /boot/config-$(uname -r) ? Regards, Andi -- gentoo-user@lists.gentoo.org mailing list
Re: [gentoo-user] libselinux.so.1 dependency problems
Hi, Vladimir G. Ivanovic wrote: [...] It turns out that many, many executables require libselinux.so.1, despite what the documentation of --depclean in man emerge says (or what I think it says -- is this a bug or operator error?) Sadly sys-apps/coreutils is one of them. Recent versions - including stable - do an autodetection for libselinux and link against it even when emerged with USE=-selinux[1]. This should be no problem for systems which never saw libselinux (i.e. installed from 2008.0) but unmerging this library on older systems can be quite problematic. I cobbled together a system that limps along thanks to the 2008.0 beta LiveCD (which does not depend on libselinux.so.1), but I am unable to emerge a large number of packages that seem to silently depend on libselinux.so.1: the ebuilds fail when ld cannot find -lselinux. [...] What gives? Where does the -lselinux come from? How can I get rid of this maddening dependency? I think that libtool is the main offender here. At least on my system somehow '-lselinux' made its way into a bunch of .la files and provoked these errors. So I searched for the packages with broken libtool archive files and manually emerged them (with --oneshot). I figured out the correct order by using the trial-and-error method but you could do something like the get_build_order() function in the revdep-rebuild script. The command I've used for searching is as follows (requires app-portage/portage-utils): grep -l -r --include='*.la' selinux /usr/lib | qfile --nocolor -f - | \ cut -d' ' -f 1 | sort | uniq Another way might be to look at the line before the error message and rebuild the package containing the library right before the '-lselinux' flag. hth, Andi [1] http://bugs.gentoo.org/show_bug.cgi?id=230073 -- gentoo-user@lists.gentoo.org mailing list
Re: [gentoo-user] Minimal Gentoo installation
Hi, Petar Dimitrijevic wrote: Hi ppl, My basic idea is to have chroot-ed environment which will be the full system and then to install separate system with only minimal stuff (without gcc, portage, ...). When I need to update the minimal system I will first update the chrooted one and the emerge the updates onto the new one. So I got the stage3 tarball, unpacked it and and chrooted it. I've updated all the packages and started installation of new packages with: ROOT=/install emerge ... I wanted to ask if somebody has done something like this, is something like this possible and are there any wiki's or howto's on this topic. I've tried searching through the handbook and google-ing but had no luck. http://gentoo-wiki.com/TinyGentoo provides a somewhat extensive introduction to this as well as a long list of links to related web pages. Note, that you have to include libstdc++.so yourself if you're not emerging gcc into your /install. If you're not cross-compiling you might want to have a look at http://www.gentoo.org/proj/en/base/embedded/gnap.xml Regards, Andi -- [EMAIL PROTECTED] mailing list
Re: [gentoo-user] Minimal Gentoo installation
Petar Dimitrijevic wrote: Andreas Niederl wrote: [...] Note, that you have to include libstdc++.so yourself if you're not emerging gcc into your /install. You mean just to copy libstdc++.so to /lib or emerge libstdc++ ? [...] Just copy it over to /usr/lib (which is the default location for this library, I believe). There is no general separate libstdc++ package. sys-libs/libstdc++-v3 only provides an older version for backwards compatibility. Regards, Andi -- [EMAIL PROTECTED] mailing list
Re: [gentoo-user] CXXABI error after gcc upgrade
Iain Buchanan wrote: On Mon, 2007-08-13 at 08:02 +0200, Andreas Niederl wrote: Hi, Iain Buchanan wrote: so I ran `revdep-rebuild --library=libstdc++.so.6` but that got stuck on eix: src/eix: /usr/lib/gcc/i686-pc-linux-gnu/3.4.6/libstdc++.so.6: version `CXXABI_1.3.1' not found (required by src/eix) [...] My own little workaround: echo 'LDPATH=/lib:/usr/lib:/usr/local/lib' /etc/env.d/04basic-ldpath env-update I tried that, but eix still fails with the same error... Or do I have to rebuild the system again?! No, eix still takes the wrong libstdc++.so.6 and rebuilding won't likely change that. Whats the output of the libstdc++ section, if you call eix with LD_DEBUG like `LD_DEBUG=libs eix`? You could directly query the exported CXXABI versions of different libstdc++ libraries in the same order as they are tried by the dynamic linker by pasting the following loop into your shell: while read dir ; do if [ -r $dir/libstdc++.so.6 ] ; then echo $dir: readelf -a $dir/libstdc++.so.6 | grep Rev | grep CXXABI fi done /etc/ld.so.conf Regards, Andi -- [EMAIL PROTECTED] mailing list
Re: [gentoo-user] CXXABI error after gcc upgrade
Iain Buchanan wrote: On Wed, 2007-08-15 at 13:23 +0200, Andreas Niederl wrote: Iain Buchanan wrote: On Mon, 2007-08-13 at 08:02 +0200, Andreas Niederl wrote: [...] My own little workaround: echo 'LDPATH=/lib:/usr/lib:/usr/local/lib' /etc/env.d/04basic-ldpath env-update I tried that, but eix still fails with the same error... Or do I have to rebuild the system again?! No, eix still takes the wrong libstdc++.so.6 and rebuilding won't likely change that. Whats the output of the libstdc++ section, if you call eix with LD_DEBUG like `LD_DEBUG=libs eix`? the current eix (0.9.9) shows this (and a whole lotta other stuff): $ LD_DEBUG=libs eix 18778: find library=libstdc++.so.6 [0]; searching 18778: search path=/usr/lib/fglrx (LD_LIBRARY_PATH) 18778: trying file=/usr/lib/fglrx/libstdc++.so.6 18778: search cache=/etc/ld.so.cache 18778: trying file=/usr/lib/gcc/i686-pc-linux-gnu/3.4.6/libstdc++.so.6 [snip] 18778: calling init: /usr/lib/gcc/i686-pc-linux-gnu/3.4.6/libstdc++.so.6 You could directly query the exported CXXABI versions of different libstdc++ libraries in the same order as they are tried by the dynamic linker by pasting the following loop into your shell: while read dir ; do if [ -r $dir/libstdc++.so.6 ] ; then echo $dir: readelf -a $dir/libstdc++.so.6 | grep Rev | grep CXXABI fi done /etc/ld.so.conf /usr/lib/gcc/i686-pc-linux-gnu/3.4.6: 0x00a4: Rev: 1 Flags: none Index: 6 Cnt: 1 Name: CXXABI_1.3 /usr/lib/gcc/i686-pc-linux-gnu/4.1.2: 0x0158: Rev: 1 Flags: none Index: 11 Cnt: 1 Name: CXXABI_1.3 0x0174: Rev: 1 Flags: none Index: 12 Cnt: 2 Name: CXXABI_1.3.1 /usr/lib/gcc/i686-pc-linux-gnu/4.2.0: 0x017c: Rev: 1 Flags: none Index: 12 Cnt: 1 Name: CXXABI_1.3 0x0198: Rev: 1 Flags: none Index: 13 Cnt: 2 Name: CXXABI_1.3.1 But what does it all _mean_ Basil? (gratuitous bad movie quote...) [...] Just that the first libstdc++.so.6 found by ld.so doesn't have CXXABI_1.3.1 which is apparently needed by eix. If you look into your /etc/ld.so.conf you'll see some directories listed which ld.so will probe from the top downwards when it searches for dynamically linked libraries. In your case the gcc library directories are in the wrong order. My guess is that you should reverse them. The search order is specified by the LDPATH variables in the /etc/env.d/* files. env-update reads these files sorted alphabetically concatenates the found LDPATHs and writes it into /etc/ld.so.conf. The LDPATH for gcc should be in /etc/env.d/05gcc, I imagine something like LDPATH=/usr/lib/gcc/i686-pc-linux-gnu/3.4.6/:/usr/lib/gcc/i686-pc-linux-gnu/4.1.2/:/usr/lib/gcc/i686-pc-linux-gnu/4.2.0/ standing in your version of this file. For reversing the search order simply reverse this variable so that it reads: LDPATH=/usr/lib/gcc/i686-pc-linux-gnu/4.2.0/:/usr/lib/gcc/i686-pc-linux-gnu/4.1.2/:/usr/lib/gcc/i686-pc-linux-gnu/3.4.6/ and do a env-update. The reason for the search path being in the wrong order is probably some bug in gcc-config or eselect-compiler, whichever you are using. At least I can't think of another reason. Regards, Andi -- [EMAIL PROTECTED] mailing list
Re: [gentoo-user] CXXABI error after gcc upgrade
Hi, Iain Buchanan wrote: Hi all, another gcc upgrade issue :) I've read and followed the gcc upgrade gentoo howto, and also a few bugs on b.g.o, but I'm getting stuck... this all went well: # gcc-config i686-pc-linux-gnu-4.1.2 # env-update source /etc/profile # fix_libtool_files.sh 3.4.6 # emerge --oneshot -av libtool then I did an emerge -eav system, also went well. Then I started upgrading, but came into this error: /usr/lib/gcc/i686-pc-linux-gnu/3.4.6/libstdc++.so.6: version `CXXABI_1.3.1' not found so I ran `revdep-rebuild --library=libstdc++.so.6` but that got stuck on eix: src/eix: /usr/lib/gcc/i686-pc-linux-gnu/3.4.6/libstdc++.so.6: version `CXXABI_1.3.1' not found (required by src/eix) libstdc++so.6 belongs to gcc 3.4.x, so I recompiled gcc 4.1.2 just in case, and then re-ran the entire above sequence, but it gets stuck again at the same place. libstdc++so.6 _still_ belongs to gcc 3.4.x. Is this a problem? [...] Unmerging the old gcc should take care of this. However, if might want to keep a 3.x gcc (e.g. app-emulation/qemu needs it to build), have a look at your /etc/ld.so.conf. Lately, I noticed on a Gentoo x86 machine with both gcc versions 3.4.x and 4.1.x installed that this file was a bit messed up, i.e. the old gcc libs got preferred over /lib /usr/lib etc. (though that shouldn't happen on an amd64 multilib system). My own little workaround: echo 'LDPATH=/lib:/usr/lib:/usr/local/lib' /etc/env.d/04basic-ldpath env-update I kinda remember wanting to search for existing bugs / opening a new report, but it seems I've forgotten about that. Regards, Andi -- [EMAIL PROTECTED] mailing list