Re: Installing xorg without all those video drivers
On Mon, Aug 06, 2007 at 07:39:05PM +1000, Robert S wrote: I'd like to install xorg, but only with the video drivers that I need. I'm doing a net install. Here's how I do it: First, I use aptitude interactively. If you haven't used aptitude before, then that's a separate project to do first. Aptitude gives you a curses interface similar to but more powerful than dselect. Aptitude keeps track of which packages were installed by you and which were installed to meet dependencies. So far, similiar to apt-get. However, when you remove a package that you installed, aptitude will remove anything that then doesn't have anything depending on it that was marked as automatic. Then, I go to the xorg meta-package. I scroll down to the things it depends on and I select them manually, and I dig down into the dependancies and adjust things. For example, somewhere there a package will depend on _either_ an -all package or a device-specific package. Left on its own, it will pull in the -all whereas I can mark the device spcific package for manual and deselect the -all. It sounds more complicated than it is. I'm comfortable with aptitude so its not. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Debian hangs randomly
On Mon, Aug 06, 2007 at 12:27:44PM +0200, Jogito nit wrote: First of all, sorry about my english. No problem. I have a problem with my Debian, it hangs randomly. I have tried to change the driver of my ati, I, ve tried without 3D acceleration. I thougth that was the memory, so i changed it, but it continue hang. When it hangs i cant do anything, no mouse, no keyboard. I?ve tried Kubuntu and Knoppix too, and the same problem. I?ve looked the logs and i cant find anything strange. In windows i dont have this problem. Can someone help me please. I ussually use Kubuntu, but I?ve tested Debian to see if the problem happend, and of course it happends. So its not a problem of Debian I know, but i dont know what can i do, because i dont want to use windows. I have tested debian 4 etch, with kernel 2.6.18. About 4 years ago i used mandrake, and I had a problem like that because the micro heat to much and the kernel try to reduce the frequency, so it hang, but I?m not sure of this, I dont know if this can help. My computer is : P4 3,06 HT Ati radeon 9700 Sound blaster live. 1024mb memory. It sounds like we need to narrow down the source of the troubles. The first big dividing line is between a text-based system and an Xwindow based system. I'll assume that when you boot your machine normally, it gives you an Xwindow login. Reboot the computer but select the single-user mode grub menu option. If you happen to be using lilo, what you want is the linux s option (the s is passed to init). This tells init to boot into single-user mode. Debian will boot but instead of giving you X, you will be prompted for the root password and given a shell. Now disable the display manager (gdm, kdm, xdm) in /etc/rc2.d See the README in that directory for instructions for how to do this. This will mean that when you boot normally, you will get a command-line login. To start X from the command line, you type 'startx'. #shutdown -r now will cause the computer to boot nomally now (out of single-user mode) and you should be in text mode. Try doing some things and see if the computer hangs. While I wouldn't touch the package management incase of a hang in the middle, you can use whatever other text-based apps you have. Try an editor, lynx or links2 or mc if you have it. Try the following to exersize the system a bit (don't worry about 'permission denied', you're only trying to read files and you don't care about the output. cd /;du find / find / |xargs md5sum If it doesn't hang by the time it usually would, then we're on the right track. If it hangs, try to find out what it was doing. Also, the first boot after a hang, look at /var/log/syslog. If it doesn't hang, you can run apt-get memtest86+. This should add a line to your grub menu whereby you can run memtest without an OS loaded. You can also apt-get lynx and mc so you have some command-line apps to use while you figure out why the box hangs. Let us know how things go. Good luck, Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: DNS problem on local network
On Mon, Aug 06, 2007 at 12:39:56PM +0100, Adam Hardy wrote: I installed dnsmasq to run DNS and DHCP servers on my little home network of 4 PCs and a couple of laptops, and everything was going fine, internet browsing, ssh, ftp by IP address etc. Now I am trying to get DNS to work for local machines but it won't co-operate. I spent the last couple of hours pouring over HOWTOs and FAQs trying to work out the problem but I'm stuck with a few errors messages which I suspect point to something totally obvious to those in the know. Please send us your /etc/hosts file. Every box needs a minimal /etc/hosts file with at least its own hostname (though I'v never used DHCP). The box running dnsmasq should have all the hosts on your network listed in /etc/hosts for dnsmasq to read. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Transitioning to 64bit, is it worth it, and how
On Mon, Aug 06, 2007 at 12:55:46PM +0100, Alan Chandler wrote: On 8/4/07, Alan Chandler [EMAIL PROTECTED] wrote: I have a Core2 Duo on which I am running a 686 kernel (from Debian unstable). It has 1GB of memory I am wondering two things a) What are the pros and cons in switching to 64 bit mode? - Is it faster? I am only responding to this particular message in the thread because my computer is still not fully operational and this is the only way (at the moment) to comment on this process. [snip many woes] When you ran the installer, did you start from scratch? Did you start by removing your existing partitions? I don't think you can reuse a 32-bit software raid setup on 64-bit (don't know, never tried). The only thing you can probably reuse is your /home, however, if /home is on a raid or LVM partition, you may not be able to reuse it either. If this is the case, at the beginning of the installer, choose a disk to partition and tell it to ignore the partition table and start from scratch. If it doesn't let you do that either, restart the installer, and go to the VC with a shell and #dd if=/dev/zero of=/dev/[drive toclear] bs=512 count=1 to clear the partition table. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: [OT] Recommended Router Hardware
On Mon, Aug 06, 2007 at 04:07:00PM +0200, Samuel B?chler wrote: Hi Everyone I want to improve my home network. I have got a Laptop and an old desktop machine. I want to use the old desktop for server services such as web- and mail-sever. I am connected to the ISP using a cable modem. Question A: Can someone recommend a router with the following properties: -NAT -port forwarding Use your old desktop machine, just give it a second NIC to go to the cable modem. Then install shorewall and read the shorewall-doc. I should clarify this a bit: how old is the desktop machine? Debian now needs at least a 486 and the installer needs 48 MB ram. If the desktop is a 386 or has less ram, then you'll need either NetBSD or OpenBSD (24 MB ram). Question B: Is it possible to install FREESCO on top of a commercial router (hardware, e.g. linksys, netgear etc.) What's FREESCO? Question C: Any recommendations for Guides, HOWTOs about networking. debian-reference, linux-HOWTOs, shorewall-doc, harden-doc, man pages (all available as debian packages). Question D: Are there any mailing lists for conceptional discussions about Computers? If those computers run debian, this is it. What you're wanting to do is something most of us do or have done. Once you get this router/firewall set up, you can put things like dnsmasq on it. Enjoy, Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: [OT] Recommended Router Hardware
On Mon, Aug 06, 2007 at 07:30:22PM +0200, Samuel B??chler wrote: I want to improve my home network. I have got a Laptop and an old desktop machine. I want to use the old desktop for server services such as web- and mail-sever. I am connected to the ISP using a cable modem. Question A: Can someone recommend a router with the following properties: -NAT -port forwarding Use your old desktop machine, just give it a second NIC to go to the cable modem. Then install shorewall and read the shorewall-doc. I should clarify this a bit: how old is the desktop machine? Debian now needs at least a 486 and the installer needs 48 MB ram. If the desktop is a 386 or has less ram, then you'll need either NetBSD or OpenBSD (24 MB ram). Well - it isn't that old. It is a Compaq Desktop EN, Pentium III with about 500MB RAM. But there is only one Ethernet-Connection which - I guess - can be solved by using a Hub or Switch, isn't it? Not if you want it to be a router. Buy a NIC; cheap ones based on Realtek cost about $10. You have oodles of capacity for what you want. I now got a Linksys WRT54GL to which Eric Reymond refers to in Linksys Blue Box Router HOWTO[2]. Never had one. Question B: Is it possible to install FREESCO on top of a commercial router (hardware, e.g. linksys, netgear etc.) What's FREESCO? FREESCO is a NAT/firewall router/server based on Linux and runs on a single 1.44MB floppy[1]. [1] http://sourceforge.net/projects/freesco [2] http://www.tldp.org/HOWTO/Linksys-Blue-Box-Router-HOWTO/index.html Your firewall will need to be totally up-to-date. If its a floppy image, unless its up-to-date, I wouldn't trust it. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Opinions XFS
On Mon, Aug 06, 2007 at 09:55:28AM -0700, David Brodbeck wrote: On Aug 4, 2007, at 2:42 AM, Ron Johnson wrote: I'd have to modify that. Instead of NIH, my worry is that since XFS was designed for a different kernel, it's been shimmed into Linux and so doesn't integrate as well as ext2/3 and ReiserFS. Same concern with jfs. I suppose that's a valid concern, but in the absence of any evidence of problems caused by it I can't say I'm going to lose any sleep. :) Given that SGI boxes now use Linux and have dropped Irix but still use XFS, I think it pretty likely that they have done a good job of ensuring that Linux's XFS is up to snuff. IBM started JFS version 1 with AIX, then ported it to OS/2 and added features to make it version 2, then ported it back to AIX where it is the standard FS. They got Linux working on their newer Power servers to meet customer demand and ported JFS to linux so that they had a common filesystem irrespective of OS. In both cases, the porting was done or directed by the origionator of the filesystem for reasons that impact their bottom line. To some extent their reputations are on the line with their filesystems. As they are right now, I would trust them both equally well. They each have their stronger points that make one more suitable than the other for certain uses of the filesystem. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Hard disk dying?
On Mon, Aug 06, 2007 at 10:59:13AM -0700, Sam wrote: Justin, Should I unmount the drive before doing the tests? No need. The tests happen at the drive level, below the level of the filesystem. When you run the test, it should tell you how long to wait before you query the drive to get the result. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: DNS problem on local network
On Mon, Aug 06, 2007 at 06:50:21PM +0100, Adam Hardy wrote: Douglas Allan Tutty on 06/08/07 14:27, wrote: On Mon, Aug 06, 2007 at 12:39:56PM +0100, Adam Hardy wrote: I installed dnsmasq to run DNS and DHCP servers on my little home network of 4 PCs and a couple of laptops, and everything was going fine, internet browsing, ssh, ftp by IP address etc. Now I am trying to get DNS to work for local machines but it won't co-operate. I spent the last couple of hours pouring over HOWTOs and FAQs trying to work out the problem but I'm stuck with a few errors messages which I suspect point to something totally obvious to those in the know. Please send us your /etc/hosts file. Every box needs a minimal /etc/hosts file with at least its own hostname (though I'v never used DHCP). The box running dnsmasq should have all the hosts on your network listed in /etc/hosts for dnsmasq to read. [EMAIL PROTECTED]:~$ cat /etc/hosts 127.0.0.1 localhost localhost.localdomain # The following lines are desirable for IPv6 capable hosts ::1 ip6-localhost ip6-loopback fe00::0 ip6-localnet ff00::0 ip6-mcastprefix ff02::1 ip6-allnodes ff02::2 ip6-allrouters ff02::3 ip6-allhosts So I should put gondor in there on the same line as the localhost? And on the dnsmasq box it's the same. I can enter the IP address of the box's NIC on the internal network into /etc/hosts because it's fixed, but all the DHCP clients? You should only have to put your hosts in the /etc/hosts file on the dnsmasq box. No, you should not touch the 127.0.0.1 line. They should go on their own line, one per IP. I don't know about the DHCP clients. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: cups yet again
On Mon, Aug 06, 2007 at 05:02:07PM +0100, graham wrote: Yet another cups problem (the one program which makes me feel like I do when running windows - like putting a foot through the computer). So why run cups? Use LPRng and Apsfilter or foomatic print filters. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: cups yet again
On Mon, Aug 06, 2007 at 07:14:25PM +0100, graham wrote: Douglas Allan Tutty wrote: On Mon, Aug 06, 2007 at 05:02:07PM +0100, graham wrote: Yet another cups problem (the one program which makes me feel like I do when running windows - like putting a foot through the computer). So why run cups? Use LPRng and Apsfilter or foomatic print filters. Cos what I'd understood from other threads was that this would mean swimming against the tide, since cups is now the default for both debian and gnome, and because I had understood that lprng was no longer supported. I'm really hoping to spend the minimum of time possible maintaining printers; they don't interest me much ;-) LPRng does look a little long-in-the-tooth. The web page is dated '5 Oct 2004' for the same version as debian ships. On the other hand, good'ol lpr is up-to-date (still the default, constantly maintaind, on Net- and OpenBSD), sourced from OpenBSD. The bug reports are rather silly, such as Installing LPR over LPRng doesn't work (of course not, command file names), lpr fails to modprobe the paralell port (of course not, that's your job), etc. This is, in fact, what I use. Lpr with apsfilter. Simple to setup, well documented. It works. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: [OT] Recommended Router Hardware
On Mon, Aug 06, 2007 at 01:49:08PM -0700, David Brodbeck wrote: On Aug 6, 2007, at 1:22 PM, Andrei Popescu wrote: On Mon, Aug 06, 2007 at 08:45:21AM -0700, Andrew Sackville-West wrote: no. to use a desktop machine as a router, you need two network connections: one for the local network to attach to and one for the internet at large. The computer then routes packets appropriately and I've read somewhere that it can be done with one card, but it's very difficult to setup. You can do it, by creating multiple IP aliases on the one adapter. There are some caveats, though. For example, this likely won't work very well if you plan to do DHCP on your local network. Other protocols that use IP broadcasts can also get confused or get into trouble. Other than that, it isn't necessarily *harder* to set up, but it's more *confusing* to set up, if that makes sense. If this box is going to be a firewall, even with IP aliases, isn't this an issue? Is there no way for an attack to bypass the level 3 router by doing something at a lower level? Anyway, its too confusing to me, especially if its only to save buying a cheap NIC. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Opinions XFS
On Mon, Aug 06, 2007 at 04:29:43PM -0400, Justin Piszcz wrote: On Mon, 6 Aug 2007, Douglas Allan Tutty wrote: On Mon, Aug 06, 2007 at 09:55:28AM -0700, David Brodbeck wrote: On Aug 4, 2007, at 2:42 AM, Ron Johnson wrote: I'd have to modify that. Instead of NIH, my worry is that since XFS was designed for a different kernel, it's been shimmed into Linux and so doesn't integrate as well as ext2/3 and ReiserFS. Same concern with jfs. I suppose that's a valid concern, but in the absence of any evidence of problems caused by it I can't say I'm going to lose any sleep. :) Given that SGI boxes now use Linux and have dropped Irix but still use XFS, I think it pretty likely that they have done a good job of ensuring that Linux's XFS is up to snuff. IBM started JFS version 1 with AIX, then ported it to OS/2 and added features to make it version 2, then ported it back to AIX where it is the standard FS. They got Linux working on their newer Power servers to meet customer demand and ported JFS to linux so that they had a common filesystem irrespective of OS. In both cases, the porting was done or directed by the origionator of the filesystem for reasons that impact their bottom line. To some extent their reputations are on the line with their filesystems. As they are right now, I would trust them both equally well. They each have their stronger points that make one more suitable than the other for certain uses of the filesystem. I would too, until I found out JFS has no maintainer. Yikes. The jfsutils copyright and README.Debian are internally dated in 2001 as if they are old packages. However, the changelog.Debian.gz and changelog.gz are June, 2006. xfsprogs have more recent changes. Stefan Hornburg is listed as responsible for this Debian package. What exactly do you mean that JFS has no maintainer. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Opinions XFS
On Mon, Aug 06, 2007 at 05:32:11PM -0400, Justin Piszcz wrote: On Mon, 6 Aug 2007, Douglas Allan Tutty wrote: What exactly do you mean that JFS has no maintainer. It has a maintainer, but he cannot work on it full-time: Date: Mon, 30 Jul 2007 15:11:06 -0500 From: Dave Kleikamp [EMAIL PROTECTED] Anyway, for enterprise use, I couldn't recommend jfs, since there is no full-time maintainer. Thanks, Shaggy -- David Kleikamp IBM Linux Technology Center Wow. IBM telling you that they don't recommend jfs for enterprise use. Now why is it still in debian stable without a warning? Perhaps I've been naive, but I expect that packages in debian main stable to be solid and reliable, especially for system-level packages (like filesystems), and less-so for inherently unstable things like web browsers (since they have to keep pace with web features, the feature set isn't stable enough for reliability). If the maintainer can't recommend it for enterprise use, that should be a release-critical bug in and of itself that prevents it from getting into stable. Thanks for the heads-up. I'll probably switch to XFS for my next build. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
JFS suitability [Was Re: Opinions XFS]
Hello Stefan, We've been having a discussion on debian-user on the differences between XFS and JFS and where one would be better than the other for different applications. In the course of the discussion, Justin Piszcz sent a copy of an email he received from Dave Kleikamp at IBM who doesn't recommend JFS for enterprise use because it no longer has a full-time maintainer. I wondered how an important package like a filesystem would end up in debian main stable when upstream doesn't recommend it. I thought that this would by itself have created a release-critical bug that would keep it out of stable. Would you care to comment or shed light on this issue? Please reply to the debian-user mailing list if possible. Thanks, Doug Tutty. On Mon, Aug 06, 2007 at 05:32:11PM -0400, Justin Piszcz wrote: On Mon, 6 Aug 2007, Douglas Allan Tutty wrote: On Mon, Aug 06, 2007 at 04:29:43PM -0400, Justin Piszcz wrote: On Mon, 6 Aug 2007, Douglas Allan Tutty wrote: On Mon, Aug 06, 2007 at 09:55:28AM -0700, David Brodbeck wrote: On Aug 4, 2007, at 2:42 AM, Ron Johnson wrote: I'd have to modify that. Instead of NIH, my worry is that since XFS was designed for a different kernel, it's been shimmed into Linux and so doesn't integrate as well as ext2/3 and ReiserFS. Same concern with jfs. I suppose that's a valid concern, but in the absence of any evidence of problems caused by it I can't say I'm going to lose any sleep. :) Given that SGI boxes now use Linux and have dropped Irix but still use XFS, I think it pretty likely that they have done a good job of ensuring that Linux's XFS is up to snuff. IBM started JFS version 1 with AIX, then ported it to OS/2 and added features to make it version 2, then ported it back to AIX where it is the standard FS. They got Linux working on their newer Power servers to meet customer demand and ported JFS to linux so that they had a common filesystem irrespective of OS. In both cases, the porting was done or directed by the origionator of the filesystem for reasons that impact their bottom line. To some extent their reputations are on the line with their filesystems. As they are right now, I would trust them both equally well. They each have their stronger points that make one more suitable than the other for certain uses of the filesystem. I would too, until I found out JFS has no maintainer. Yikes. The jfsutils copyright and README.Debian are internally dated in 2001 as if they are old packages. However, the changelog.Debian.gz and changelog.gz are June, 2006. xfsprogs have more recent changes. Stefan Hornburg is listed as responsible for this Debian package. What exactly do you mean that JFS has no maintainer. It has a maintainer, but he cannot work on it full-time: -- Forwarded message -- Date: Mon, 30 Jul 2007 15:11:06 -0500 From: Dave Kleikamp [EMAIL PROTECTED] To: Justin Piszcz [EMAIL PROTECTED] Cc: [EMAIL PROTECTED], [EMAIL PROTECTED], [EMAIL PROTECTED], [EMAIL PROTECTED] Subject: Re: bonnie++ benchmarks for ext2,ext3,ext4,jfs,reiserfs,xfs, zfs on software raid 5 On Mon, 2007-07-30 at 10:29 -0400, Justin Piszcz wrote: Overall JFS seems the fastest but reviewing the mailing list for JFS it seems like there a lot of problems, especially when people who use JFS 1 year, their speed goes to 5 MiB/s over time and the defragfs tool has been removed(?) from the source/Makefile and on Google it says not to use it due to corruption. The defragfs tool was an unported holdover from OS/2, which is why it was removed. There never was a working Linux version. I have some ideas to improve jfs allocation to avoid fragmentation problems, but jfs isn't my full-time job anymore, so I can't promise anything. I'm not sure about the corruption claims. I'd like to hear some specifics on that. Anyway, for enterprise use, I couldn't recommend jfs, since there is no full-time maintainer. Thanks, Shaggy -- David Kleikamp IBM Linux Technology Center -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED] -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: replacement for apt-listchanges?
On Tue, Aug 07, 2007 at 01:42:25AM +0200, Vincent Lefevre wrote: On 2007-08-06 15:13:10 -0500, Ron Johnson wrote: because breakage due to python occurs too frequently), it is important I still don't have any evidence that python breaks frequently. This is based on personal experience. Unfortunately I haven't kept statistics, but I'd say, taking into account the serious bugs that directly affected the use of my machines and those I could avoid (e.g. thanks to apt-listbugs), this should be about 4 per year. This may seem low, but this is much more than what I get with the other base packages (libc6, perl, xorg...). Can you describe the breakage? I've found that packages that rely on python for a GUI break (GUI freezes), I've always blamed it on the GUI. Base python itself I've never had break. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: less, exit but left content on screen
On 8/4/2007 6:06 PM, Tong Sun wrote: Package: less Version: 394-4 Severity: wishlist I remember that I used to be able to exit 'less' by command key 'x' or something so that the content just viewed is left on screen, instead of being cleared and restored to the screen before invoking 'less'. But I found there is no such capability in current less (version 394). Or is there? I'm running standard Etch, with less version 394-4. I've always exited less with 'q', which leaves the screen intact. If -X works, I wonder what is different between our terminals. My $TERM = linux. In any event, less(1) doesn't show a command key 'x' that I can see. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Get display back?
Keep the reply on the list please. On Sun, Aug 05, 2007 at 02:44:45PM +1200, Jeff wrote: Douglas Allan Tutty wrote: Another way to get into the box is to boot the install CD in rescue mode. From the menus, you can get a shell that is chrooted into the box. From there, you can either edit xorg.conf or you may even be able to run dpkg-reconfigure xserver-xorg. Also, there's an option inbetween a normal boot and an init=/bin/sh, that is 's', for single. This will run through the /etc/rcS.d scripts and prompt for the root password. Thanks for that! I booted the single user mode from the Grub Menu and got the command prompt from there. Ran dpkg-reconfigure xserver-xorg as root then changed the driver back to nv and the display is back. So, on to why the nvidia driver isn't working for you. Are you running Etch with stock kernel and have the matching nvidia kernel package, nvidia-xconfig, and nvidia-glx? How did you try to set it up before? Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Transitioning to 64bit, is it worth it, and how
On Sun, Aug 05, 2007 at 12:28:14AM -0600, Bob Proulx wrote: But even so there was huge resistance because it was not a perfect solution and did not support 32-bit applications out of the box. I run Etch amd64 on may Athlon64 3800+ and its great. Then again, this box was an upgrade from my 486 :) The only thing I need ia32 for is iceweasel/flash. For this, I had to setup a chroot which wasn't that bad and takes up 442 MB of disk space. The only thing that would have been convenient would have been an amd64 package that brought in debootstrap and schroot and whatever else and did all the setup ready for installing the chroot. There was a lot of reading of howtos and man pages to bring myself up-to-speed on setting it up. Now I see in Sid there's a wrapper for mozilla plugins to make using a chroot for this unneccessary, however there could still be value in a 'ia32-schroot' meta-package. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: less, exit but left content on screen
On Sun, Aug 05, 2007 at 08:54:52AM -0400, Douglas Allan Tutty wrote: On 8/4/2007 6:06 PM, Tong Sun wrote: Package: less Version: 394-4 Severity: wishlist I remember that I used to be able to exit 'less' by command key 'x' or something so that the content just viewed is left on screen, instead of being cleared and restored to the screen before invoking 'less'. But I found there is no such capability in current less (version 394). Or is there? I'm running standard Etch, with less version 394-4. I've always exited less with 'q', which leaves the screen intact. If -X works, I wonder what is different between our terminals. My $TERM = linux. In any event, less(1) doesn't show a command key 'x' that I can see. I just tried man in xfce-terminal and find that q does clear the screen. How unhelpful; another good reason to use the text console :) Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Weird cron behavior
On Sun, Aug 05, 2007 at 09:26:28AM -0400, Frank McCormick wrote: I left my machine on overnight ( I usually don't ) and got these messages in my mail this morning: You mean that you don't run anacron when you don't leave the compuer on overnight? There are _lots_ of default scripts in /etc/cron.daily that need to run. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Get display back?
On Mon, Aug 06, 2007 at 09:03:42AM +1200, Jeff wrote: So, on to why the nvidia driver isn't working for you. Are you running Etch with stock kernel and have the matching nvidia kernel package, nvidia-xconfig, and nvidia-glx? How did you try to set it up before? I am using kernel 26.18.-4-K7, card is nvidia FX 5200 . I used a step by step guide from a website (which i cant find now) I would need to use an easy better guide. Any suggestions? I'm assuming Etch then. You don't need a website, you don't need module-assistant. If you downloaded the driver from nvidia and ran their installer, then you have nvidia polution in your filesystem that you'll have to get out somehow (never did it, don't know how, but remember threads here bemoaning it). Then you should have the following packages installed: i linux-image-2.6-k7 always depends on latest kernel for the k7 iA linux-image-2.6.18-4-K7 the latest kernel, which you have installed i nvidia-glx Read its docs to ensure that your card is supported. If not, try the '-legacy' series of glx and kernel modules. i nvidia-kernel-2.6-k7 always depends on latest nvidia module iA nvidia-kernel-2.6.18-4-K7latest nvidia module for the K7 iA nvidia-kernel-common i nvidia-xconfig Now get X working with the free nv driver. cp xorg.conf xorg.conf.nv.works Then update the man-db (as root, run /etc/cron.daily/man-db), apropos nvidia and read all the man pages and all the documentation in /usr/share/doc/nvidia* Get the system so that it boots to command line rather than X. This generally means temporarily disabling the startup of your display manager (gdm or whatever, never used one). You should just have to run nvidia-xconfig. This will alter your xorg.conf file for use of the driver. Note that it does more than change 'nv' to 'nvidia'. cp xorg.conf xorg.conf.nvidia.test1 Test X with startx. You may need to fitz with parameters to nvidia-xconfig. Once its working, if you use a display manager, re-enable it. cp xorg.conf xorg.conf.nvidia.works Good luck, Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Public PC
On Sun, Aug 05, 2007 at 03:06:10PM -0500, Dave Walker wrote: I hope to place a PC running Etch in our small (tiny, actually) airport building for use by pilots. The PC will be used to access perhaps 10 web sites (plus or minus) over a DSL connection and to run a few utilities to show GMT and a calculator and maybe a few additional applications. The box will also serve as a real-time data collection platform and web server for a local weather observing system, so I want to prevent it from being re-booted. Be very careful. These are two very different applications to put on one box. You don't want the weather reports to stop going out if the desktop freezes the box. A box that runs X, is almost by definition more prone to needing to reboot than one that doesn't. Ditto if you use a Desktop environment instead of a simple window manager. If you really need it to be one box, here's what I would suggest: Set up the server stuff first. Setup a chroot for the desktop stuff, one that gets copied when used. Use the pam module so that when the desktop user logs in, they get a fresh clean chroot which gets distroyed when the user logs out. Use icewm with the toolbar configured with the common tools. Also note that Xfce seems to use more and more memory; I don't know if it has been fixed. Determine if you need a full-fledged web browser for those 10 sites or if something like links2 or dillo will work. Hint: dillo is plain with a gtk interface, links2 adds javascript but doesn't have a gtk interface, whereas for flash you need a full browser like iceweasel or Konqueror. Provide an alternate means for an administrator to log into the machine; ssh or serial console (or both). You may want to setup syslog to send logs to another machine and put something on that machine to monitor the logs. Good luck, Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Syntax error in bitops.h:244
On Sun, Aug 05, 2007 at 09:42:29PM +0100, Adam Gray wrote: By the magic of aptitude, I installed a package that wanted udev, and as a result got rid of hotplug by accident. So I removed this package (can't remember what it was, nothing terribly important), reinstalled hotplug and thought nothing of it. If you're running Etch (stable), the release notes say that udev is required now. udev conflicts with hotplug since it provides the same service. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Opinions XFS
On Sat, Aug 04, 2007 at 09:43:46AM +0200, J?rg-Volker Peetz wrote: Debian Administration :: Filesystems (ext3, reiser, xfs, jfs) comparison on Debian Etch http://www.debian-administration.org/articles/388 Great article. As Ron points out, a UPS answers many of the power-failure questions, however, should a machine freeze up and require a hard reset, a UPS doesn't help. I suppose to answer this, the tests should have added a calculate md5 hash on all files with a power-cycle at one point during a vulnerable point in a second round of the test. We all know that power-cycling a mounted FS can result in loss of data that was being written at the time. It suggests that at least system binaries should remain intact, unless you're doing an upgade at the time :( Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: converting file system
On Sat, Aug 04, 2007 at 07:42:57AM +0300, Andrei Popescu wrote: On Fri, Aug 03, 2007 at 08:23:17PM -0400, Douglas Allan Tutty wrote: This is a great reason to be using LVM with separate LVs. Need to convert? Create a new LV, make the new filesystem, tar the data over to it. When everything is working, delete the old LV. If you need extra space during the transition, borrow a drive and add it to the VG. What's the benefit of LVM when I add a separate drive? It depends on the relative size of the filesystem you want to convert to the size of your spare drive. If your spare drive is larger, it doesn't matter. If the filesystem is larger, the drive can be added to the VG to donate its space. This could be a real issue if your spare drive is an older, smaller one. E.g: filesystem is 60 GB, you only have 20 GB spare in the VG (call it an 80 GB drive), however, you can borrow an older 40 GB drive. Add the 40 GB drive to the VG, you now have 60 GB free to make a temp LV. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Transitioning to 64bit, is it worth it, and how
On Sat, Aug 04, 2007 at 12:19:09PM +0100, Alan Chandler wrote: On Saturday 04 Aug 2007, Ron Johnson wrote: On 08/04/07 03:25, Alan Chandler wrote: b) Is there a simple transition path? Reinstall from scratch, and pray that your /home is on a seperate partition. Well the only home that matters - ie mine is - this is essentially just a personal machine. I am using raid1 with LVM on top, so I can easily create additional partitions and move things about as I wish. My only slight reservation is I am not sure if the installer supports raid, so it might be more complex than just installing. I also want to take the opportunity to enlarge my /boot partition (also raid 1) as 32M is just not enough these days - frequently seem to fill it up as I upgrade kernels. I don't think that you can keep your 32-bit raid/LVM setup, but I don't know. Your safest option is to backup your /home (you do this anyway, right), and do a clean reinstall. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Issues with ethernet in testing/lenny
On Sat, Aug 04, 2007 at 09:35:50AM -0400, Wayne Topa wrote: percy tiglao([EMAIL PROTECTED]) is reported to have said: Hello, I've recently purchased a Dell Inspiron 530 and wanted to get Debian onto it. I've installed Lenny mostly without any issues, but it did not autodetect my ethernet card. It is a 82562V-2 integrated card, at least according to the Windows Vista devise manager. Anyway, I do have a very basic system up and running. No internet support or anything, so I can really only do stuff from the Windows side of my system right now. I'm not sure how I can get logs of what I do to my email from Debian so I'm sorry about the lack of the logs. I'll do my best to describe the error messages however. If you have a device, such as a USB stick that can be read and written by Linux and at least read by Windows, you can transfer the logs via this; sort of a sneaker-net that doesn't go anywhere. As I stated earlier, the installer did not detect the network card. I tried loading the e1000 module inside the installation program which according to intel should be the module I need: http://downloadmirror.intel.com/9180/ENG/README.txt The installer wouldn't acccept it, it would just loop back to the select module screen. I tried to modprobe e1000 manually during the installation, but that didn't work either. No error messages come up. After you do this, look in /var/log/syslog and see if any eth lines appear. It may not be numbered eth0. After that, the rest of the system installed without any hassle, but obviously I cannot connect to the internet right now without any network card, so I don't have anything installed aside from whatever is on the base netinstall on Lenny. You could use the apt-zip package only use a CD or USB stick instead of a Zip (unless, of course, you have a zip disk drive). You can download individual packages with your windows box, put them on the stick, then install them on the debian box with dpkg. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Opinions XFS
On Sat, Aug 04, 2007 at 02:14:28PM -0700, Andrew Sackville-West wrote: magic sysrq key... Alt-SysRq-S to sync the filesystems followed by Alt-SysRq-U tu remount readonly. then reboot. http://www.debian-administration.org/articles/457 doesn't *always* work, but I've gotten to work a couple times on otherwise totally unresponsive machines... If I remember right, the magic key is disabled in stock debian kernels. Do you compile your own? Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Get display back?
On Sat, Aug 04, 2007 at 08:42:15PM -0500, Kent West wrote: Kent West wrote: ... that's fairly indicative that the either the video system or the system as a whole is locked up. If your CAPS lock or Numlock indicator toggles when you press on the corresponding key, that's indicative that the entire system is not locked up; you might can ssh into your box from another box and make your changes without having to do the boot-up hacks I mentioned in the previous email. That would probably be easier. If you can ssh in, you can just manually fix the /etc/X11/xorg.conf file without killing the /etc/init.d/[gxwk]dm script first. (Of course, you could bypass this step anyway and just edit the xorg.conf file in your recovery boot.) Another way to get into the box is to boot the install CD in rescue mode. From the menus, you can get a shell that is chrooted into the box. From there, you can either edit xorg.conf or you may even be able to run dpkg-reconfigure xserver-xorg. Also, there's an option inbetween a normal boot and an init=/bin/sh, that is 's', for single. This will run through the /etc/rcS.d scripts and prompt for the root password. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Adaptec 2100s Debian stable
On Fri, Aug 03, 2007 at 12:29:38PM +0200, Fabrizio Pedersoli wrote: 2007/8/3, Douglas Allan Tutty [EMAIL PROTECTED]: On Thu, Aug 02, 2007 at 06:54:51PM +0200, Fabrizio Pedersoli wrote: i have a big problem with an Adaptec 2100s scsi controller, which manages 2 disks in RAID-0. the Debian installer found correctly that raid and the installation process terminates fine. But when i reboot, after linux loads i20 modules and founds hdd... it stops saying waiting for root filesystem. any suggestions??? thanks (and sorry for my english :D). What is your kernel command line, from /boot/grub/menu.lst? Doug. root (hd0,0) kernel /boot/vmlinuz root=/dev/sda1 ro initird savedefault that's grub menu.lst. The strange thing is that with the old stable Debian (sarge) everthing work s fine. Is there any way to boot Debian etch witch a 2.4 series kernel? thanks root=/dev/sda1 gives a single partition, which is fine if the hardware raid card is really recognized as hardware raid since it will present what looks like one dist to the kernel. You need to find out if the module for the card is being loaded by the initrd; I don't know how to do that. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Start booting hard drive, from a CD
On Fri, Aug 03, 2007 at 05:42:22PM +0530, Masatran, R. Deepak wrote: I am unable to boot from my hard drive due to some problem. As a temporary solution, I want to make a CD, and use it to start booting the hard drive. How can I make such a CD? Thanks! The answer will depend on why you can't boot from hard drive in the first place. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
design focus [was Large initrd, was booting problem (udev related?)]
On Fri, Aug 03, 2007 at 05:54:57PM +0300, Andrei Popescu wrote: On Thu, Aug 02, 2007 at 08:34:00PM -0400, Douglas Allan Tutty wrote: However, don't all those modules in the initrd end up staying in the kernel anyway, or do they get unloaded during boot? If they stay, and 'most' modules get added, how is that different than having a huge monolithic kernel? It may not matter on a box with huge memory, but I have mostly small-memory boxes. I may be wrong, but I think that only the needed modules are actually loaded. As for xorg-video-foo, that's why I don't install the xorg metapackage. I choose from its dependencies what I need. Same here All these extra packages together take a lot of disk space, a lot of download bandwidth to install and maintain. /rant There's a growing kitchen-sink approach in Debian (perhaps all of Linux, I don't know). There's the kernel/initrd size, there's the variable device name problems, to name two. It suggests to me that there's a missing piece of infrastructure. Perhaps the installer system should create a hardware inventory file that initrdtools (or whatever the nom de jure) can access to generate a tailord initrd, that apt can consult for what drivers to download, etc. The installer rescue mode could offer a tool to regenerate the inventory file for times when one changes hardware. /end rant True, but you have to consider the competition. I guess the problem is related to this notion of trying to compete with MS. If people 'buy' brand A because they like features x,y, and z, and brand B has the goal of gaining market share, it will tend to morph into a clone (feature-wise) of brand A. However, it will tend to take on some of the compromises of brand B that go with features x, y, and z. I stick with debian on my big box because of inertia, the debian policy, the debian security support for all packages in debian/main, and the absolute ease of applying bug fixes with aptitude. Debian also supports my trackball mouse's scroll wheel (IMPS/2) whereas OpenBSD does not. However, my older computers are transitioning away from Debian to BSD because of the newer debian (perhaps all linuxes) being so much slower on them than either older debians or new BSDs. If you plug a new device into a Windows machine the driver gets installed automatically or you get prompted for the drivers if Windows doesn't have them. You have to admit that this is pretty convenient functionality which has been there at least since Windows 2000 (how this is cluttering the registry and the fact that it isn't always working is a totally different topic). That convenience comes at a huge price in terms of system resource utilization on boxes with few resources. Compare it to OpenBSD, for example, where there is no such thing as eth0, but network interfaces based on driver name (eg. ne) and configuration; my 486 has one NIC as ne1. Its not convenient to have to look up in a file for the supported configurations of different hardware to ensure that your NIC is set up to match one of them then configure networking based on ne1. However, its only done once. The big advantage on linux (and especially Debian) is that power users still have the possibility to customize the setup (like using a different mkinitrd, different options, purge unneeded packages, ...) that a Windows user doesn't have. True, but rather than hotplugging, I would prefer a program that can be run as needed each time a new piece of hardware is attached for the first time, which would create the device node and load the appropriate module and parameters. Once done, it would get out of the way. On subsequent attachment of a device, everything would be pre-existing. It all comes down to the notion of competition and market share. If Debian is going to focus on market share and competing with MS it will have to target MS's target market. Since I'm not in that market, Debian will be shifting its focus on the market I'm in. It won't be that I'm drifting away from Debian but that Debian is drifting away from me. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Security newbie?
On Fri, Aug 03, 2007 at 08:15:38AM -0500, Rodney Richison wrote: Art Edwards wrote: I've been running debian @ home and @ work, for years, had no indication of attacks. Over the last few days, my iptables firewall seemed simply to stop. I checked my auth log file to find many, many attempts to break in. My firewall was very simple. I have since added rules to drop packets from offending IP addresses. So, I have a couple of very basic questions: 1. Are there repositories of offending IP addresses to block? Can/should one contribute to these? First ask if you need to ssh into your box from the internet, if not then limit the interfaces to which ssh listens. 2. The attacks never use the same user name more than once. Is there a way to block access, even temporarily, from an IP address after a set number of attempts, even if the attempts use different user names? If you are using good strong passwords then it shouldn't matter how many times someone tries. However, you should consider using public-key ssh logins where you can totally disable password logins. 3. Are there other obvious things I should be doing? ssh, by it's design is insecure. This seems a little harsh. It SHOULD incorporate some means of limiting password attempts. It does not! Using alternate ports can be a pain in the butt as some programs (like webmin filesystem backup) do not support alternate ports. I suggest 2 methods, fail2ban and a firewall if you must allow password logins. You can set the firewall to allow only certain ip's or ip ranges. But do not get to comfortable with a firewall ONLY solution. The first time the local firewall goes down, or is taken down and forgotten to re-enable, you'll get compromised. Yes, a firewall is the first line of defence in that it blocks things before they reach your daemons but the last line of defence in that everything else should be relied on first. You can use the firewall to limit the rate of connection attempts that go to ssh. Also, if you know the range of IPs from which you need to connect, you could limit attempts to that range. You can also look at the sshd_config options of MaxAuthTries and MaxStartups. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: nfs problem while internet access broken
On Fri, Aug 03, 2007 at 02:47:14PM +0200, Stephane Durieux wrote: I have noticed that my clients are not declared in /etc/hosts if I put them in /etc/hosts (and in nis table to be useful) everything is find. Nevertheless I don t understand how the nfs can make a resolution via dns of a a client whith a private address Perhaps does the server reply it it s impossible and everything is fine to log it /etc/nsswitch.conf (name service switch conf) has the line: hosts: files dns This means that when the resolver tries to find a host it will first look in the files on the system (e.g. /etc/hosts) before doing a dns search. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: converting file system
On Fri, Aug 03, 2007 at 11:23:06PM +0530, Bhasker C V wrote: Is there a method to convert an ext3 file system created with -T largefile4 to a normal ext3 file system with normal block sizes ? (without losing data). I don't think that -t largefile changes the size of block sizes but changes the number of blocks per inode (but I'm hazy on such details). Looking at tune2fs, I don't see any related options. So I doubt it. How you proceed depends on what mount point we're talking about. Hopefully, its not /. Anything else you can 'fix' by doing a backup, going single-user, unmount the partition, remake the filesystem, mount it, and restore the backup, then shutdown back to multi-user. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: ext3fs errors with kernel 2.6.18 but not with 2.4.27
On Fri, Aug 03, 2007 at 02:28:58PM -0400, Brad Sawatzky wrote: On Fri, 03 Aug 2007, Francois Duranleau wrote: [ . . . ] Now, when I boot my computer using the new kernel, I get an error message during filesystem check saying it contains errors (sorry, I don't have a more precise error message, and I have no logs) and it then mounts the filesystem read-only. If I boot back with the 2.4.27 kernel, no problems. I do get often though an error message like this: hda: dma_intr: status=0x51 { DriveReady SeekComplete Error } hda: dma_intr: error=0x84 { DriveStatusError BadCRC } and then sometimes: ide0: reset: success I have been having this for a long time (even after I changed to a new hard drive), and only when DMA is active, but I never noticed anything bad happening except for those error messages. [ . . . ] You have either: 1) a bad ide cable, or 2) a bad motherboard (IDE connector, timing, flakey chipset, who knows). Try using a new IDE cable and see if the error messages go away. Be sure to use an 80 wire cable and not an old ATA/33 era 40 wire cable. If you have two devices on the same cable, it could be that the second device is causing problems too. (2) is pretty unlikely unless it's a really old board. But why would he only get the errors with a new kernel? To get the 'logs', with the system in single-user with the fs mounted ro, can you mount a usbstick or floppy and copy dmesg to it? If the messages aren't there, set up a serial console and capture the messages there. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: converting file system
On Fri, Aug 03, 2007 at 03:41:52PM -0300, Jeronimo Pellegrini wrote: On Fri, Aug 03, 2007 at 02:35:09PM -0400, Douglas Allan Tutty wrote: So I doubt it. How you proceed depends on what mount point we're talking about. Hopefully, its not /. Anything else you can 'fix' by doing a backup, going single-user, unmount the partition, remake the filesystem, mount it, and restore the backup, then shutdown back to multi-user. Or use convertfs. (apt-cache show convertfs) I don't see any such package (aptitude search convertfs) in either i386 or amd64. http://packages.debian.org/convertfs doesn't show any such package. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: ext3fs errors with kernel 2.6.18 but not with 2.4.27
On Fri, Aug 03, 2007 at 03:26:11PM -0400, Francois Duranleau wrote: On 2.4.27, I get the CRC errors (not at boot time, later, and all the time thereafter), and I've been having them for many years. I mentionned just in case there might be a link with my problem. Otherwise, I am not trying to solve this particular problem. On 2.6.x, at boot time, errors are reported on the initial filesystem check. I do not know if I still have those CRC errors. CRC errors are nothing to be ignored. Either there's a drive problem or the driver hasn't worked for a while. It could be that 2.6 is less forgiving of CRC errors since they are errors that suggest that the data isn't reliable and the drive is failing or failed. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: MS Word under wine/crossover office
On Fri, Aug 03, 2007 at 08:44:31PM +, Tyler Smith wrote: I am aware that the problem lies entirely in the inconsistent and undocumented .doc format. And I know that rtf suffers from many of the same deficiencies. And I know that pdf is far preferable for most purposes. For reasons that are beyond my control I am required to submit manuscripts in .doc format, and they must conform to very explicit formatting rules. We all know this is dumb, but at the moment I have to deal with it. Please believe that I will be bringing this issue up with the academic societies that I am working with. I don't suppose that, while it has to be in .doc format, it doesn't have to be editable? Could you make each page an .eps (or other graphic image) and plonk it down on a page in OO and then save it in .doc? It would then _look_ correctly when viewed with Word and it would be a .doc file, just not an editable one. Just an idea; I've never used Word or OO. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: converting file system
On Fri, Aug 03, 2007 at 05:15:42PM -0700, Andrew Sackville-West wrote: from the RM bug report: - The final step in converting a filesystem, reordering the blocks of the target filesystem, is apparently programmed in a very inefficient way, and it can take weeks for large filesystems to complete convertfs. --^^ imagine if you didn't know that going in... In order for that to appear in the report, someone must have found out the hard way. To know that it took weeks (instead of, say, a weekend), someone had the patience of Job to know that after weeks it did complete. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: converting file system
On Fri, Aug 03, 2007 at 06:44:39PM -0500, Hugo Vanwoerkom wrote: Is there a way to convert from ext2 to anything? I think its the tried-and-true method of tar by way of spare space on some device. This is a great reason to be using LVM with separate LVs. Need to convert? Create a new LV, make the new filesystem, tar the data over to it. When everything is working, delete the old LV. If you need extra space during the transition, borrow a drive and add it to the VG. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Opinions XFS
On Fri, Aug 03, 2007 at 07:04:44PM -0500, Sam Leon wrote: Sergio Belkin wrote: Hi I was reading http://oss.sgi.com/projects/xfs/index.html and was amazed because XFS powerful features. But I'd like opinions if xfs should be a good alternative to ext3 in typical cases, or if it should be relegated to critical missions servers. From what I have read xfs and jfs can corrupt data quickly if the drive is not properly unmounted first (ie, forced reboot, power outage) People generally stick with ext3 because there is more support for it. Not this thread again. I went from ext3 to JFS because I have frequent power failures and Sarge's ext3 would get invisible mysterious errors that ended up with a corrupted file system, especially if the power failed during a fsck. At the time, I didn't go with XFS because at that time there were problems with XFS and loosing data. I haven't done a recent comparison but both file systems were developed by their companies to do slightly different things. IBM was focused on transaction-oriented servers for e-commerce. If the power failed or the server crashed, they wanted the fsck to go as fast as possible. So the filesystem will come up quickly in a good state; that some files could possibly be missing is a good reason for backups. So the notion that JFS isn't good at unclean shutdowns goes against one of the design criteria. SGI's XFS was more for compute-oriented boxes (XFS is used in the new Cray super-computers) and grahpics workstations. The filesystems can be staggeringly huge and so also need a quick fsck in the event of power failure. In both cases, down time either during fsck or fixing of missing or broken files represents real financial burden. So they're both designed to do basically the same thing from companies with two different target markets. At any given point, the difference will be how well Linux handles them; what subset of the features are implemented. When I chose JFS, XFS had some problems. Based on posts to recent threads on this topic, I believe that they both work fine now. As for the features, there was an article in the Linux Gazatte that I'm looking up now... http://linuxgazette.net/102/piszcz.html that does some real-world benchmark comparisons. Its from May, 2004 and the kernel is a 2.4. However, it may be useful. Try a google search for 'XFS JFS Linux' You've got sgi's site for XFS. Here's IBM's JFS site: http://www-128.ibm.com/developerworks/linux/library/l-jfs.html Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Opinions XFS
On Fri, Aug 03, 2007 at 08:37:31PM -0400, Jose Luis Rivas Contreras wrote: I love XFS!! The only thing I cannot do with XFS is boot from that disk so I need to make a `/boot' partition to get it to boot but the rest is works great!! Smoothly :) Why can't you boot from it? Doesn't GRUB support booting from XFS? It does boot from JFS. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Opinions XFS
On Fri, Aug 03, 2007 at 07:44:03PM -0700, David Brodbeck wrote: I have a MythTV system where there are lots of large (2+ gigabyte) video recordings. With ext3, deletes would block all writes to the filesystem until they completed, causing skips if there was an ongoing recording. With XFS I haven't had any such problems. I think there's a bit of Not Invented Here syndrome with XFS that causes people to be wary of it, but in my experience it's a rock- solid filesystem. However, it doesn't journal data, only metadata, so you may lose a bit of data if the system goes down uncleanly. The filesystem will be protected from corruption, however. (Ext3fs can also be configured this way, but its default is to journal data as well as metadata.) I think that this is another reason for having different mount points each with the best filesystem for it. At the current state (history is history), for reliability you probably want ext3, so use that on system filesystems and for storing backups. For high performance, expecially in a graphic environment, use XFS (which makes sense given XFS comes from Silicon _Graphics_), which really shines at handling large files (e.g. movies). For databases, you may want JFS, which is designed for transactional data and handles many small files well too; many databases don't do a sync so a power failure can mess up the data even if the underlying filesystem survives intact, so any loss caused by JFS would be moot. Based on this discussion, I'll be re-evaluating my current set up. Thank you. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Iceweasel 2.0.0.5 and /dev/hda
On Thu, Aug 02, 2007 at 07:01:57AM +, Steven wrote: Also, if I `dd if=/dev/hda of=test bs=512 count=16384 strings -a test | less` I see several explicit module paths, a reference to an old kernel which isn't on the system anymore (that I know of), and what looks like an irc proxy. [snip related to iceweasel] I don't know. Perhaps you should get the iceweasel source and search for the strings of interest and see why they're there. The relevant output from dd /dev/hda: [snip] What is the question re /dev/hda? 'deleting' a file on a filesystem simply removes the file name out of a directory listing and marks the inodes as free. The data is still left on the drive untill written over. Actually, even after being written over its still possible to retrieve it forensically. What you are showing is a prime example of why its a good idea to have an encrypted swap and /tmp on tmpfs so that nothing from memory or /tmp can end up on the drive enclare. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Ownership Question
On Thu, Aug 02, 2007 at 01:25:47AM -0600, Telly Williams wrote: Most of my files/folders are owned by root. It's starting to seem like that's not such a great idea because a few of the programs that I use don't like you to be root when you run them. So I'm thinking that I should start recursively changing the ownership of my files to the local user, as I'm the only user of this computer anyways. I'm not sure if I said what I said correctly, but is changing my ownership drastically like this a good idea? Isn't it bad to have so many files owned by root? If my system got compromised then they could do whatever. You should only run root when you need to run root, and your umask should be conservative. You should probably read harden-doc although realize that it is a little out of date. You should _never_ run the xserver as root. Run as a normal user and use su (or some such) to become root only as needed. Are these files/folders in /root? If so, the easiest way would be to, as root, move them into your normal user's home directory then do a chown -R user.user on the whole directory. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: nfs problem while internet access broken
On Thu, Aug 02, 2007 at 01:46:40PM +0200, Stephane Durieux wrote: --- Douglas Allan Tutty [EMAIL PROTECTED] a ?crit I encounter a problem whith my nfs stations. If the server has not access to internet, they cannot mount their directories ! I think what he means is that: client - server - internet Client can ping server but if server is not connected to internet, client can not mount nfs shares from server. client - server -X- internet That is exactly my issue . Excuse me for not being clear ! Anyway, has somebody a solution . I have read that client and server must be able to make a reverse resolution of the client and the server, but I am not sure of it and I don t have an in depth knowledge of nfs . With the internet off, check: ping from client to server by name ping from server to client by name Send us /etc/hosts, /etc/host.conf, /etc/hosts.*, /etc/nsswitch.conf Do you have a firewall that perhaps is changing config depending on internet status? What firewall do you use? Relavent error messages: 1. take a copy of /var/log/syslog on each machine. 2. Try to mount an NFS directory 3. Record any visible error messages on either box 4. Take another copy of /var/log/syslog on each machine and diff with the previous copy, and send the diffs, snipping off extraneous or sensitive data. Good luck, Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: how to set network io priority for a process?
On Thu, Aug 02, 2007 at 06:02:34PM +0300, Andrei Popescu wrote: On Wed, Aug 01, 2007 at 11:03:50PM -0400, Douglas Allan Tutty wrote: On Wed, Aug 01, 2007 at 09:49:41PM -0300, Henrique de Moraes Holschuh wrote: It certainly can do what you want, if you leave it running and use it as a shell and not as a single-command download tool. lftp can carry as many transfers in parallel as needed, to as many sites as needed, and bounce from one to another as needed. Pause them, queue them, stop them, etc. If you need to detach lftp from terminals and access it remotely, screen is your friend. It sounds like it will do everything I need except be persistant over reboots. I'm also going to look into curl. According to the description it is scriptable. You could write an initscript ... I'm considering creating some kind of user interface to pick URLs to put into a system queue and have a queue runner start on ppp/ip_up.d and end on ip_down.d. To do properly, its a big project. Similar to writing a new print spooler. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: cheap scsi card recomendation
On Thu, Aug 02, 2007 at 03:04:03PM -0500, Rodney Richison wrote: I've about 20 of old netserver lpr machines that have the old megaraid raid card. The kernal has dropped support for these cards. I'm tired of dealing with it. Can someone recomend a card that will work well with a stock debian that I might find on ebay? The drives are ultra3 18g (only raid 1 needed, do NOT need raid 5) 10rpm hard drives I don't know specifically what MegaRaid cards you have, but just FYI and not any sort of endorcement, I see that OpenBSD includes support for several MegaRaid cards. I have an old 486 that is a dog under Etch but is quite zippy under OBSD. I've found that Linux development focuses on newer and newer stuff at the expense of old stuff. I have lots of old stuff, always looking for more (free). Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: MS Word under wine/crossover office
On Thu, Aug 02, 2007 at 11:41:54PM +, Tyler Smith wrote: Which brings me to my question: do any of you have any experience using wine or crossover office with MSWord? Does it work? If it is possible to produce true, well-formatted .doc files this way then it will definitely be an improvement over hunting down a computer running Word under MSWindows to check my submissions. But it needs to be completely identical to regular Word - if I have to check the crossover/wine Word version with native Word I won't have saved any time. Is this possible? Any other suggestions regarding producing true .doc formats without resorting to installing Windows would be welcome, but in this case OpenOffice is definitely not good enough. It's great, of course, but not when it comes to exchanging .doc files with critical formatting intact. Do the people to whom you send the word docs have to edit them? If not, what about sending pdf? I never send anyone a doc file, and I've never used OpenOffice; I just make a pdf from LaTex. Just lucky, I guess. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Bridged Network Question
On Thu, Aug 02, 2007 at 03:09:48PM -0400, Wayne Topa wrote: I have been using an old laptop as an Access Point for our laptops to connect to the internet through the main box - modem connection. The Lan (eth0) is bridged with a Netgear WG511U PCMCIA card (ath0) to connect to the gatway computer. It works fine, as an AP, but has one problem. The AP itself can not connect to the internet through the gatway. The only way to update the AP's software is to change /etc/network/interfaces from the bridged setup to just a Lan setup, reboot, do the update/upgrade, change the interfaces file back to the bridged mode and reboot again. I would like to move the AP to an old 500Mhz headless box so that I can setup the firewall and a mailserver, etc on it and free up the laptop. I can't see how to do that with the above problem. I've never needed bridge and I'm wondering why you do? Bridging makes the two networks look like one, but expecially when you start firewalling, they need to be different. For details on setting this up, read the shorewall-doc package, even if you don't want to use shorewall. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Adaptec 2100s Debian stable
On Thu, Aug 02, 2007 at 06:54:51PM +0200, Fabrizio Pedersoli wrote: i have a big problem with an Adaptec 2100s scsi controller, which manages 2 disks in RAID-0. the Debian installer found correctly that raid and the installation process terminates fine. But when i reboot, after linux loads i20 modules and founds hdd... it stops saying waiting for root filesystem. any suggestions??? thanks (and sorry for my english :D). What is your kernel command line, from /boot/grub/menu.lst? Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Large initrd [Was: Re: booting problem (udev related?)]
On Fri, Aug 03, 2007 at 12:19:36AM +0300, Andrei Popescu wrote: On Thu, Aug 02, 2007 at 10:35:01AM -0700, Andrew Sackville-West wrote: So what is the significance of initrd size? (other than the obvious filling up /boot issue). Is it really a problem to have most modules in there? I can think of some situations where it might be nice to have most of them -- mobo fails catastrophically and you want to be able to just boot, for example. This is about it. Debian wants to provide an initrd that works even ehn changing hardware. Same reason for installing all -xorg-video-foo packages. However, don't all those modules in the initrd end up staying in the kernel anyway, or do they get unloaded during boot? If they stay, and 'most' modules get added, how is that different than having a huge monolithic kernel? It may not matter on a box with huge memory, but I have mostly small-memory boxes. As for xorg-video-foo, that's why I don't install the xorg metapackage. I choose from its dependencies what I need. /rant There's a growing kitchen-sink approach in Debian (perhaps all of Linux, I don't know). There's the kernel/initrd size, there's the variable device name problems, to name two. It suggests to me that there's a missing piece of infrastructure. Perhaps the installer system should create a hardware inventory file that initrdtools (or whatever the nom de jure) can access to generate a tailord initrd, that apt can consult for what drivers to download, etc. The installer rescue mode could offer a tool to regenerate the inventory file for times when one changes hardware. /end rant Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Bridged Network Question
On Thu, Aug 02, 2007 at 09:47:58PM -0400, Wayne Topa wrote: Douglas Allan Tutty([EMAIL PROTECTED]) is reported to have said: On Thu, Aug 02, 2007 at 03:09:48PM -0400, Wayne Topa wrote: I have been using an old laptop as an Access Point for our laptops to connect to the internet through the main box - modem connection. The Lan (eth0) is bridged with a Netgear WG511U PCMCIA card (ath0) to connect to the gatway computer. I think I understand your setup, although I've never used wireless. I too am on dialup and at one point had a laptop that I connected with a serial cable to a box that then connected to the box that had the modem. I set up the access box with NAT at the time by simply installing ipmasq and dnsmasq and setting forwarding. All boxes could access the internet. I never needed bridge and wonder why you're using it. As I understand it, bridging is for making stand-alone boxes that perform the same function as the various ubiquetous blue ones: routers, switches, etc. They themselves don't need to access the internet. Anyway, if you're happy to bridge, good, but I don't know how to make it do what you want. Good luck. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Bridged Network Question
On Thu, Aug 02, 2007 at 10:45:43PM -0400, Wayne Topa wrote: Douglas Allan Tutty([EMAIL PROTECTED]) is reported to have said: On Thu, Aug 02, 2007 at 09:47:58PM -0400, Wayne Topa wrote: Douglas Allan Tutty([EMAIL PROTECTED]) is reported to have said: On Thu, Aug 02, 2007 at 03:09:48PM -0400, Wayne Topa wrote: I have been using an old laptop as an Access Point for our laptops to connect to the internet through the main box - modem connection. The Lan (eth0) is bridged with a Netgear WG511U PCMCIA card (ath0) to connect to the gatway computer. I think I understand your setup, although I've never used wireless. I too am on dialup and at one point had a laptop that I connected with a serial cable to a box that then connected to the box that had the modem. The AP laptop -is- connected to the gatway machine via Cat5 cable to the gateway. I understand that. I set up the access box with NAT at the time by simply installing ipmasq and dnsmasq and setting forwarding. All boxes could access the internet. I never needed bridge and wonder why you're using it. Because I needed a way to connect the wireles laptops to the internet. The AP has the the wireless adapter and an ethernet adapter in it do does what you did with your laptop except it accepts wireless signals in/out. I'm about as happy with the bridge, as it is now, as I am with my 2.5KBs dialup connection. ~6 minutes for each Meg downloaded is a bigger PITA but I don't have any other alternative, currently. You don't need to bridge. If it were me trying to set this up, here's how I'd do it. IP numbers are examples only. Internet 209.181.8.8 = ppp0 Gateway box, masqurade using iptables eth0: 192.168.1.1 = eth0: 192.168.1.2 AP laptop, masqurade using iptables. ath0: 192.168.2.1 = wireless0: dhcp e.g. 192.168.2.2 user's laptop A bridge gives all boxes the same network number but this isn't required for your setup. All your user's laptops would be on network 192.168.2.0, and their default gatway would be 192.168.2.1 The only things different from what I do all the time is the dhcp and wireless (never used either). The iptables setup I've always done with either ipmasq (no config required) or for more control shorewall. Good luck, I hope this helps. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: new Etch install fails to boot
On Mon, Jul 30, 2007 at 10:19:09PM -0400, Steve Kleene wrote: [I wrote that my fresh Etch install calls grub and then stops.] Here's a fresh start, just to verify that your machine will actually boot properly. 1. Connect your drives to /dev/hda and /dev/hdc, set the jumpers on both drives to master. Alternatively, just remove hdc for now. 2. Boot the installer and go to a shell. 3. Clear the beginning of the drives, which includes the MBR: dd if=/dev/zero of=/dev/hda bs=512 count=2 ;sync dd if=/dev/zero of=/dev/hdc bs=512 count=2 ;sync 4. exit the shell and return to the installer. 5. Run the install, just the base system (don't select any tasks). 6. Partition the drives thus: hda1/boot 32 MB hda2swap128 MB hda3/ remainder You don't need hdc for this. 7. Install the grub onto hda (not hda1 or other partition). 8. Try to reboot. If it doesn't work, reboot the installer in rescue mode and tell it to install grub again in hda. As for the grub-disk, if you mount it you should see a default menu.lst file. Therefore, when you boot it, you should get a menu on the screen. You may need to ensure that it got copied correctly. Use dd to make an image file of the floppy you created and then compare the md5sum of both images. They should be the same. Good luck, Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Debian Linux in Chroot
On Tue, Jul 31, 2007 at 02:31:07PM +0530, Masatran, R. Deepak wrote: I have two Linux installations in my hard drive, and I want to modify Linux-2 from Linux-1, using Chroot. Basically dpkg-reconfigure and similar stuff. How do I tell DPKG of Linux-2 to not disturb the daemons that are running in Linux-1? The easiest way is to, on Linux-1, install schroot. Then tell schroot how to access the chroot. It then takes care of mounting anything required. For example, on my amd64, I have an i386 chroot installed under /srv/chroot, and here's my schroot.conf: [etch-ia32] type=directory description=Debian Etch ia32 groups=ssh run-setup-scripts=true run-exec-scripts=true personality=linux32 location=/srv/chroot/etch-ia32 The groups=ssh is a security feature; only people who I trust to run ssh can run schroot. I generally schroot into the chroot and then run the command. To do so its just: $ schroot -pc etch-ia32 The -p means to bring in my environment: usefull if I want to run an X app. The -c is telling schroot which chroot to use. I understand that your chroot in this case is also a fully bootable instalation on its own, with its own kernel. However, that kernel and its daemons will not run under a chroot, only when it is booted. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Debian 4.0 on AMD64 and SATA with multi-arch DVD
On Tue, Jul 31, 2007 at 01:07:32PM +0530, Masatran, R. Deepak wrote: ALSO: I want to try the 486 linux-image, but dpkg refuses to install it, saying that that package is only for I386. How can I install the 486 linux-image on my AMD64 computer? You can't since the 486 kernel is 32-bit and you've installed a 64-bit system. If you want to run an amd64 with a 486 kernel, you'll have to reinstall i386 instead of amd64. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: mounting usb frustration
On Mon, Jul 30, 2007 at 11:37:52PM -0500, Ron Johnson wrote: On 07/30/07 20:13, Carl Fink wrote: USB detection has been broken on Debian for years, literally. It works fine for me with removable drives, but my Testing system will detect my Palm device once -- and never again, until I reboot. Then I can sync once more. It always works perfectly for me plugging thumb drives and a digital camera. I've taken to using UUIDs for permanent mount points, since pmount seems to bypass udev. And udev is (was?) in such flux and the rules changed on me. Here's a snippet from my /etc/fstab: UUID=c207a86c-91ac-4733-9760-93b0389e193d /media/backup \ ext3,ext2 defaults,noauto 0 2 For a shorter fstab entry, can you label a digital camera and then use LABEL=camera instead? Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Part3: More problems. [What was the subject?]
On Tue, Jul 31, 2007 at 02:03:54AM -0500, Brad B wrote: Thanks for helping me. I installed both of those, and it got past requesting libc! Now, it's requesting the kernel source, which i can't seem to find the appropriate version of. Would you care to change the subject line to something meaningful, and where possible to stick to one thread? Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: how to set network io priority for a process?
On Wed, Aug 01, 2007 at 11:08:06AM -0300, Henrique de Moraes Holschuh wrote: On Sun, 29 Jul 2007, Douglas Allan Tutty wrote: What we need is a multi-protocol proxy server that does proper throttling of download requests. Squid delay pools? Will work for http and ftp. But not rsync, which I use whenever I can for large downloads due to errors creeping in for some reason over my noisy phone line and freqent line drops (and susequent redials by pppd). Such a multi-protocol proxy server doesn't need to do any caching, it just needs to put a variable delay on the outgoing ACKs that trigger the next download block on ftp and rsync protocols, variable on whether there's an active http connection. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Udev. Problems with ordering hardware using /dev/video
On Tue, Jul 31, 2007 at 11:21:27PM +0200, Florian Kulzer wrote: This flip-flopping is, AFAIK, more the fault of the newer kernel then the fault of udev. If the kernel would always load the modules in the same order then udev would probably assign the device nodes in a consistent manner. I disagree. Udev is responsible for creating device nodes. It should remember what devices nodes it has made in the past for different devices, and recreate them the same way in the futuer. Yes, this could mean that if a webcam is video0 and a TV card is video1 that if the webcam is not connected there will be no video0, but that would be fine. Over the life of a box, I could see that there may be videos starting with video20. The only problem with this is that we enumerate drives with letters instead of numbers. This whole udev mess, and devfs before it, was to address the rampant growth in device nodes/names. It seems to have brought with it a whole slew of its own problems. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: ntfs mount errors
On Wed, Aug 01, 2007 at 04:49:04AM +0200, pinniped wrote: (quote) I still have the problem. ie. The windoze partition is mounted automatically fine, but I can only cd to it if I am root. (end quote) Do: man mount Look at the 'Mount options for ntfs'. All your mysteries are explained there - for example: ... By default, the files are owned by root and not readable by somebody else. So you need to set the uid/gid and umask. You really want 'root' to be the owner anyway so I guess you only want to change the gid to the 'disk' group and make sure you have a sensible umask. Be careful there, members of the 'disk' group can do nasty things on your debian box too. Perhaps create a new group who can access the windows partition, put those users into the group, then use the gid= parameter in fstab. However, this doesn't affect the uid. I can't test this further since I haven't run windows since 3.1 and never played with ntfs. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Swap configuration for 16GB of RAM, 8 cores
On Tue, Jul 31, 2007 at 08:18:09PM -, [EMAIL PROTECTED] wrote: I am installing two servers, each with 16GB of RAM, two quad-core Xeon processors, and a SATA hard drive. The machines will be compute servers, meaning lots of concurrently logged in users, each running an assortment of jobs, and various long-running processes. The jobs will be vastly dissimilar with regards to using the memory, disk, and CPU, and to their run time, so the configuration should be generic, general purpose. What are the current best practices with regards to swap partitions? Is it better to create one big, or several smaller swap partitions? Is the rule of thumb still RAM*2 for the total size? We are running Debian 4.0/Etch with the stock -i686-bigmem kernel. Try this: Use LVM for everything. Actually, grub may or may not work on LVM so put a regular partition (32 MB is more than ample) at the start (sda1) and the remainder of the disk as sda2, as physical device for LVM. Then set up your usual separate 'partitions' under LVM. To be generous, this means a / of 512 MB, a /usr of 4 GB, a /var of 6 GB plus whatever other stuff you need for your /var, a separate /srv if needed of whatever size, and /home. Your choice of filesystem is another matter; it should be growable and reliable, shrinking is less of an issue. I use JFS. Then put swap on LVM as well. Since you may need to increase swap size, and I don't know how to do that with the swap online, you may want two swap partitions so that one can be off line for a couple of minutes. Total swap size depends on how much VM gets assigned. Since disk space is cheap you may want to start with RAM*2 so you could start with two 16 GB swap partitions. Note, however, that I run amd64. I don't know what the swap size limits are on a Xeon. In combination with your swap partitions, you could use swapd to dynamically make swap files. You can monitor their existance and as needed, take a swap partition off-line (swapd should then automatically increase swap-file size), increase it with LVM, mkswap again on that partition, then swapon that partition. Good luck, Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: how to set network io priority for a process?
On Wed, Aug 01, 2007 at 01:03:52PM -0300, Henrique de Moraes Holschuh wrote: On Wed, 01 Aug 2007, Douglas Allan Tutty wrote: But not rsync, which I use whenever I can for large downloads due to errors creeping in for some reason over my noisy phone line and freqent line drops (and susequent redials by pppd). Why do you allow for damaged packets at all? I used analog async ITU-T V42 modems for a *long* time (fortunately, I was able to move away before V9x hit the market). You really want an error-free channel without compression for regular Internet over PPP domestic use, and any modem still on the market should be able to deliver that to you just fine. Just configure it for error correction without compression, and have a proper error-free short and well-shielded serial connection to it if it is an external modem. Its an external 3Com Courier modem, that's not the problem. I don't know _where_ the errors happen, but the more times the download is interrupted the more frequently there are issues somewhere in the iso. Not a problem if I can find an rsync server with the file, but a royal pain with plain ftp. I _once_ had to download three defective copies of an ISO (took a couple of weeks) and use that gnu tool that creates a single good file out of three bad ones. It has never been a problem when downloading packages via aptitude, even if I have to interrupt it. However, under sarge, I found that gftp would die out and not resume properly, corrupting the file, and that sometimes wget will also corrupt the file. For all I know, the errors are happening upstream of my ISP's modem. I'm using porchlight which gives me unlimited for 9.99 per month. This never happened when I lived outside of Parry Sound, farther away from the telephone exchange, but paid $24.95 per month. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: IM on a home debian network
On Tue, Jul 31, 2007 at 10:16:32PM -0400, Kevin Mark wrote: I just started to use screen and it has 'control-a ?' which given you the 'cheatsheet'. But you dont need all the commands to start using it. I use: 'screen MYCOMMAND' to start a new screen session control-a for menu-based session selection control-a n for 'go to next screen session' 'screen -list' to list the sessions there are many more, but it is easy to try and then ask here for more advanced options. Yea, I just tried it again and it only took me 20 seconds to hit something (wrong key?) and its frozen that vt, nothing works. I tried killing its process and it remained frozen. Luckily, I had ssh'd in to the box so I could kill the ssh session from the originating box. I know, the GNU folks are in love with Ctrl-* * commands (witness info) but why they can't give a screen like in minicom with drop-down menus or something... Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: nfs problem while internet access broken
On Wed, Aug 01, 2007 at 11:00:19AM -0600, Bob Proulx wrote: Stephane Durieux wrote: Of course the network connexion is good ! I can ping the server But you said that the server cannot access the network: I encounter a problem whith my nfs stations. If the server has not access to internet, they cannot mount their directories ! This is a contradiction. You say that the server cannot connect to the internet and you say that you can ping your server. Please explain. I think what he means is that: client - server - internet Client can ping server but if server is not connected to internet, client can not mount nfs shares from server. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: how to set network io priority for a process?
On Wed, Aug 01, 2007 at 01:41:56PM -0300, Henrique de Moraes Holschuh wrote: Try lftp. I know of no better ftp client. But it is command-line, which is just as well: the transfer engine is well cared for, and not a secondary thing to the GUI. I've got lftp installed but haven't tried it. I will. However, I would have thought that wget would be the cat's meow since its what debootstrap uses. Perhaps they only use it because its small. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: syslog help
On Wed, Aug 01, 2007 at 12:54:23PM -0400, Tony Heal wrote: I am trying to move my postgresql (v7.4) logging out of the syslog and into a postgres.log. I want to do this so I can turn on logging full bore and have logrotate keep the logs to a reasonable size. Here is my problem. Postgresql v7.4 will either log to STDOUT, syslog or none. I have been looking at syslog.conf and trying to set this up and I have what looks good, but I still get the entries in the syslog. Here is what I have for syslog.conf Local0.* /var/log/postgresql/postgres.log I also tries this with cron and that did not work either. Is this even possible? Can I redirect entries out of the syslog into another file, or does this facility only copy the entries to another file, while maintaining them in the syslog. My /etc/syslog.conf has the following line: *.*;auth,authpriv.none -/var/log/syslog The *.* means that _everything_ logged to the syslogger goes into /var/log/syslog. man syslog.conf tells us that with one line where terms are separated by ';', latter terms override earlier terms. Assuming that your Local0.* gives you the log you want, you should be able to append ';Local0.none' to the *.* line to exclude them from /var/log/syslog. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: mounting usb frustration
On Wed, Aug 01, 2007 at 12:14:01PM -0500, Ron Johnson wrote: On 08/01/07 08:05, Douglas Allan Tutty wrote: For a shorter fstab entry, can you label a digital camera and then use LABEL=camera instead? You can do it with fixed disks that have decent filesystems, but I doubt that anything partitioned with vfat would work. Can't you use mlabel? Or mkfs.vfat with the -n parameter? Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: how to set network io priority for a process?
On Wed, Aug 01, 2007 at 03:33:14PM -0300, Henrique de Moraes Holschuh wrote: On Wed, 01 Aug 2007, Douglas Allan Tutty wrote: On Wed, Aug 01, 2007 at 01:41:56PM -0300, Henrique de Moraes Holschuh wrote: Try lftp. I know of no better ftp client. But it is command-line, which is just as well: the transfer engine is well cared for, and not a secondary thing to the GUI. I've got lftp installed but haven't tried it. I will. However, I would have thought that wget would be the cat's meow since its what debootstrap uses. Perhaps they only use it because its small. wget is for script use. lftp is for interactive use :-) Except that for a download that I have to restart 5 or 10 times, its easier to put the url in a file and use wget, or for rsync I put the whole command line in a file, pound-hack it, chmod +x and away it goes. If lftp had a download queue that was persistant between invocations, it would be useful. And AFAIK curl is supposed to be better than wget for script use, but I won't go into that topic. I'll look into it. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: how to set network io priority for a process?
On Wed, Aug 01, 2007 at 09:49:41PM -0300, Henrique de Moraes Holschuh wrote: It certainly can do what you want, if you leave it running and use it as a shell and not as a single-command download tool. lftp can carry as many transfers in parallel as needed, to as many sites as needed, and bounce from one to another as needed. Pause them, queue them, stop them, etc. If you need to detach lftp from terminals and access it remotely, screen is your friend. It sounds like it will do everything I need except be persistant over reboots. I'm also going to look into curl. Thanks, Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Caps Lock don't work
On Wed, Aug 01, 2007 at 09:12:26PM +0200, Salve H?kedal wrote: I run Etch on 3 machines, and have trouble with Caps Lock in the linux console. (In X it's ok.) machine[1a] Installed Woody in 2004, upgraded to Sarge and then Etch. Caps Lock works. machine[2] Installed Sarge, later upgraded to Etch when stable. Caps Lock won't work. Gives a mix of upper and lower case. machine[3] Installed Etch some months ago. Caps Lock won't work. Gives a mix of upper and lower case. machine[1b] Installed Etch some days ago fresh on another partition. Caps Lock won't work. Gives a mix of upper and lower case. My default locale is nn_NO.utf8, but I have tried switching to en_GB.utf8 or C with no better result. As the two installations on machine[1], behaves different, it seems to me that it must be a software issue? I used the same problem on my i386 since I switched to Etch. The caps lock worked but but numbers were also shifted. The only time I noticed it was when typing Canadian postal codes (mine is K0K 3N0) which is easy with caps lock but not with the shift. I was about to say I _have_ the same problem but I guess its been a while since I typed a postal code, testing just now I don't have that problem. The only change I've made is that I have changed local to C and purged the locales package, and I've switched back from terminus font and purged the console related packages (e.g. console-setup). Based on this, you could test by purging the console- stuff if you're using a regular legacy keyboard (as I am) but I don't know if its needed for a USB one. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: booting problem (udev related?)
On Thu, Aug 02, 2007 at 08:50:08AM +0900, Miles Bader wrote: I seems like it may be related to udev because if I look in /dev, the disk device nodes which should be there _aren't there_, even though the disk hardware is recognized fine by the kernel. Udev isn't running yet. The boot devices/modules are loaded in the initramfs. I've never compiled a kernel so I haven't had to fitz with initramfs. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Swap configuration for 16GB of RAM, 8 cores
On Wed, Aug 01, 2007 at 07:47:46PM -0700, [EMAIL PROTECTED] wrote: Have you considered using the AMD64 port?, or perhaps you have some custom 32 bit stuff that can only run on x86? Even if he does, it would probably run better in an i386 chroot. Unless he's developing the i386 kernel. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: new Etch install fails to boot
On Sun, Jul 29, 2007 at 10:28:04PM -0500, Mumia W.. wrote: If you can, try to get the boot files placed before the 1024th cylinder boundary. Sometimes this is at 0.5GB, 2.1GB or 8GB. Try a partition layout like so: /boot (primary #1, 2.1GB) / (primary #2, 37.8GB) swap (logical #5, 1.5GB) You shouldn't need a 2.1GB /boot. I find that 24 MB is fine. Hey, splurge and make it 32 MB. Since we don't know what the problem is, better be safe and assume that the boundary is as 512 MB. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: new Etch install fails to boot
On Mon, Jul 30, 2007 at 08:24:08AM -0400, Steve Kleene wrote: On Sun, Jul 29, 2007 at 11:00:30PM -0400, Steve Kleene wrote: [I wrote that my fresh Etch install calls grub and then stops.] On Sun, 29 Jul 2007 23:19:40 -0400, From: Douglas Allan Tutty replied: What happens if you reboot the installer in rescue mode and tell it to install grub again? I don't know how to do this yet, but it sounds like it's worth looking into. I'm hoping not to have to run the whole build again. The installer's rescue mode (at the boot prompt, instead of typing 'install', just type 'rescue') is designed to rescue an already installed system. It will not reinstall from rescue mode. I also gives you the option of a shell chrooted into your installation where you can run commands as if it had booted normally. Does the box have a floppy and do you have a grub-disk (I've never made a grub-stick)? Will that get you to a grub command line? It does have a floppy. I do not have a grub-disk. I do have a second (newer) box that is happily running Etch. Then on that box, install the grub-disk package. It gives you a disk image which you write to a floppy with dd: dd if=grub-disk.img of=/dev/fd0 bs=1024 conv=sync; sync If that box has grub installed and you have the grub-doc package, there are instructions for putting grub onto a floppy from within the grub command line. And on Sun, 29 Jul 2007 22:28:04 -0500, Mumia W.. wrote: If you can, try to get the boot files placed before the 1024th cylinder boundary. Sometimes this is at 0.5GB, 2.1GB or 8GB. Try a partition layout like so ... This is exactly what I always did with Red Hat and lilo on a drive that shared Windows and Linux. I could easily try this again but thought it should be unnecessary for two reasons. First, I am using grub now, which I thought supported lba by default. Second, without the whole drive allocated to Etch (i.e. no Windows partition at the start of the drive), I imagined the files needed by grub would not be placed past cylinder 1024. But maybe that's unpredictable. Just because grub can find something doesn't mean that your bios can boot it. Just to save the headache later, especially if I move the drive from one computer to another, I _always_ put /boot in the first partition on its own. If I have two drives, I'll put it on a raid1 partition for good measure. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: GTK+ File dialog hangs
On Mon, Jul 30, 2007 at 12:35:08PM -0500, Ron Johnson wrote: The closest to that which I have seen is that the larger a directory gets, the longer it takes to open a GTK File dialog. And it just burns up the CPU. If there's more than 12000 files, the File dialog will never open. Wow. I've never had a single directory with 12000 files. What file dialog have you tried that does work? How do KDE file dialogs handle it? Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: dpkg purge problem
On Tue, Jul 31, 2007 at 01:15:19AM +0300, Sasho Angelov wrote: When I try to purge old linux image: sudo dpkg --purge --force-all linux-image-2.6.14-2-k7 Why do you need to force things? This may be a clue. What package manager do you use and why don't you use it to remove this? [snip] Updating /boot/grub/menu.lst ... done And stop here. Only I can do is to press Ctrl+C then result is: dpkg: error processing linux-image-2.6.14-2-k7 (--purge): subprocess post-removal script killed by signal (Interrupt) Errors were encountered while processing: linux-image-2.6.14-2-k7 /var/lib/dpkg/info/linux-image-2.6.14-2-k7.list is empty file -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: mounting usb frustration
On Mon, Jul 30, 2007 at 09:13:03PM -0400, Carl Fink wrote: USB detection has been broken on Debian for years, literally. It works fine for me with removable drives, but my Testing system will detect my Palm device once -- and never again, until I reboot. Then I can sync once more. Care to give us the relavent sections of your /var/log/syslog and dmesg? What software are you useing to connect to your Palm? What device name have you told the software to use? Its been a while since I hooked up my palm (the data hasn't changed in 6 months), and I've reinstalled Debian since so don't have it set up. However, I remember that in gnome-pilot there was some confusion over the actual device to tell it. If I was too specific, and the kernel alternated between two device names, it couldn't find it the second time. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Not able to find qwtplot3D
On Mon, Jul 30, 2007 at 07:46:29PM -0500, Ashivni Shekhawat wrote: I am new to Debian. I have trying to install qwtplot3d on my machine; however I get the following error: # apt-get install qwtplot3d Reading package lists... Done Building dependency tree... Done E: Couldn't find package qwtplot3d http://packages.debian.org/qwtplot3d says that the package doesn't exist. However, if you follow the link to search for package names containing that word, you find lots of libqwtplot3d. However, they are all in either testing or unstable. Since you're running Etch, apt won't find it. All packages for all branches are kept in pool; hense the name. Its the packages file that tells apt what packages are valid. The package qwtplot3d is available on ftp.debian.org at http://ftp.debian.org/debian/pool/main/q/qwtplot3d/ my sources.list reads cat /etc/apt/sources.list # deb cdrom:[Debian GNU/Linux 4.0 r0 _Etch_ - Official i386 CD Binary-1 20070407-11:55]/ etch contrib main deb http://ftp.debian.org/debian/ etch main deb http://ftp.us.debian.org/debian/ etch main deb http://security.debian.org/ etch/updates main contrib deb http://http.us.debian.org/debian/ etch main Why do you have three lines pointing to three identical mirrors? Why do you want security support for contrib when you don't have contrib from the mirrors? Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Asus P5K WS motherboard / Marvell IDE - CDrom not detected by installer
On Sun, Jul 29, 2007 at 03:55:18AM -0800, Mike Messick wrote: I'm using the latest debian-testing-amd64-businesscard installer on an Asus P5K motherboard with an ICH9 Southbridge and Marvell 88SE61xx SATA / PATA chipset. I'm using an HP dvd1040i CDrom drive for installation. The installer boots fine but cannot find any cdrom device to use. When I look at dmesg, I get the following output relating to the ide devices: [snip dmesg] an lspci yields the following IDE devices: [snip lspci] So it looks like my 2 hard drives are showing up just fine, but I'm fairly clueless as to why the kernel (2.6.21-2-amd64) isn't finding the CDrom drive. I've verified that the Marvell IDE/eSATA controller is enabled in the BIOS as well (hence the initial booting of the installer CD). I'm hoping someone on the list can point me in the right direction; if this isn't the appropriate list then please accept my apologies for the noise. Its not noise. However, check out the installation manual. It tells you how to submit an installation report via the BTS with email to bugs.debian.org to the (I think) installation-reports psudo-package. The 'maintainer' for the reports is the debian-boot team who write the installer. I would suggest that you subscribe to debian-boot, follow the instructions in the manual and submit an installation report. Since they know how the installer works, they can be of more specific help. They certainly were when I was in your shoes 8 months ago. Good luck, Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: /bin/login listening?
On Sun, Jul 29, 2007 at 12:48:16PM +, Tyler Smith wrote: On 2007-07-29, Jeff D [EMAIL PROTECTED] wrote: I ran rkhunter again, and then for good measure I aptitude --purged it, reinstalled, and ran again. And then I thought maybe the whole thing was compromised, so I purged it again, installed rkhunter 1.30 from sourceforge, and ran again. And I also ran chkrootkit. In all cases they showed nothing happening, except for warning me that some of my /bin executables had been replaced by scripts -- stuff like egrep, fgrep etc. So perhaps it was just a false positive. I'm going to read up on security stuff now, so maybe I'll have some idea how to proceed the next time. Its tricky. If you have been rooted, you can't trust anything on the system, including aptitude. As for reading, try the package harden-doc. Good luck. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: lenny: getting non-free nvidia drivers issue
On Sun, Jul 29, 2007 at 06:19:47AM -0500, Hugo Vanwoerkom wrote: Yes I'm running amd64. However, since I'm on slow dialup and the amd64 box is the only one with a CD burner and lots of drive space, it would be a major headache to have something happen that caused stuff to not work. But what could happen if you tried it on a duplicate partition? So it doesn't work. You go back to the working partition. The box uses LVM so I can't give Lenny its own actual disk partitions. Also, downloading a whole new installation is non-trivial at 28.8 Kb/s (sometimes 14.4, sometimes 9600). I'm many miles from the telephone exchange. It took a few days to install an ia32 chroot with just base, mc, lynx, and iceweasel. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: /bin/login listening?
On Sun, Jul 29, 2007 at 03:56:08PM +, Tyler Smith wrote: So if I'm compromised nothing is safe, and the only guaranteed way to clear this up is to format my harddrive and reinstall. Given that the only evidence of a problem is a warning about /bin/login listening from rkhunter, which happened only once, and I have had no other problems with my net connection or general performance of my laptop, let alone mysterious withdrawals from my bank account or other signs of stolen passwords, what should I be doing? From the advice received and what I'm reading, I'm getting two very different messages - I must reinstall to be 100% certain that I'm safe, and while I can't be 100% certain I'm safe it's pretty unlikely that I have a real problem. What would you do in my situation? Try this: Boot the box from something like the install CD, go to a shell, mount your / partition ro, noexec. I think the install CD has md5sum installed. Run: #md5sum /bin/login. On my i386, I get: 2ee32ff74e474c4d9fc9df6f1460980f /bin/login If /bin/login is fine, then I'd forget about it. If it differs, I'd wipe the drive and reinstall; from backups before your first indication of a problem. Then examine the difference between that backup's data and your most recent backup. Actually, to put your mind at ease, I've attached a file bin-MD5SUMS which is the output of: $md5sum /bin/* bin-MD5SUMS Put this onto a floppy and mount it when you boot your install CD. Then edit it so that, for example the /bin/login reads /mnt/bin/login. You can then verify the whole /bin with #md5sum -c bin-MD5SUMS Here's the file, and good luck. Doug. be2bfd8feb6bfb826593c087817be9d5 /bin/arch 72e1a7bbf8478e3dd08693bec6f4c50e /bin/bash 01fcfa4919953518bbbc97b2637a27ad /bin/bunzip2 a60f3c2c4dcedeec5b0e6cce4fd777c8 /bin/busybox 01fcfa4919953518bbbc97b2637a27ad /bin/bzcat dfaba3a92070a1881dd8ec64a26069a4 /bin/bzcmp dfaba3a92070a1881dd8ec64a26069a4 /bin/bzdiff 2b11565d85da178b3a1942a22d20c624 /bin/bzegrep ea97408418bc4c3a77c0048003198acc /bin/bzexe 2b11565d85da178b3a1942a22d20c624 /bin/bzfgrep 2b11565d85da178b3a1942a22d20c624 /bin/bzgrep 01fcfa4919953518bbbc97b2637a27ad /bin/bzip2 d231db40e391032509c4c4782653cb6e /bin/bzip2recover e243255b6cf3b9403df53cb9cd6176e1 /bin/bzless e243255b6cf3b9403df53cb9cd6176e1 /bin/bzmore c12e12da393d90fba841aa678aef5094 /bin/cat 117baf5142bb451a8a0c501cdbf43726 /bin/chgrp aa1ab822de26dd9d455c8ac9163ba30e /bin/chmod b28ba00d8345041e4955ed970ed174ee /bin/chown a096cd237ee340b66f84a7867a2da2a7 /bin/cp 901cc68b293e3249a681ab4f396d1cd4 /bin/cpio a9a89a3beefb30729ea4ae80d6335cb6 /bin/csh 2af9162bd0c10ecd3b77983a56d79f6c /bin/date 02aec16981ffee391d957a28cd1190af /bin/dd 53f20746bb14718e54a65b86510bcb82 /bin/df 1c4d91adb9b1fa383247d0334a389975 /bin/dir 5c54d6f8b6af629e4be985f52c21adb6 /bin/dmesg 638cead25982bc413a287e30a6b3fea4 /bin/dnsdomainname 177e77531159a20fbcf741136c02ce05 /bin/echo 73a8a6f1948231171a6586aef43f26a6 /bin/ed 1a1c4e75e82a51bc570350aa22184913 /bin/egrep 28b23332333e80869b5810c4105392c6 /bin/false 01b9524c8e60a5e167132a6e85452cd0 /bin/fgrep 5d3ff43e62be5f980abeb4100a018ff1 /bin/fuser d274e7a42d015822ea25fb08ed19262c /bin/grep df40328a2c30b3dd195ef2f55d60cef4 /bin/gunzip cd4aee768f1e3db05aac2b3f5a6219ae /bin/gzexe df40328a2c30b3dd195ef2f55d60cef4 /bin/gzip 638cead25982bc413a287e30a6b3fea4 /bin/hostname 01c8af0fc0fe16eab70368389a5482bb /bin/ip aca6202f58b4e514ac9c0501505c2076 /bin/kernelversion 083ec3e06bc9de75e00fcb6d6292b378 /bin/kill 2f67f424360319c65ab68c27984f4d06 /bin/ln 2ee32ff74e474c4d9fc9df6f1460980f /bin/login 3a409d2e7d87fa96c89650c6aec35ac7 /bin/ls 8903244917679b8f5a19909e7e5d0fcc /bin/lsmod 432c653790fe9d2562f0894bb922d46d /bin/lsmod.modutils e89d8739e436bf722668b838476d65cb /bin/lspci 2b71253ac2aa883f6b65cc4d636fe8c8 /bin/mkdir 95887a0809f5a6de47e26d8b60ae28b1 /bin/mknod 641ec128955d32c613c201d45a9bf224 /bin/mktemp cc51af5002e2d41a84aecb14fc9cbd79 /bin/more 27c66448968d6775d3f61ee07938938c /bin/mount dcfe6fa0df8251d56c7f6cd738181003 /bin/mountpoint 0658725a01811e897497f24838c79e75 /bin/mt 0658725a01811e897497f24838c79e75 /bin/mt-gnu 45fc16400d06a4cf9d69c8d619f9104b /bin/mv 68de2870b06443403332c81022010a24 /bin/nano f0169e77f969e17e013c295cd74346a6 /bin/nc f0169e77f969e17e013c295cd74346a6 /bin/netcat e00b5e934dfa34a968b33cb2566ecdec /bin/netstat 3aba7c43d7978452e790220b0deb0e4e /bin/pidof 7001afa26625989c85d05be0d4f93e4e /bin/ping d420db19497b56e632756884efd244e9 /bin/ping6 6140d156296de35a86fd154081b00f26 /bin/ps b7ec22f9d3040fff114acfd4f6d226e7 /bin/pwd 72e1a7bbf8478e3dd08693bec6f4c50e /bin/rbash 07e433957de1c39329ebd81d61ca44a2 /bin/readlink bdd022ca8ec797544b3eddb817ce97f5 /bin/rm 34dd0e07f6abdd1531c7c0953752ab1d /bin/rmdir 68de2870b06443403332c81022010a24 /bin/rnano 1622c90a9570641dd182d0eff4e9d95b /bin/run-parts d9be68996d0b87faeb83d1ad8951a481 /bin/sash
Re: /bin/login listening?
On Sun, Jul 29, 2007 at 06:40:05PM +0200, Mathias Brodala wrote: You should also tell the exact version of the login package you are using. Otherwise this number is useless. Sorry. Stock, up-to-date Etch. Aptitude shows it as version 1:4.0.18.1-7. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: essential services? ssh, nfs?
On Sun, Jul 29, 2007 at 04:11:55PM +, Tyler Smith wrote: I'm working through the security quick start how to, and I'm not clear on what services are required and which ones I can safely remove. I'm running a single laptop, which I connect to the net via wireless at home or at cafes, and via an ethernet cable at work. 1) I never login remotely, so I think I can safely do away with openssh-server? If you don't need it, and a package isn't there to meet a dependancy, get rid of it. tcp6*:ssh *:* LISTEN 3026/sshd 2) The how-to suggests that for my setup I don't need anything to do with NFS - netstat reports rpc.statd and portmap as listening. Can I just purge nfs-common and portmap? tcp *:37381 *:* LISTEN 2603/rpc.statd tcp *:sunrpc*:* LISTEN 2578/portmap Ditto. 3) I have apache installed as a dependency of doc-central. netstat shows it to be listening to all interfaces. Is there a way to set it to listen only for local connections? I don't understand this very well, but it seems I shouldn't need to listen to anyone from the outside to connect to my docs. tcp *:www *:* LISTEN 3826/apache I've never run apache so don't know. 4) The only remaining listeners I have are: tcp localhost:929 *:* LISTEN 3721/famd tcp *:auth *:* LISTEN 3661/inetd tcp localhost:smtp *:* LISTEN 3385/exim4 What is auth? Since famd and exim4 are only listening to localhost, can I conclude they are not a security risk? What do you have uncommented in /etc/inetd.conf? I don't have anything, so inetd doesn't start up at boot. Finally, as the last defence, do you have a good firewall setup? I use shorewall with a default net to all DROP and everything else REJECT, then open ports as needed in rules. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
how to set network io priority for a process?
I'm on slow dialup. Downloads of iso's take days. Yet, I still want to be able to browse the internet. I would like to set up something like trickle that will run something but limit its bandwidth so that it lower's its priority. For example, wget and rsync allow one to limit the bandwidth to a set rate. However, if nothing else wants the connection, this wastes available bandwidth. Conversely, if I try to view a big web site, it would like the wget or rsync to get out of the way while I'm doing it. Trickle allows one to set a rate for programs that don't give the option. ionice does something similar for disk io. What I want is something like ionice but for network bandwidth. Any suggestions? Thanks, Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: essential services? ssh, nfs?
On Sun, Jul 29, 2007 at 05:52:12PM +, Tyler Smith wrote: Now that I have nothing listening to the outside world, do I still need a firewall? I have been using firestarter, but not consistently. I look at a firewall as icing on the cake. If you reject and log local to net, anything trying to go out that you haven't configured gives you something to trace. For example, I recently put etch onto my 486 and installed ntp. Between the time I installed it and when I configured it to look to my main box, it came configured to try to access the debian ntp pool. Those packets were rejected but I go an email from my main box warning me about it. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: how to set network io priority for a process?
On Sun, Jul 29, 2007 at 06:14:46PM -, Paul Johnson wrote: On Jul 29, 11:10 am, Douglas Allan Tutty [EMAIL PROTECTED] wrote: I'm on slow dialup. Downloads of iso's take days. Yet, I still want to be able to browse the internet. I would like to set up something like trickle that will run something but limit its bandwidth so that it lower's its priority. [...] What I want is something like ionice but for network bandwidth. I'm not sure if that's workable, but if you have your own home network, you might look into setting up your own squid cache. Squid is a web proxy that maintains it's own cache of items people on your network have already visited. If you use offline mode, you may have to force a refresh on sites you know have new content (shift-refresh in most browsers), but items you've already downloaded for viewing before will load much more quickly on subsequent visits. I already have polipo cache set up and working. Perhaps this is something else I'll have to write from scratch. I'll have to see how, at any given time, one sees what is using the bandwidth. Thanks, Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: how to set network io priority for a process?
On Sun, Jul 29, 2007 at 03:32:44PM -0400, Celejar wrote: I have issues similar to Doug's, and I have also wondered whether kernel based traffic shaping is what I need. Since we both use shorewall, which has an interface to the kernel's shaping capabilities, I suppose we ought to read shorewall-doc/html/traffic_shaping.htm That's what I'm about to do. I've only recently (within the past hour or two) determined that traffic shaping is what I'm trying to do. Ideally, I would just set ftp and rync to second priority and let everything else be normal. I'll keep the list posted. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: dumb question about Adobe Acrobat....
On Sun, Jul 29, 2007 at 12:32:18PM -0700, Alan Ianson wrote: The next version of Evince, due this fall, will also support form filling. That is such good news.. exactly what I have been hoping to read.. I've been looking for a way to fill in pdf forms for an amd64 box I have here. I have been under the impression that evince and kpdf use xpdf behind the scenes. Does this mean that xpdf will be able to do this too, or are evince (and perhaps kpdf) more functional than xpdf? Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: /bin/login listening?
On Sun, Jul 29, 2007 at 12:44:56PM -0700, Jeff D wrote: On that note, one thing that you might want to consider as part of the hardening process is to install aide or some other file integrity checker. Using something like that greatly helps in detecting and identifying issues such as this. I use samhain. However, since a compromised system can't reliably check for an intrusion, I use it as a check agains JFS. Since JFS doesn't journal data (just meta-data), it is possible that after a power failure, a file may be missing. Samhain would detect this. For security, you should have the samhain on a live-CD or something with the checksums stored on a CD or USB stick. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: curses-interface ftp client with resume?
On Sun, Jul 29, 2007 at 10:51:14PM +0300, Andrei Popescu wrote: On Sat, Jul 28, 2007 at 09:44:28PM -0400, Douglas Allan Tutty wrote: I'm on dialup and often access the internet via a slow computer by sshing into my fast computer (which has the modem). Right now, if I want to download something like an iso file via ftp (there being no rsync mirror available), I put the url in a file, e.g. No torrents either? No. For my current download, I found a mirror that also offers rsync (although it doesn't advertise it). However, lots of things are available via FTP that aren't available any other way. 3. It would be nice to have a curses interface like mc that lets me browse to the correct file, then tag the file for downloading which puts it into the above queue. AFAIK mc can do ftp. I didn't use it much, but it might do most of what you need. mc can't resume; it only wants to overwrite, append, or cancel. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: IM on a home debian network
On Sun, Jul 29, 2007 at 10:54:34PM +0300, Andrei Popescu wrote: On Sat, Jul 28, 2007 at 11:30:05PM -0400, Douglas Allan Tutty wrote: and leave this VC 1 open. VC 2 is ssh'd to the box and either mutt or $watch -n 20 from -c VC 3 is for net-related curses stuff (currently a wget session). VC 4 is where I type startx when I need to. There was a time when I had Ctrl-Alt-UP set in inittab to open a new vt. Never heard of 'screen'? I have heard of it, but it relies, like so many other gnu tools, on the operator (me) having either a good memory or a cheat-sheet for keystrokes. I have a poor memory (learning disability) and a cheat-sheet makes using it very slow. Whereas, Ctrl-Alt-UP is simple. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: how to set network io priority for a process?\
On Sun, Jul 29, 2007 at 01:54:36PM -0700, Mike Bird wrote: Traffic shaping usually applies to output. Policing[0] usually applies to input. Since we often can't shape on the router transmitting data to us, in such cases we instead have to police on the receiving end and rely on the sender's TCP flow control to reduce rates to the point where packets aren't lost. This doesn't work for UDP and ICMP and works poorly for varying loads. Right, but downloads happen by blocks. Couldn't something sit somewhere on the data stream and delay the requests for the net ftp or rsync block if there's an active http transfer in progress? In other words, controll the download bandwidth by throttling the rate of upbound requests. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Unable to handle Kernel paging request
On Sun, Jul 29, 2007 at 03:56:26PM -0500, Brad B wrote: I recently did the network installation of Debian to a spare HD, and tried running it by itself in my PC, which usually runs windows. It boots into grub, but I get serveral different error messages at different times. I'm never able to load the kernel, I believe. Here're the most common errors: Unable to handle kernal paging request Kernal Panic -- Not Syncing Are you saying that you installed with the drive in one computer but are trying to boot it in another computer? My guess is that the two different computers would need different initramfs. Since I've never run into this, I've never had to fitz with initramfs; just a pointer. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Lost /home partition
On Mon, Jul 30, 2007 at 02:05:38AM +0300, Dotan Cohen wrote: In a Debian-based distro (Ubuntu) I have had a corruption of my /home partition, which resides on sda4 of my Dell Inspiron laptop. Not knowing what to do, and with no way to boot and google the situation, I played Y, Y, Y to all fsck's questions. Now, /home is empty. I do have a backup from 4 weeks ago, as I backup the first of every month, but I have done quite a bit of work this past month. I'm very interested in recovering the data. A corrupted /home should not keep you from booting. You may need to go single-user or init=/bin/sh but it should boot. Probably should have backed up more recently. It sounds like you made things worse with the YYY. Good luck. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Lost /home partition
On Mon, Jul 30, 2007 at 02:34:55AM +0300, Dotan Cohen wrote: On 30/07/07, Douglas Allan Tutty [EMAIL PROTECTED] wrote: On Mon, Jul 30, 2007 at 02:05:38AM +0300, Dotan Cohen wrote: In a Debian-based distro (Ubuntu) I have had a corruption of my /home partition, which resides on sda4 of my Dell Inspiron laptop. Not knowing what to do, and with no way to boot and google the situation, I played Y, Y, Y to all fsck's questions. Now, /home is empty. I do have a backup from 4 weeks ago, as I backup the first of every month, but I have done quite a bit of work this past month. I'm very interested in recovering the data. A corrupted /home should not keep you from booting. You may need to go single-user or init=/bin/sh but it should boot. Maybe something additional was corrupted, bu I only remember seeing references to sda4, which is /home. Any idea how to get the data back? Undeletion in *NIX is either very difficult, expensive, or impossible. Unless you got lucky and they ended up in lost+found only slightly mangled. If you want to try recovery, unmount sda4 and remove it from fstab. With it mounted, things change. Then aptitude search ~drecover and look at some tools. Try something like foremost or magicrescue. Read the documentation, follow the instructions, and only mount the partition again if it says to. Often such tools work by reading the block device itself, bypassing the filesystem. Doug. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]