Re: AMD64-generic doesn't see all 4GB RAM?
Lennart Sorensen wrote: On Mon, Nov 06, 2006 at 05:54:00PM +0100, A J Stiles wrote: There's your problem; you're still running the installer kernel. The installer kernel is only supposed to work well enough and for long enough for you to build yourself a new one. Install kernel-package, libncurses5-dev (menuconfig needs it); then you can just get sources from kernel.org, and compile them into a .deb package to install with dpkg -i. Note: unless you're *very* lucky, you *will* at some point turn off something you should have left on and your new kernel won't boot. Save all your config files, have a bootable CD handy, and learn how to use it to alter your LILO or GRUB configuration to boot the installer kernel. If you're still running a stock kernel, you're only using about half the power of Linux . There is almost never a reason to not run one of debian's prebuilt kernels. They work perfectly and optimally for probably 99% of users. The 3.2GB problem has to do with memory remapping which is a BIOS problem. The etch installer is quite good at installing the optimal kernel for the system. I'm running a stock kernel on a Sun Fire V40z (4 x Opteron 852) with 16 gigs of RAM - the kernel sees all 16 gigs just fine. Regards, Ozz. signature.asc Description: OpenPGP digital signature
Lightweight GUIs [was Re: sources.list ???]
Dean Hamstead wrote: enlightenment is awesome. nor bars or docks or loading bays etc just screen real estate. Dean I would recommend xfce, it's a lot lighter, faster and works just fine, if colours and things are a must, try enlightenment... IceWM rocks too. Most of the benefits of GUIs such as KDE, without the footprint. Put it this way - I have an old 100MHz Pentium laptop with 16Mb RAM, and it can do IceWM. It's not fast, but it's certainly usable. Put IceWM on a slow machine and it's fast (compared to many GUIs) - put it on a fast machine and it SCREAMS! I run it on my own AMD64 at home... Regards, Austin. signature.asc Description: OpenPGP digital signature
Re: quiet dual dual-core opteron
Brendan Corkery wrote: I have put a few together for engineers where I work. The low power single core units were expensive, but made a huge difference. I used a Quiet PC power supply and case fan, a large Chenbro tower case, and two of the Zalman 7000 (?) CPU sinks. Almost silent except for, yes, the SCSI drive noise. I will probably be doing a few dual core units in the next month and I will be using the new Zalman 9000-something cooler and essentially the same setup. I can list the specific part numbers later today if anyone is curious, but the low noise case fan, power supply, and CPU sinks were key, and the low power (HE, I think) CPUs made all of that possible by not forcing me to move a ton of air. Remember too that these are AMD chips, not Intel. They use a fraction of the juice (and therefore generate a fraction of the heat and require a fraction of the cooling) of the Intel offerings. My dual-Opteron 246 tower is quieter and cooler than my PentiumIII laptop. The Opteron CPUs in this box hover around the 40c mark (~105f). Regards, Ozz. signature.asc Description: OpenPGP digital signature
Re: Mirroring the AMD pool
Steven Dobson wrote: I am thinking about mirroring the AMD64 archive and was wondering how much space was required. The amd64-specific files are about 13 Gigs, but for the mirror to work properly you will also need the binary-independent packages which bump it up to around 43 Gigs. There will be some big changes in the way the mirrors work soon (some arch being split out, others moved in) to help control the size as a full Debian mirror is HUGE. The changes will make it a lot easier for people to mirror specific arch without all the cruft of other arch that they don't need. You may want to hold off until the changes are completed (should be a few weeks). Regards, Ozz. signature.asc Description: OpenPGP digital signature
Re: Mirroring the AMD pool
Steven Dobson wrote: Ozz On Tue, 2006-03-07 at 09:05 -0500, Austin Denyer wrote: There will be some big changes in the way the mirrors work soon (some arch being split out, others moved in) to help control the size as a full Debian mirror is HUGE. The changes will make it a lot easier for people to mirror specific arch without all the cruft of other arch that they don't need. I know it's hugh - that's why I don't mirror it all just the i386 arch ATM. I currently mirror three arch locally: i386, amd64 and SPARC. Unfortunately my mirror is currently off-line due to a RAID failure. (I still believe it's the card, but MegaRAID insist that I really did have THREE drives out of a 6-drive RAID5 fail simultaneously...) You may want to hold off until the changes are completed (should be a few weeks). Being lazy I will hold off so I don't have to do the work twice. :-) Sounds like the way to go. As I need to track what changes I need to make to the rsync script (anonftpsync) for my i386 mirror what is the best list to subscribe to to find out what is going on and what changes I need to make. I'm not sure myself yet what the changes will be. A good list to subscribe to would be the following: debian-mirrors-announce@lists.debian.org debian-mirrors@lists.debian.org See http://lists.debian.org/debian-mirrors-announce/ and http://lists.debian.org/debian-mirrors/ to subscribe. I will forward a copy of the info I have concerning the changes to you off-list. Regards, Ozz. signature.asc Description: OpenPGP digital signature
Re: Mirroring the AMD pool
Lennart Sorensen wrote: Not sure if rsync support is available from the amd64.debian.net server. That is usually the best way to mirror things. I was using a modified version of this script: http://amd64.debian.net/~joerg/sync-debian-amd64.sh It uses rsync. Regards, Ozz. signature.asc Description: OpenPGP digital signature
Re: Can't mount CD/DVD drives
On Fri, 17 Feb 2006 12:44:14 -0800 Andrew Sharp [EMAIL PROTECTED] wrote: On Thu, Feb 16, 2006 at 10:01:15AM +0100, Gabor Gombas wrote: On Wed, Feb 15, 2006 at 03:45:01PM -0500, Austin Denyer wrote: Unless you have the RAID code compiled into the kernel (not modules) and you are using kernel-based RAID autodetection, the /dev/md* entries are created by mdadm, not by udev (udev can not create them until the array is created, and mdadm needs to create the device first to be able to build the array). The raid code doesn't have to be compiled in, it can be modularized in an initrd image. So long as grub can load the kernel and the initrd image, you're good to go with whatever modules you want in the initrd image. I believe current versions of grub can do raid0 and raid5 now, but, well, you know, test it first ~:^) I was running a vanilla Debian kernel. Unfortunately I didn't have time to poke around too much, as I can't take it out of service for very long at a time before the users start complaining. Regards, Ozz. pgpEDFAltrwRz.pgp Description: PGP signature
Re: Can't mount CD/DVD drives
On Wed, 15 Feb 2006 12:28:01 -0800 Andrew Sharp [EMAIL PROTECTED] wrote: On Wed, Feb 15, 2006 at 01:51:54PM -0500, Lennart Sorensen wrote: On Wed, Feb 15, 2006 at 08:39:11AM -0500, Carl Brown wrote: Would this mess perhaps have something to do with udev/hotplug? Dpkg says: rc udev 0.056-3 ii hotplug 0.0.20040329-25 There was a problem around 2.6.12 days as far as I remember where a bad initrd tool generated initrd images which loaded ide-scsi, and once that is done the initrd tool will make future ones to match your setup. I think 2.6.15 kernel will force it to not use ide-scsi at all, but I am not sure. All true. I was a faithful user of 2.6.12 debian kernel up until yesterday. If you unloaded the ide-scsi, it would at least oops, sometimes harmlessly, sometimes fatally. It sounds like there is some extra step to using debian kernels now? It would be nice to get an FYI at least when installing one of these. I tried the 2.6.15.x debian kernel yesterday, and it couldn't mount root, I think because it wasn't loading the nv_sata module or the md/raid1 or something. My feeling is along the lines of Carl's in that I think it is a udev issue. I had a similar problem to you, Andy, with the 2.6.15 kernel and a RAID1 setup - it could not find /dev/md3 to mount it (/dev/md3 was the root partition). It looked like udev had not yet created the device. Unfortunately I didn't get to experiment much as the box in question, whilst not mission-critical, could not be off-line for very long or people would start whining #;-D I just rolled it back to 2.6.12 to get it back up and running. Regards, Ozz. pgpMow6z7O1Oa.pgp Description: PGP signature
Re: Unable to eject cdrom???
On Tue, 07 Feb 2006 14:15:36 +0100 Joost Kraaijeveld [EMAIL PROTECTED] wrote: Intermittently I am unable to eject my dvd/cdrom. Is there a way to determine whether Linux is refusing to eject the dvd/cdrom or if it is a hardware related problem? I would check to see if there are any files open on the drive. In particular, Konqueror was bad for leaving files open, preventing you from un-mounting or ejecting CDs/DVDs. If your CD is mounted on /media/cdrom then try lsof | grep /media/cdrom Regards, Ozz. pgpBuwam54cPM.pgp Description: PGP signature
Quanta 3.5.x crashing
Hi Guys. Ive recently started experiencing crashes with Quanta. They started when I upgraded to 3.5, but at the time were very infrequent. Quanta had been rock-solid up to that point. Today I've already had 2 crashes in 30 mins. I apt-get upgraded to 3.5.1 this morning, and the crashes seem to be identical. I took a look at Bugzilla, and 120983 seems similar to my issue, but they're telling me: quote Your crash seems to be completely independent of Quanta. Might be a: - libc bug - gcc misscompilation issue - hardware error /quote Has anyone else experienced this? Regards, Ozz. pgpYrh90IJDk5.pgp Description: PGP signature
Re: Stable vs Etch
On Fri, 3 Feb 2006 19:08:09 +0100 Gian Domeni Calgeer [EMAIL PROTECTED] wrote: Stable is Sarge and the most stable one; Testing is Sid and a bit less stable; Unstable is Etch and even less stable. You are correct in that Stable is currently Sarge. However, Testing is currently Etch, not Sid. Unstable is Sid, not Etch. Unstable will always be Sid. For those who don't know, the names are from characters in the movie Toy Story, and Sid was the boy who broke toys #;-D Regards, Ozz. pgpAsm8mHs1qN.pgp Description: PGP signature
Debian Security Advisory DSA 947-1 (ClamAV)
Hi Guys. I was perusing the latest security advisories and noticed this one that refers to ClamAV. To quote the advisory: quote A heap overflow has been discovered in ClamAV, a virus scanner, which could allow an attacker to execute arbitrary code by sending a carefully crafted UPX-encoded executable to a system runnig ClamAV. In addition, other potential overflows have been corrected. The old stable distribution (woody) does not include ClamAV. For the stable distribution (sarge) this problem has been fixed in version 0.84-2.sarge.7. For the unstable distribution (sid) this problem has been fixed in version 0.86.2-1. We recommend that you upgrade your clamav package immediately. /quote I checked the version I'm running on my amd64 boxen and it's version 0.88-2. Anyone know it this version is vulnerable, and if so, when a fix will be available? Regards, Ozz. pgpae9uHeU6JE.pgp Description: PGP signature
Re: Debian Security Advisory DSA 947-1 (ClamAV)
On Fri, 20 Jan 2006 13:33:37 + Stephen Gran [EMAIL PROTECTED] wrote: This one time, at band camp, Austin Denyer said: I checked the version I'm running on my amd64 boxen and it's version 0.88-2. Anyone know it this version is vulnerable, and if so, when a fix will be available? 0.88 is not vulnerable. Thanks Steven - I appreciate it. I was running it on (among other things) our proxy server, so I was more than a little concerned #;-D Thanks again. Regards, Ozz. pgphXZoMRHnjg.pgp Description: PGP signature
Re: apt get for AMD64
On Thu, 12 Jan 2006 09:34:58 -0700 Sean Roe [EMAIL PROTECTED] wrote: I emailed the group about this last week, but I didn't get a response (but to be fair I probably mis worded it), I am having problems with getting updates via apt-get. I am using the the repository: deb http://mirrior.espri.arizona.edu /debian_amd64/debian sarge main contrib But I keep getting errors when it downloads . Like the OS cant stat the file when I do it in aptitude. I am using a tyan K8 thunder board with 2 opteron cpus, 4 G of Ram, 2 320G sata drives, software raided. Any help would be great. What were the error messages? Do another apt-get update and copy/paste the output - it's kinda hard to know what the problem is without seeing the error message #;-D Regards, Ozz. pgp35Bye6tFVB.pgp Description: PGP signature
Re: apt get for AMD64
On Thu, 12 Jan 2006 10:18:55 -0700 Sean Roe [EMAIL PROTECTED] wrote: OK, thanks for the input. I just got ssh working so I can connect to this box and copy and pate the output: Get:1 http://security.debian.org stable/updates/main Packages [164kB] Get:2 http://mirror.espri.arizona.edu sarge/main Packages [3223kB] Hit http://security.debian.org stable/updates/main Release Hit http://security.debian.org stable/updates/contrib Packages Hit http://security.debian.org stable/updates/contrib Release 15% [1 Packages gzip 4096] [2 Packages 367415/3223kB 11%] gzip: stdin: invalid compressed data--crc error gzip: stdin: invalid compressed data--length error snippage I agree with Len - do you connect via a proxy? If so, does the proxy do virus checking? Sometimes the virus checking can get in the way. I've seen similar things happen with Squid/DansGuardian before. I don't use the Arizona mirror - I use debian.csail.mit.edu and a local internal mirror - but security.debian.org is the same for all of us... Regards, Ozz. pgpUTiZsKu0ba.pgp Description: PGP signature
Re: nvidia-glx removes x-server???
On Thu, 12 Jan 2006 12:25:57 -0500 Andrew Syrewicze [EMAIL PROTECTED] wrote: On 1/12/06, Lennart Sorensen [EMAIL PROTECTED] wrote: On Thu, Jan 12, 2006 at 12:18:45PM -0500, Andrew Syrewicze wrote: Well The packages are there, whether or not their up to date still remains to be seen. I'm gonna look first thing when i get out of work. The only nvidia packages on amd64.debian.net appear to be 7174. Those are too old, and hence not compatible with the latest x.org packages. I wonder if anyone has access to update non-free, since I don't think it is auto built. Len Sorensen Well, That's interesting. Guess i'll have to use the nv driver until a developer ports the latest version. Guess i could always go back to testing too. That still works as far as I know. Alright, thanks for the help guys... I get the same thing here. For me it's no big deal as I don't need 3d or 3000+FPS in GLXgears, so I'm fine with the vesa driver... Here's the output for me: dev04:/upgrade/Thu# apt-get install -s nvidia-glx Reading package lists... Done Building dependency tree... Done The following packages will be REMOVED xserver-xorg The following NEW packages will be installed nvidia-glx 0 upgraded, 1 newly installed, 1 to remove and 6 not upgraded. Remv xserver-xorg [6.9.0.dfsg.1-3] Inst nvidia-glx (1.0.7174-3 Debian AMD64 archive:unstable) Conf nvidia-glx (1.0.7174-3 Debian AMD64 archive:unstable) dev04:/upgrade/Thu# apt-cache show nvidia-glx | more Package: nvidia-glx Priority: optional Section: non-free/x11 Installed-Size: 10072 Maintainer: Randall Donald [EMAIL PROTECTED] Architecture: amd64 Source: nvidia-graphics-drivers Version: 1.0.7174-3 Replaces: nvidia-glx-src Provides: xserver Depends: nvidia-kernel-1.0.7174, xserver-common (= 4.0.3), xlibmesa-glu | libglu | libglu1, libc6 (= 2.3.2.ds1-4) Recommends: nvidia-kernel-source (= 1.0.7174) Suggests: nvidia-settings Conflicts: nvidia-glx-src Filename: pool/non-free/n/nvidia-graphics-drivers/nvidia-glx_1.0.7174-3_amd64.deb Size: 3571596 MD5sum: 954b404f7ad0af3b96094887da3a8d2e Description: NVIDIA binary XFree86 4.x driver These XFree86 4.0 binary drivers provide optimized hardware acceleration of OpenGL applications via a direct-rendering X Server and support the TNT, TNT2, TNT Ultra, GeForce, nForce and Quadro chipsets. AGP, TV-out and flat panel displays are also supported. . Please see the nvidia-kernel-source package for building the kernel module required by this package. Tag: devel::library, hardware::video == Regards, Ozz. pgpJRd4W9OZt2.pgp Description: PGP signature
Re: Please don't laugh...
On Tue, 10 Jan 2006 11:53:40 + Graham Smith [EMAIL PROTECTED] wrote: Ok, you can maybe have a bit of a chuckle. I have just done an upgrade and, because I wasn't paying attention, I installed a load of KDE 3.5 packages which has left we with a less than perfect KDE experience. I have managed to get most things working again but due to dependency issues I can get the menu bar to work. The way I see it I have three options 1)Attempt a complete downgrade of all the KDE stuff. I would rather not do this. 2)Add alioth to my sources list and cross my fingers that the packages in there will fulfil the dependencies (they mainly have version numbers along the lines of 3.5.0-01 where are packages in the official repositories are 3.5.0-03). 3)Wait it out and hope that the other 3.5 packages enter unstable soon. I really don't want to do this as I need to get on and work. Do I have any other options? What would you recommend? (Be more careful when upgrading would be my recommendation to myself). One thing I always do is this: Run apt-get upgrade with the -s option to see what it's gonna do. Run dpkg-repack [list of packages to be replaced/removed] in a temp directory. Then run apt-get upgrade as normal. If the upgrade goes south for any reason, just cd to the temp directory and run dpkg -i * and you're back to the way you were. It doesn't guard against every screwup, but it's saved my tail many times #;-D Regards, Ozz. pgplzrfefTWiO.pgp Description: PGP signature
Re: Please don't laugh...
On Tue, 10 Jan 2006 12:48:53 + Graham Smith [EMAIL PROTECTED] wrote: Anyway, the problem seems to be that there are some packages that are fairly central to KDE (kcron for instance) that are still at 3.4 in unstable. I am pretty much resigned to just sitting tight now. I tried to use packages from experimental but they have version numbers slightly too low. Yep. It's a pity, because from what I've seen it is real nice. I feel kinda lonely right now as I'm the only one in my department still running 3.4 - but then again, I'm also the only one in my department using amd64 on my desktop (the rest are still in 32-bit land) #;-D Regards, Ozz. pgp443VOOz0bP.pgp Description: PGP signature
Re: Panic - Kernel 2.6.14.4
On Wed, 28 Dec 2005 14:50:02 -0600 Russ Cook [EMAIL PROTECTED] wrote: I'm running a stock 2.6.12 kernel (2.6.12-1-amd64-k8-smp) on an ASUS A8N32-SLI Deluxe motherboard. Processor is AMD64 4800 dual core, with 4 MB ram. I'm trying to compile and install kernel 2.6.14.4. To start, I copied the same .config file from the 2.6.12 kernel, went through the usual make-kpkg clean, menuconfig, kernel-image process, changing nothing in the .config file. After installing using dpkg -i and rebooting, I get a kernel panic because VFS can't mount the root partition (hda1). I'm using Grub. The 2.6.12 kernel boots fine, and I used the same .config file. My hard disk is a Maxtor ATA. Can anyone offer pointers to check? I don't understand why one should boot, and the other won't. I'll be happy to send .config file on request. As an aside to this, when doing a regular 'make' you can specify the number of simultaneous jobs with the -j option (great for smp systems). I have not found any similar option for make-kpkg. Does anyone know if such an option exists for make-kpkg? Regards, Ozz. pgpZcBztKfrHi.pgp Description: PGP signature
Re: Local mirror problems.
On Wed, 14 Dec 2005 20:57:17 -0500 Anthony DeRobertis [EMAIL PROTECTED] wrote: Austin Denyer wrote: Now, if I ssh to the mirror, debian-amd64/debian-amd64/pool/main/m/manpages/manpages_2.16-1_all.deb is a symlink to the file in the regular mirror (debian-amd64/debian/pool/main/m/manpages/manpages_2.16-1_all.deb) and I can follow the link manually. Confirm that your FTP/HTTP/whatever server is set to follow symlinks. Some don't by default as a security measure. As far as I'm aware, the symlink configuration issue you mention is usually caused by the ftp server running in a chroot environment. Ours is not, as this particular server is internal only. If it matters, it is running proftp. Regards, Ozz. pgprIiJiwkiJB.pgp Description: PGP signature
Re: Local mirror problems.
On Fri, 16 Dec 2005 14:20:42 +0100 Goswin von Brederlow [EMAIL PROTECTED] wrote: Austin Denyer [EMAIL PROTECTED] writes: Hi. As we are getting more and more amd64 boxen here, we've decided to run our own internal mirror (we currently have an internal mirror of i386). Since the mirror is internal and rsync puts quite some stress on the server I recommend using reprepro instead. The drawback is that you loose the archive signature (and get a new signature by reprepro) and reprepro doesn't mirror install images. The advantage is that you can mix debian.org, security.debian.org and amd64.debian.net all into one local archive. Thanks - I'll take a look at that. Regards, Ozz. pgpnkY6ONNZSv.pgp Description: PGP signature
Re: Problems installing on AMD Athlon 64 system
On Mon, 19 Dec 2005 11:44:03 -0500 [EMAIL PROTECTED] (Lennart Sorensen) wrote: I think that board is one of the ones that does much better with 2.6.12 than 2.6.8 kernel. Unfortunately the official sarge installer uses 2.6.8. The unofficial one uses 2.6.12 and can be found here: http://www.tinyplanet.ca/~lsorense/amd64/ Might even fix the USB keyboard problem too. Len's install image above really does cure a multitude of problems. I used it recently to install a dual-Opteron, and it was FAR better than the official installer. You did a great job there, Len - Thanks! Regards, Ozz. pgpaq0tut8ZHb.pgp Description: PGP signature
Local mirror problems.
Hi. As we are getting more and more amd64 boxen here, we've decided to run our own internal mirror (we currently have an internal mirror of i386). I set the mirror up according to the How-To at: http://amd64.debian.net/~joerg/mirror.html I used the secondary mirror setup, as we already have the /debian tree mirrored, and there was no sense in pulling another ~18Gb of duplicate files. That section of the How-To has a real-life scenario at the end that exactly matches my setup: quote Real-Life Scenario: You mirror :debian-amd64/debian-amd64 to /home/ftp/debian-amd64, so the files end up living in /home/ftp/debian-amd64/debian-amd64. Assume you have your regular Debian mirror in /home/ftp/debian, then you need to create a symlink called debian in /home/ftp/debian-amd64 which points to /home/ftp/debian to actually get the link farm to work. /quote Now, this works fine for the amd64-specific files, but anything it has to pull from the regular mirror (non-amd64-specific) produce errors. Here is an example from the update I did today: === Get:1 ftp://mirror unstable/main libkpathsea4 3.0-12 [78.3kB] Get:2 ftp://mirror unstable/main tetex-bin 3.0-12 [4014kB] Get:3 ftp://mirror unstable/main manpages 2.16-1 [407kB] Err ftp://mirror unstable/main manpages 2.16-1 Unable to fetch file, server said '/debian-amd64/debian-amd64/pool/main/m/manpages/manpages_2.16-1_all.deb: No such file or directory ' Get:4 ftp://mirror unstable/main manpages-dev 2.16-1 [1116kB] Get:5 ftp://mirror unstable/main libcupsys2-dev 1.1.23-13 [91.3kB] Get:6 ftp://mirror unstable/main libcupsys2 1.1.23-13 [80.0kB] Get:7 ftp://mirror unstable/main libcupsimage2 1.1.23-13 [57.7kB] Get:8 ftp://mirror unstable/main cupsys 1.1.23-13 [8975kB] Get:9 http://debian.csail.mit.edu unstable/main manpages 2.16-1 [407kB] Get:10 ftp://mirror unstable/main menu 2.1.27 [387kB] Get:11 ftp://mirror unstable/main menu-xdg 0.2.2 [4210B] Get:12 http://debian.csail.mit.edu unstable/main manpages-dev 2.16-1 [1116kB] Get:13 http://debian.csail.mit.edu unstable/main menu-xdg 0.2.2 [4210B] Fetched 15.2MB in 5s (2706kB/s) == Now, if I ssh to the mirror, debian-amd64/debian-amd64/pool/main/m/manpages/manpages_2.16-1_all.deb is a symlink to the file in the regular mirror (debian-amd64/debian/pool/main/m/manpages/manpages_2.16-1_all.deb) and I can follow the link manually. Any clues? Regards, Ozz. pgpeHhlz3Gh0t.pgp Description: PGP signature
Re: Full mirror size ?
On Wed, 14 Dec 2005 16:21:20 +0100 Bool [EMAIL PROTECTED] wrote: Hello, I would like to create a full mirror for i386 and amd64, each one in stable and testing version. Do I need 18*2 + 25*2 Go of disk, or 18*2 + 6*2 Go ? or other sizes ? Thanks, Olivier http://amd64.debian.net/~joerg/mirror.html If you opt for the secondary mirror type (near the end of the page) then you can do 18 + 6. Otherwise, you're looking at 18 + 25. Regards, Ozz. pgprb8YLxhQ8m.pgp Description: PGP signature
Re: DualCore Dual-Opteron Board - suggestion?
On Thu, 08 Dec 2005 13:12:49 -0500 Adam C Powell IV [EMAIL PROTECTED] wrote: On Wed, 2005-12-07 at 16:02 -0500, Austin Denyer wrote: I believe there is an issue with kernels 2.6.14-2 with Opteron SMP. That is why I'm running 2.6.14-2 myself. The issue was an AMD errata rather than a bug with the kernel - 2.6.14-2 contained a workaround. I have a dual/dual machine (265/S940, Tyan S2891) which has trouble net-booting, while other dual/single core machines do fine: on the dual/dual, dhclient sends a request, the server replies, dhclient doesn't seem to get the reply. Could this kernel issue be causing my problem? I don't know, but it's possible. What kernel are you running at the moment? You may want to try it on 2.6.14-2. apt-get install linux-image-2.6.14-2-amd64-k8-smp Regards, Ozz. pgpJK1dLJGoon.pgp Description: PGP signature
Re: DualCore Dual-Opteron Board - suggestion?
On Wed, 07 Dec 2005 16:25:38 +0100 Lars Schimmer [EMAIL PROTECTED] wrote: Just for a quick note: anyone has a suggestion for a dual-Opteron board for Dualcore opterons which works flawless under debian amd64? E.G. the Tyan Tiger K8WE looks nice for me, but does it run flawless with debian amd64? At least this board has to power up 4 firewire bus for us here. I'm not running dual-core, but I do run dual-processor (Opteron 246) on the K8WE, and it runs great. You will probably want to use Len's 2.6.12 install cd though. Regards, Ozz. pgp5qjQqJKKKI.pgp Description: PGP signature
Re: DualCore Dual-Opteron Board - suggestion?
On Wed, 7 Dec 2005 13:04:43 -0800 mike [EMAIL PROTECTED] wrote: I believe there is an issue with kernels 2.6.14-2 with Opteron SMP. That is why I'm running 2.6.14-2 myself. The issue was an AMD errata rather than a bug with the kernel - 2.6.14-2 contained a workaround. I think that's all that NUMA stuff right? It wasnt optimized properly or something... It was the TLB flush filtering issue. Here's a link. It's long, so I've reproduced the clip below. http://www.kernel.org/pub/linux/kernel/v2.6/testing/ChangeLog-2.6.14-rc2 quote commit bc5e8fdfc622b03acf5ac974a1b8b26da6511c99 Author: Linus Torvalds [EMAIL PROTECTED] Date: Sat Sep 17 15:41:04 2005 -0700 x86-64/smp: fix random SIGSEGV issues They seem to have been due to AMD errata 63/122; the fix is to disable TLB flush filtering in SMP configurations. Confirmed to fix the problem by Andrew Walrond [EMAIL PROTECTED] [ Let's see if we'll have a better fix eventually, this is the QD let's get this fixed and out there version ] Signed-off-by: Linus Torvalds [EMAIL PROTECTED] /quote And here's another link: http://lkml.org/lkml/2005/9/20/207 quote Date Tue, 20 Sep 2005 10:30:48 -0700 (PDT) From Linus Torvalds Subject Re: x86-64 bad pmds in 2.6.11.6 On Tue, 20 Sep 2005, Charles McCreary wrote: Another datapoint for this thread. The box spewing the bad pmds messages is a dual opteron 246 on a TYAN S2885 Thunder K8W motherboard. Kernel is 2.6.11.4-20a-smp. This is quite possibly the result of an Opteron errata (tlb flush filtering is broken on SMP) that we worked around as of 2.6.14-rc4. So either just try 2.6.14-rc2, or try the appended patch (it has since been confirmed by many more people). Linus /quote Regards, Ozz. pgpBdE6RAnhqH.pgp Description: PGP signature
Re: Suggestions for a new AMD64 system
On Tue, 29 Nov 2005 17:09:23 +0100 Goswin von Brederlow [EMAIL PROTECTED] wrote: After 22 days: total used free sharedbuffers cached Mem: 10255801014156 11424 0 13740 614136 -/+ buffers/cache: 386280 639300 Swap: 979832 204780 775052 Galeon eats a lot of ram and old tabs get swapped out after a while. Another MAJOR ram-killer is Evolution. It seems to increase memory usage exponentially with message size. I recently received a 10mb log file that took forever to load, and _almost_ crashed my box - and I have a dual-Opteron 246 with 2Gigs of ram. I run Sylpheed now, and that same e-mail hardly causes it to skip a beat. Evolution leaks like a sieve too... Regards, Ozz. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]