Re: make-kpkg doesn't work properly when compiling i386 kernel on amd64 host
In article you write: >Hello, I am trying to build a kernel for my i386 PC on a amd64 server. > >I follow a old instrusion at >http://lists.debian.org/debian-user/2008/07/msg00926.html . It can build >the kernel itself correctly, but the problem is script tools such as >genksyms are build as amd64 format. >I notice that Linux Makefile mentions an environment variable named HOSTCC. >I tried to set the variable, but it still failed because of linker issue. > >Can make-kpkg do this job correctly? What extra parameter I should add? Yes, you can do this. 1. add "ARCH=i386" to "make config" and similar commands, for example: cd linux- cp .../where/ever/your/dotconfig/file/is/dotconfig .config make oldconfig ARCH=i386 2. Run make-kpkg with an i386 specific environment and flags, like: CONCURRENCY_LEVEL=3 DEB_HOST_ARCH=i386 make-kpkg \ --cross-compile - --arch i386 --rootcmd fakeroot \ --revision --append-to-version \ --initrd kernel_image The important thing is to set DEB_HOST_ARCH=i386 in the environment of make-kpkg and add the flags "--cross-compile - --arch i386" Mike. -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/201304172256.r3hmuwnn019...@xs8.xs4all.nl
Re: Dosemu
In article <[EMAIL PROTECTED]>, Ken Heard <[EMAIL PROTECTED]> wrote: >Is there anyone on this list who uses dosemu? I had it working on Sarge >ever since Sarge came out, but I am having trouble getting it to work on >Etch. If your PC is fast enough, why not use dosbox instead. It uses cpu-emulation instead of vm86 mode, so it's a bit slower, but it's a lot easier to get going (dos is built in, and it can access the linux file system). Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: OT: Comparison of filesystems
In article <[EMAIL PROTECTED]>, Ron Johnson <[EMAIL PROTECTED]> wrote: >On Mon, 2006-04-24 at 20:54 -0500, Sumo Wrestler (or just ate too much) >wrote: >> Ron Johnson wrote: >> > [...] >> > Besides, you can't "wipe" files on a journaling fs. So, you re- >> > mount your ext3 partition as ext2, wipe the file(s) and then re- >> > mount as ext3. >> > [...] >> >> Huh? >> >> Are you suggesting that you can't permanently delete a file's data by >> overwriting the file before deleting it? > >Not in any journaling fs. Not in any fs that journals that _data_, but ext3 doesn't do that by default. It only journals meta-data, and you can overwrite files just fine. If you do use ext3 in data-journalling mode (mount -o data=journal) that yes, it will be a lot slower than ext2. That's why the default is data=ordered. You can make it even faster, and still more safe than ext2, by using data=writeback. See "man mount". Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: No shell, Only FTP access
In article <[EMAIL PROTECTED]>, Ron Johnson <[EMAIL PROTECTED]> wrote: >On Thu, 2006-04-06 at 21:56 -0400, Roberto C. Sanchez wrote: >> Ron Johnson wrote: >> > On Thu, 2006-04-06 at 23:30 +0200, Øyvind Lode wrote: >> > [snip] >> > >> >>And yes I considering switching to proftpd but havn't got the time yet >> >>and I also need to setup my own DNS service using BIND... >> > >> > >> > Isn't BIND kinda insecure? >> > >> >> In what way? > >Old, crufty, written before security was a concern. That's why bind9 is a rewrite from scratch. See http://www.linuxsecurity.com/content/view/117551/49/ On the other hand, even in bind8 there hasn't been a serious (remote execution) security hole reported in the last 3 years. Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: amd64: why is sizeof(int) =4? why not =8?
In article <[EMAIL PROTECTED]>, Ron Johnson <[EMAIL PROTECTED]> wrote: >On Wed, 2006-04-05 at 06:48 -0500, Mitchell Laks wrote: >> Hi, >> >> I wanted to find out more about amd64, so I installed the debian distro > >Did you install the 64 bit version? > >> (etch/sid) on my amd64 dual core system. > > >Which one? Etch or Sid? > >> I wanted to see if it is really 64 bit. So naively i compiled the following >> example program I found on the internet: >> >[snip] >> Size of double is 8 >> >> So how do I see the 64 bittness of the operating system if not this way >> with >> gcc? I would think tha size of int would be 8? >> uname -a >> Linux Rashi 2.6.16.1-meshulum-2006-4-5 #1 SMP Wed Apr 5 13:34:46 EDT 2006 >> x86_64 GNU/Linux > >If you did actually install the 64 bit build, do a sizeof(int *) >and you'll see that it is 64 bits. > >AMD decided that, to help with compatibility with s/w written when >sloppy programmers assumed that sizeof(int) == sizeof(*), integers >will be 32 bits, even in 64 bit mode. Hmm? That's a decision of the GCC guys. They decided to make int 32 bits. Remember it's just a C data type. It makes sense because this way, you have a standard C data type for every size of integer: char8 bit short 16 bit int 32 bit long64 bit And this does not help compatibility with s/w written by sloppy programmers who assime that sizeof(int) == sizeof(*), in fact it breaks that compatibility, because sizeof(*) == 8. You can generally assume that sizeof(long) == sizeof(*) on most Linux platforms nowadays though. That will probably break on future 128-bit systems. >64 bit integers are of type "long long" and int64, which are just >different names for the same type. On most current Linux platforms, yes, but on 128 bit platforms, long long will likely be 128 bits. And on amd64 with gcc, "long" is also 64 bits. If you need exactly 64 bits, int64 is your best bet. If you need at least 64 bits, long long will probably work fine. Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Cleaning /tmp regularly using bootclean.sh + crond
In article <[EMAIL PROTECTED]>, Mike McCarty <[EMAIL PROTECTED]> wrote: >Josep Serrano wrote: >>>On Friday 17 February 2006 07:29, Josep Serrano wrote: >>> Do you guys clean regularly your /tmp ? >>> >>>Yup. The tmpreaper package is great for that. >>> >> >> >> Yes, tmpreaper does the job. But here I was preteding to simply reuse the >> bootclean.sh by simply making a symlink. Why installing an addittional >package if >> you already have the stuff to the the job? >> >> Maybe I am wrong and it is a bad idea using bootclean.sh in crond ? > >My understanding (I am not a Linux guru) is that it is risky >to remove things from /tmp while the system is "up", as >some apps, for example X Window system, store stuff in there >which they don't need to survive a boot, but do need to keep >clickin' > >So using a script may not be a Good Thing. That's true. Bootclean.sh is just a step in the bootup sequence, it wasn't designed to be called manually or by cron. Tmpreaper probably does 'the right thing' Mike. -- Freedom is no longer a problem. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: GMail like offline email client
In article <[EMAIL PROTECTED]>, Ð ÐžÑ Ð°Ð»ÐžÐ¹ Ð Ñ ÐµÐœÐºÐŸ <[EMAIL PROTECTED]> wrote: >But it stores all threads like plain lists, not as treeview and that's >uncomfortable for me, or i'm mistaken an evo has the abiliti to handle >mailing-list and display them as trees? View -> Group by Threads (or Control-T) Mike. -- Freedom is no longer a problem. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Most directories locked read-only: how to unlock them?
In article <[EMAIL PROTECTED]>, Ken Heard <[EMAIL PROTECTED]> wrote: >Daniel B. wrote: > >> That looks like you mistyped "defaults" as "drfaults" in your >> /etc/fstab file. > > I was well aware that I mistyped "defaults" as my original post >indicated. I was also aware that this error caused the root directory >to be mounted read-only. Also as my original post indicated, my first >query was how to change the mount to read-write without having to >reinstall the whole operating system. Simply specify both the directory _and_ the device on the command line, in that case mount doesn't read fstab. Eg something like: mount -n -o remount,rw /dev/sda1 / Mike. -- Freedom is no longer a problem. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Question: Maximum number of directories? OT question...
In article <[EMAIL PROTECTED]>, Paul E Condon <[EMAIL PROTECTED]> wrote: >On Tue, Dec 06, 2005 at 11:46:14PM +0000, Miquel van Smoorenburg wrote: >> In article <[EMAIL PROTECTED]>, >> Gerorge Reece-Howe <[EMAIL PROTECTED]> wrote: >> >> "Jochen Schulz" <[EMAIL PROTECTED]> >> > >> >>You can see that / and /srv are ext3 and /var and /home are xfs. I >> >>chose xfs for these because they contain directories with a lot of >> >>files (the already mentioned Maildirs and a news spool). So far, I had >> >>no problems with xfs. >> > >> >I'll see how I go with ext3 for now, but I'll keep that in mind. >> >> Ext2 and ext3 get really slow when you have lots of files (or >> subdirectories) in one directory. >> >> However that was fixed some time ago. If you're running 2.6 you >> can enable the dir_index on an ext3 filesystem. See man mkfs.ext3: >> >>-O feature[,...] >> Create filesystem with given features (filesystem options) >> >>dir_index >> Use hashed b-trees to speed up lookups in >> large >> directories. >> >> >> You can enable this on existing filesystems using tune2fs. See man >> tune2fs(8) for more information. >> > >I've looked at the man page. I'm interested in trying this, but I have a >question: The -O feature... option allows one to set and clear file system >features, but how can I query the system to discover which features are >set on which volumes? $ man tune2fs -l List the contents of the filesystem superblock. # tune2fs -l /dev/sda1 tune2fs 1.37 (21-Mar-2005) Filesystem volume name: / Last mounted on: Filesystem UUID: 71c1a321-ec0e-45c1-967a-59529cc8def0 Filesystem magic number: 0xEF53 Filesystem revision #:1 (dynamic) Filesystem features: has_journal filetype needs_recovery sparse_super Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Question: Maximum number of directories?
In article <[EMAIL PROTECTED]>, Gerorge Reece-Howe <[EMAIL PROTECTED]> wrote: >> "Jochen Schulz" <[EMAIL PROTECTED]> > >>You can see that / and /srv are ext3 and /var and /home are xfs. I >>chose xfs for these because they contain directories with a lot of >>files (the already mentioned Maildirs and a news spool). So far, I had >>no problems with xfs. > >I'll see how I go with ext3 for now, but I'll keep that in mind. Ext2 and ext3 get really slow when you have lots of files (or subdirectories) in one directory. However that was fixed some time ago. If you're running 2.6 you can enable the dir_index on an ext3 filesystem. See man mkfs.ext3: -O feature[,...] Create filesystem with given features (filesystem options) dir_index Use hashed b-trees to speed up lookups in large directories. You can enable this on existing filesystems using tune2fs. See man tune2fs(8) for more information. Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: [OT] SATA vs. SCSI
In article <[EMAIL PROTECTED]>, Gene Heskett <[EMAIL PROTECTED]> wrote: >On Thursday 27 October 2005 20:59, Allan Wind wrote: >>On 2005-10-27T19:33:22-0400, Gene Heskett wrote: >>> Our first foray into using a scsi based commercial server resulted in >>> its getting converted to ata disks fairly rapidly as the scsi raid >>> lost a drive at 2 week intervals. A single big atapi/eide drive >>> turned out to be faster, and a heck of a lot more reliable. >> >>While I feel for you, it's not a good idea to make decisions based on a >>single installation. If you have disks dying every 2 weeks, something >>else was up. Heat would be my first guess. Seagate, I think, had a >>batch of bad SCSI drives recently. > >This was back when scsi drives were 8GB max, so quite a bit of water has >passed by now. If I've learned one thing about disks in the last few years, it's that you should never ever buy the largest disks available. I remember when 2 GB SCSI was common and 4 GB was brand-new, we bought 4 GB disks. Lots of problems with heat, reliability .. replaced them with 2 2 GB disks each, problem gone. Seagate, BTW. I've seen the same pattern over the years with both SCSI and IDE disks. Just buy disks half the size of the current maximum capacity. Those models have the kinks worked out. One other thing is that lots of people don't get SCSI termination right. And that can cause lots of trouble that you don't have with ATA/SATA, including things like (apparently, but not really) failing disks. Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: minicom display problems
In article <[EMAIL PROTECTED]>, Marty <[EMAIL PROTECTED]> wrote: >I mercifully forgot everything I knew about terminals in the '80s, >But now I am puzzled by two purely theoretical questions: > >1) why don't LINE and COLUMN get listed when I run "printenv"? Because the resize command prints out the commands you need to execute; it doesn't actually do anything else. This is because a child process cannot set the environment variables of its parent (or any other process). If you run `resize` (note the backticks) the shell will take the standard output of the command and execute it. Oh. I see that resize actually sets the lines/columns property of the tty using the TIOCSWINSZ (I used strace to see that) ioctl. too Which means you don't have to set LINES and COLUMNS. You can just run resize as "resize >/dev/null". You see, in the old days, with fixed-size terminals, there was no way to change the lines/columns of a tty. Then when windowing systems were invented, and xterms, a way to tell the text-application the number of lines and columns was needed. For that, the apps looked at the LINES and COLUMNS enviroment variable which override the lines and columns set in the termcap or terminfo file. Later, a tty property was added for lines/columns. You can see it when you do "stty -a". The ioctls TIOCGWINSZ and TIOCSWINSZ were added to get/set that property from the ttyfd. Now if you resize a tty window, the window manager also sets the lines/columns property of the terminal, and sends the app running in the terminal the SIGWINCH character so that it can resize. So modern apps first look at LINES/COLUMNS, then at lines/columns from the tty properties (TIOCGWINSZ), then at termcap/terminfo. >2) How does "resize" retrieve the TTY properties via minicom, and why >can't the remote shell automatically do likewise? The resize command uses VT100 escape sequences to query the terminal size. Why shouldn't the remote shell do this ? Why doesn't the shell have all of /bin and /usr/bin built-in ? Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: minicom display problems
In article <[EMAIL PROTECTED]>, Marty <[EMAIL PROTECTED]> wrote: >Is there any way to get minicom to properly display the color graphics >of an ncurses app like iptraf? I am using the multi GNOME terminal as my >X terminal, but xterm seems to have the same problems. > >So far I've tried: >-setting the minicom terminal type to ANSI >-setting the remote shell TERM type to xterm (it starts up as VT100) >-turning the minicom color on using the "-c on" option minicom -c on should start minicom in color. If ctrla-z shows the command summary in color, minicom supports color. Then you need the right terminal type on the remote system to tell its apps to use ansi-color. Try TERM=xterm-color or TERM=linux >Another, possibly unrelated issue, is that iptraf and other programs >like top or man which output to the entire terminal screen, only use >a small portion of the available X terminal screen area (about 23 >lines of text) regardless of how large I make the X terminal window. >Man even puts pages of text on different areas of the X terminal and >is nearly useless. My current workaround is to use a tiny X terminal >window with these programs, but ncurses graphics is still a problem. That is because the remote system doesn't know about the terminal size and there is no way for the windowing system to relay that information through minicom to the remote system on the other side of the serial connection. So you have to change that manually. Use "stty rows 40 columns 120" on the remote system, or run "resize" if that command is installed. (See "man stty", "man resize"). [I notice that resize on my Linux box doesn't set the tty properties but prints out COLUMNS= / LINES= .. how 1980's ..] Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Every fifth character blurred in text
In article <[EMAIL PROTECTED]>, Xeno Campanoli <[EMAIL PROTECTED]> wrote: >I'm not that familiar with the kind of words I should use for this video >problem with every sixth character blurred. I wonder if there's another >word I should use for google searches? Can anyone make a suggestion? Do you have an LCD screen? Is the resolution your card is running under the same as the resolution of the screen ? The effect you're describing can happen when e.g. you have a 1024x768 monitor and your graphics card is running in 800x600. Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: I messed up bootmisc.sh - now can't log in!
In article <[EMAIL PROTECTED]>, Bryan Donlan <[EMAIL PROTECTED]> wrote: >On 8/30/05, [EMAIL PROTECTED] <[EMAIL PROTECTED]> wrote: >> >> Can anyone suggest a tutorial on writing what I think are called init >> scripts? > >Take a look at the files in /etc/init.d, just use one of them as a >template, and symlink it in as /etc/rc.d/S99yourservice Well, "one of them" .. there's /etc/init.d/skeleton which is exactly what you'd think it would be... Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: kPDF
In article <[EMAIL PROTECTED]>, Bill Day <[EMAIL PROTECTED]> wrote: >Thanks for that source, put it in, ran apt-get update and then apt-get sintall >acroread when it was done I sintaleld the suggested acroread-plugins as well. >opened up each file continuously of TUX, with out it crashing. It's somehow fairly unknown, but one of PDF viewers I like best is "evince" and it's in sarge. Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: why do PPP connections die with heavy disk usage?
On Thu, 2005-07-21 at 14:04 -0400, Daniel B. wrote: > Miquel van Smoorenburg wrote: > > > In article <[EMAIL PROTECTED]>, > > Daniel B. <[EMAIL PROTECTED]> wrote: > > > >>Does anyone know why Linux can't seem to maintain a diald-dialed serial > >>PPP connection while performing lots of disk I/O, especially on a PIO-mode > >>IDE disk? > > > > > > Because the IDE driver turns off interrupt processing completely > > when issuing commands to / reading data from the drive. With DMA, > > you don't really notice, in PIO mode, you do .. > > If it turns off interrupt processing completely, how would, say, > keyboard and timer interrupts get processed? Wouldn't I be seeing > keystrokes getting dropped occasionally? It blocks them, ofcourse. The hardware still sets the interrupt bit. > > Do you mean that interrupt handling is _suspended_ when writing > to/reading from the drive (that is, other interrupts aren't ignored > entirely; just deferred)? Same difference :) > > > >>Does the kernel not handle serial port interrupts with higher priority > >>than disk interrupts to make sure that serial data doesn't overrun > >>buffers? > > > > > > Because the IDE driver simply turns all interrupts *off* ... > > I would think that even if the kernel IDE code had to disable > interrupts while performing a PIO-mode transfer, as soon as it > re-enabled interrupts (at the end of the IDE operation), pending > interrupts would be handled. > > Does one IDE operation (in PIO mode) take so long that the > serial port buffer overflows before the disk operation is done? Indeed. > Or are pending interrupts not all processed before a subsequent > IDE operation is started? > > > > To change this, see "man hdparm", -u option. > > Well, given that the manual page says: > >-u ... Use this feature with caution: some drive/controller > combinations do not tolerate the increased I/O latencies > possible when this feature is enabled, resulting in massive > filesystem corruption. > > I'm not about to try that, since the reason I'm using PIO mode > in the first place is because I get massive file system corruption > when I use DMA mode with IDE controllers on my motherboard (Asus > A7M266-D; AMD 762(?) chipset). Have you reported this to the kernel IDE maintainers ? > > Would setting one of hdparm's sector-count options to a small value > decrease the length of time interrupts are disabled and reduce the > chances of problems with serial communication? I don't know, I try to avoid broken hardware :) Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: why do PPP connections die with heavy disk usage?
In article <[EMAIL PROTECTED]>, Daniel B. <[EMAIL PROTECTED]> wrote: >Does anyone know why Linux can't seem to maintain a diald-dialed serial >PPP connection while performing lots of disk I/O, especially on a PIO-mode >IDE disk? Because the IDE driver turns off interrupt processing completely when issuing commands to / reading data from the drive. With DMA, you don't really notice, in PIO mode, you do .. >Does the kernel not handle serial port interrupts with higher priority >than disk interrupts to make sure that serial data doesn't overrun >buffers? Because the IDE driver simply turns all interrupts *off* ... To change this, see "man hdparm", -u option. Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: fsck on ext3 drives. Do we need to routinely? how to shut off or change 30 boot/180 day settings or why should I not?
In article <[EMAIL PROTECTED]>, Mitchell Laks <[EMAIL PROTECTED]> wrote: >Maybe I am really not supposed to reboot my machine? >But how can I stop 180 days from passing? >:) > >1. How do I modify this to 100 boots or 500 days or never? man tune2fs. Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Quad-port Fast Ethernet on Debian
In article <[EMAIL PROTECTED]>, Michelle Konzack <[EMAIL PROTECTED]> wrote: >But using VLAN is much more administaration and risking >mal configuration. Depends. Once you're used to handling VLANs, it makes _so_ many things _that_ much simpler that you cannot imagine ever having lived without it. Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Quad-port Fast Ethernet on Debian
In article <[EMAIL PROTECTED]>, Simon Phillips <[EMAIL PROTECTED]> wrote: >I am after some general advice or info on people's experiences with quad >port ethernet cards. Any assistance would be much appreciated. > >For this, I am specifically looking to use *quad* port ethernet cards, >due to the number of ports and physical size of the hardware required for my >configuration. I will probably need two per box (+ onboard port = 9 ports >total). I don't think GigE quad port ethernet cards exist, so you're talking about 9 100 mbit ports, right ? Why not use a single GigE port ? If you need multiple interfaces, use VLANs. Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: MySQL Performance Woody Vs Sarge
In article <[EMAIL PROTECTED]>, Marty <[EMAIL PROTECTED]> wrote: >Jochen Schulz wrote: >> Simon: >>> Jacob S wrote: >>> >>> # hdparm /dev/hde >>> [...] >>> # hdparm /dev/hdg >> >> Again, I think you are searching in the wrong direction. Your 'hdparm >> -tT' results clearly showed that the great difference between your >> servers doesn't lie in hard disk performance (48 to 43 MB/s), but in >> Memory/CPU performance (278 to 58 MB/s). > >That would be a very gross misconfiguration. To me it seems far more >likely that raid caching is simply disabled, possibly in the driver. There is no such thing as "raid caching". There is something called "write caching", but that doesn't have anything to do with reads. I also think that the hdparm -tT shows that there is something wrong with the board/ram/cpu/bios. Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Move from i386 to ia64?
In article <[EMAIL PROTECTED]>, Paolo Alexis Falcone <[EMAIL PROTECTED]> wrote: >On 5/26/05, Chris Boot <[EMAIL PROTECTED]> wrote: >> Can I upgrade the kernel without upgrading the userspace apps? > >If you'd run in 64 bit mode - No. In fact, you can. I built a 64 bits kernel on my amd64 workstation, then installed that kernel (dpkg -i --force-architecture linux.deb) on a 32 bits sarge installation. Ofcourse that box does have an em64t capable processor. It actually boots, and works fine. You then have a 32 bits userland with a 64 bits kernel. Advantages are that you can use 4 GB of memory per process (no more 3/1 GB split), and you can better take advantage of a lot of memory - no more low memory, high memory, PAE, etc. Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Use of dual CPU?
In article <[EMAIL PROTECTED]>, Marco Calviani <[EMAIL PROTECTED]> wrote: >Adam Mercer ha scritto: > >>On 23/05/05, Marco Calviani <[EMAIL PROTECTED]> wrote: >> >> >>>So, normally, even if i use an SMP kernel if i'm not telling the program >>>to use both CPU it will continue using only one CPU, right? >>> >>> >> >>Yes, if the program isn't written to use both CPUs, it won't. >> >>Cheers >> >>Adam >> >> >Do you know if it is possible to instruct the g77 (fortran) compiler to >use both CPU's, or maybe where to ask for? You cannot run one program on two CPUs. It's simply not possible. What you can do is split your program up into multiple programs or threads, that you can run simultaneously - in that case, you will use more than one CPU. But you need to redesign/rewrite your programs, ofcourse. Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: How to enable a user access to fsck at boot up.
In article <[EMAIL PROTECTED]>, Ivan Teliatnikov <[EMAIL PROTECTED]> wrote: >Good day. > >When machine reboots and there is problem with file system one has an >option to enter root password and user fsck utility to check and fix >file system errors. > >I think this is done via sulogin utility and /etc/init.d/fscheck.sh >script. Actually, via /etc/init.d/checkroot.sh and /etc/init.d/checkfs.sh. Those check the root filesystem and all other filesystems, and start a shell through sulogin if something went wrong. >On some machines I have trusted users who can su via sudo, but I am >reluctant to give away root password. Is it possible to allow trusted >users to run fsck without knowing root password. No, that is not possible. Why not set FSCKFIX to y on those machines in /etc/default/rcS (see "man rcS"). If you do that fsck will be able to repair a disk without manual intervention in most cases. By the way, what are you doing with those machines? Running ext2 filesystems and then turning off the power during heavy writes? Consider upgrading to a journalling filesystem like ext3. Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: sarge apt and init daemons
In article <[EMAIL PROTECTED]>, Phil Dyer <[EMAIL PROTECTED]> wrote: >-BEGIN PGP SIGNED MESSAGE----- >Miquel van Smoorenburg said: > >>>> If you want to keep updates from starting the daemon, just chmod 644 it. >>> >>>That sounds reasonable...and simple. :) thanks. >> >> Reasonable, simple, and wrong :) >> >> As long as one start or stop link is still present, update-rc.d will >> not change the configuration. So just remove all start/stop links >> except in, say, runlevel 5 or so. Since the default runlevel is >> 2, the service will not be started ever. >> > >OK, although both solutions work, (I guess - I haven't tried the second >solution) it still seems kludgy to me. If I use the debian supplied tool >to remove a service from startup _totally_, and I use a debian supplied >tool to update the system, shouldn't the latter honor my current config? Nope. Because that is not how it works or has ever worked. Your expectation is skewed from reality (sorry). >I don't think I should have to remeber to chmod this or fake out that. I >have no problems doing either, it just seems odd that the best system >update tool available can check to see if my conf files have been >modified and let me choose whether to overwrite, but it doesn't check to >see how I've configured services. It does. Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: sarge apt and init daemons
In article <[EMAIL PROTECTED]>, Phil Dyer <[EMAIL PROTECTED]> wrote: >-BEGIN PGP SIGNED MESSAGE- >Hash: SHA1 > >David Clymer wrote: >> >> The debian post install script probably doest go through the rc.* >> directories looking for runlevel entries since these are all just >> symlinks to a script in /etc/init.d/. All that it probably checks for is >> that the /etc/init.d/ script exists and is executable. Then, >> on the assumption that its generally a good idea to stop and start >> daemons when upgrading them, it does so. > >I just think it would be a good idea to not start something that's not >running anyway. Check to see if the proc is running and if it is, restart. In a lot of cases, the service is stopped in the preinst script, and (re-)started in the postinst script. So your suggestion doesn't work. >> If you want to keep updates from starting the daemon, just chmod 644 it. > >That sounds reasonable...and simple. :) thanks. Reasonable, simple, and wrong :) As long as one start or stop link is still present, update-rc.d will not change the configuration. So just remove all start/stop links except in, say, runlevel 5 or so. Since the default runlevel is 2, the service will not be started ever. Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: kernel2.6 I cannot login
In article <[EMAIL PROTECTED]>, Massa Takeuti <[EMAIL PROTECTED]> wrote: >My keyboard is a PS/2 American style 104 keyboard. >My mouce is a PS/2 wheel mouce. >I can use them for 2.4 kernel without any trouble. > >I show you my kernel configuration. Hmm, looks like the keyboard/mouse driver is not compiled in. Look: ># ># Input I/O drivers ># ># CONFIG_GAMEPORT is not set >CONFIG_SOUND_GAMEPORT=y >CONFIG_SERIO=y ># CONFIG_SERIO_I8042 is not set The i8042 is the keyboard/mouse UART. You need to enable that. In fact it is enabled by default on a PC. Did you simply disable _all_ serial drivers ? Not a good idea :) Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Sarge box not rebooting..
On 2004.12.29 16:20, Robert Waldner wrote:
On Tue, 28 Dec 2004 18:30:27 +0100, Robert Waldner writes:
>>No, kill(-1, SIGWHATEVER) is guaranteed to kill all processes
>>/except/ the caller. "man 2 kill" on any unix/linux box. What kernel
>>are you using, this might be a kernel bug. Is this an i386 or
>>another architecture ?
>2.4.27, from kernel-image-2.4.27-1-386, i386 arch, straight Sarge
> from d-i RC2, `apt-get upgrade` up-to-date as of now.
>I've put more info (`dpkg -l`, `ps auxwww`, cpuinfo, meminfo, lsmod) at
> http://www.waldner.priv.at/temp/machine.txt (it'd make for one long
> email otherwise).
I've now tested through a couple kernel-images, and found that the
problem does NOT manifest itself when it's an SMP-kernel, eg
2.4.26-1-686-smp and 2.4.27-1-686-smp are fine, but the default
2.4.26/7-1-386 and their respective -686 siblings aren't.
What remains is the question if I should file this as a (grave?) bug
against kernel-image-2.4.27-1-386.
And, for people googling this up later: DO NOT run the standard Sarge
kernel 2.4.27-1-386 on HP/Compaq DL380 machines, you won't be able to
reboot/shutdown them.
Well, it would be perfect if you could reproduce this. Try something like:
#include
#include
void sigcld(int sig)
{
printf("Caught SIGCLD\n");
}
int main()
{
signal(SIGCLD, sigcld);
kill(-1, SIGCLD);
sleep(2);
return 0;
}
If this prints "Caught SIGCLD" then that is a severe kernel bug.
Try both "cc foo.c" and "cc foo.c -lpthread" please.
Oh and what is the output of "ldd /sbin/killall5" ?
Mike.
Re: Sarge box not rebooting..
On 2004.12.28 17:42, Robert Waldner wrote:
On Tue, 28 Dec 2004 08:40:48 +0100, Robert Waldner writes:
>>>Hmm, can it be that killall5 doesn't actually manage to *not* kill
>>> itself?
>>Ofcourse it goes through great lengths to do exactly that - NOT
>>kill itself. It kills all processes _except_ itself and its
>>caller.
>Any hints on what it _could_ be, or on what I can do to further narrow
> down the problem?
Well, I expanded killall5.c with a couple printf's:
...
int main(int argc, char **argv)
{
...
signal(SIGTERM, SIG_IGN);
signal(SIGSTOP, SIG_IGN);
signal(SIGKILL, SIG_IGN);
/* Now stop all processes. */
// changes rw
printf("now doing kill(-1, SIGSTOP);\n");
kill(-1, SIGSTOP);
sent_sigstop = 1;
printf("done with kill(-1, SIGSTOP);\n");
...
and the last thing I see on the console is the first printf.
Screenshot (thanks to iLO) at
http://www.waldner.priv.at/temp/killall5.jpg
So to me it seems like "signal(SIGSTOP, SIG_IGN);" either isn't
honored, and killall5 itself killed, or else it kills something else
essential, but what could that be?
No, kill(-1, SIGWHATEVER) is guaranteed to kill all processes
/except/ the caller. "man 2 kill" on any unix/linux box. What kernel
are you using, this might be a kernel bug. Is this an i386 or
another architecture ?
(You're not running bootlogd somehow at shutdown time are you ?)
Plus, I've discovered 3 other boxen, various DL360/380, with the same
problem. Isn't there anyone else with Compaq/HP gear and this problem?
I doubt it is compaq specific, but there must be something else
out of the ordinary here or everybody would have this problem.
Mike.
Re: Sarge box not rebooting..
In article <[EMAIL PROTECTED]>, Robert Waldner <[EMAIL PROTECTED]> wrote: >On Mon, 27 Dec 2004 12:25:08 +0100, John Smith writes: >>On Mon, 2004-12-27 at 10:48 +0100, Robert Waldner wrote: >>> I have a HP DL380 here with Sarge (current as of now) on it. Problem > >> It must be one of the most mentioned boxes on these lists, a >>distinction I wouldn't crave, as it's manufacturer... (Hey, HP, are you >>listening? You are using Debian inhouse, why don't you contribute a bit >>more, we are convinced of the hardware quality!) > >Tell me about it :( - unfortunately I've no control whatsoever about > the choice of hardware. > >>As it's not even displaying "Rebooting", did you check >>the /etc/init.d/reboot permissions? How about calling it directly with >>a sh -x? > >It doesn't come as far. I've checked further, and what's holding it up > is /etc/rc6.d/S20sendsigs, the `killall5 -15`. I strace'd it, and the > last thing I get is "rt_sigaction(-1, SIGSTOP" (note the missing ")"). Ofcourse, by then the strace process is sigSTOPped too. Heisenbug. > If I background both killall5's, it comes as far as "Saving random > seed... done", eg S30urandom finishes. > >Hmm, can it be that killall5 doesn't actually manage to *not* kill > itself? Ofcourse it goes through great lengths to do exactly that - NOT kill itself. It kills all processes _except_ itself and its caller. >This would be a quite grave bug. Ehm, no. Wrong conclusion. Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: crontab every 5 minutes?
In article <[EMAIL PROTECTED]>, Joao Clemente <[EMAIL PROTECTED]> wrote: >Now, I must wonder: Is there a way to "shift" this by a wanted value? >(for instance, when x MOD y = 1, something like putting into cron the >instance (*/5)+1 ? It would run at 1,6,11,16,21, ... Is it possible? Perhaps 1-59/5 ? From the manpage I get the impression that that would work. Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: howto know if raid is really working?
In article <[EMAIL PROTECTED]>, Harland Christofferson <[EMAIL PROTECTED]> wrote: >At Thursday, 16 December 2004, you wrote: > >>Thanks for the reply, but I still haven't cleared my doubt. >>I'll cut the relevant part of the reply (relevant for what I'm trying >to >>understand) and comment at the end: >> >*snip* > >you mention mirroring partions. i have the entire disk mirrored, >not the individual partions. is it better to mirror individual partions? The current read-balancing code in RAID1 works better on whole-disk raid, since different MD devices don't know anything about eachother. Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: run levels 1 and S
In article <[EMAIL PROTECTED]>, martin f krafft <[EMAIL PROTECTED]> wrote: >Are there any differences between the run-levels 1 and S? I know >that S10single is not executed for S, but init spawns sulogin >directly. Any functional differences? Runlevel S doesn't have start/stop scripts. Runlevel 1 is used to shutdown all current running processes. Runlevel 1's last start action is to switch to runlevel S. So "init S" is _very different_ from "init 1", since the first will keep all current processes running and just spawn a shell on the console, while the latter will terminate all running processes, _then_ switch to runlevel S. What about /etc/rcS.d/ you say? Well, the stuff in /etc/rcS.d is run at "sysinit" time (which happens just _once_ after boot), not in single user mode. The 'S' stand for sysinit, not runlevel S. See "man init", "man shutdown", and "man inittab". Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: M$ Frontpage Replacement ?
In article <[EMAIL PROTECTED]>, Nayyar Ahmed <[EMAIL PROTECTED]> wrote: >Hello All, > >Is there any as good as M$ Frontpage and flexible >website development tool, which could give all >those luxarires present in Frontpage. Nvu, http://nvu.com/ ? Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: LC_LANG and Init
In article <[EMAIL PROTECTED]>, Christian Schaefer <[EMAIL PROTECTED]> wrote: >Hi, > >how can I set environment variables for processes listed in >/etc/inittab? Especially LC_LANG which is needed for correct running >(german umlauts) of a backup client. aa:2:respawn:/bin/sh 'LC_LANG=de_whatever exec /bitte/bitte/ja' Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: OT? hypertread on or off for SMP kernel
In article <[EMAIL PROTECTED]>, michael <[EMAIL PROTECTED]> wrote: >However, for one simple test (v.low memory, v.low comms, high compute) >case which is Fortran90 compiled using mpich, I actually found turning >HT off helped -- when it was on then for -np 2 there was no speed up >(top implied they ran on same physical processor). I presume this was >because the CPU was fully busy anyhow so the 2nd thread had to wait (you >seem to imply the opposite but I don't follow why!). Turning on the "SMT scheduler support" in the kernel is supposed to prevent exactly this. If two jobs are runnable, they will be run on 2 different CPUs (if you have them), not on 1 CPU with 2 threads. Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: login into a particular directory using ssh
In article <[EMAIL PROTECTED]>, Kamaraju Kusumanchi <[EMAIL PROTECTED]> wrote: >On Sun, 05 Dec 2004 17:18:54 -0500, David Mandelberg ><[EMAIL PROTECTED]> wrote: >> ssh [EMAIL PROTECTED] "sh -c 'cd dirname; exec somecommand'" >> >> somecommand is the command you want to run after cd'ing. E.g. if you >> want to run bash in / as joe at shell.example.net: >> >> ssh [EMAIL PROTECTED] "sh -c 'cd /; exec bash'" >> > >I tried this. It is not working. The machine seems to be hanging. Is >there any typo there? Try this: ssh -t [EMAIL PROTECTED] 'cd / && exec /bin/bash' The '-t' forces allocation of a tty, which is probably what confused you wrt "it seems to be hanging" - you didn't get a prompt. The "sh -c" isn't neccesary as ssh already executes a shell, and the '&&' makes sure that the final shell is only run if the cd / (or cd /what/ever) succeeds. Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: .ini files in bash?
In article <[EMAIL PROTECTED]>, David Baron <[EMAIL PROTECTED]> wrote: >How might one save and restore setting of variables in a bash script to and >from a file in one's home (or sub-) directory? Okay so everybody's trying to outdo eachother trying to write sed scripts that do proper quoting. I'd like to recommend reading the bash or any other POSIX shell manual instead. To save variables: export -p > envfile To read them back: . envfile See, for example, "man ash": export -p With no arguments the export command lists the names of all exported variables. With the -p option specified the output will be formatted suitably for non-interactive use. One minor nit though: the output of 'export -p' is not guaranteed to be readable for every POSIX shell. Bash and zsh generate "declare -x" statements that can't be read by a standard POSIX shell like ash. Perhaps a bug report is in order ... in the mean time, make sure you only import variables into the same shell you exported them from or use this tiny code fragment I'm posting here trying to outdo all the other sed l33t [EMAIL PROTECTED]: export | sed -e 's/^\(export\|declare -x\|typeset -x\|\)/export /' > envfile That would be reasonably portable. Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Making my USB based "webcam" work
In article <[EMAIL PROTECTED]>, stan <[EMAIL PROTECTED]> wrote: >Just picked up a cheap USB based webcam, and I'm trying to figure out >how to use it with Debian. > >lsusb reports 04fC;0561 Sunplus Technology Co, Ltd. > >When I plug it in it's detectd but not claimed by any driver. > >How do I go about geting this to work? 15 seconds with Google turned up http://spca50x.sourceforge.net/ Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: sysv-rc-conf breaks sysvinit
In article <[EMAIL PROTECTED]>, Miquel van Smoorenburg <[EMAIL PROTECTED]> wrote: >I've filed a bugreport against sysv-rc. I meant sysv-rc-config, ofcourse. Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: sysv-rc-conf breaks sysvinit
In article <[EMAIL PROTECTED]>, Greg Trounson <[EMAIL PROTECTED]> wrote: >I recently tried to apt-get install sysv-rc-conf, since rcconf seems to >have dropped off the repos, and got this: > >Unpacking sysv-rc (from .../sysv-rc_2.86-5_all.deb) ... >dpkg: error processing /var/cache/apt/archives/sysv-rc_2.86-5_all.deb >(--unpack): > trying to overwrite `/etc/init.d/rc', which is also in package sysvinit > >Does sysv-rc replace sysvinit or something? Yes it does, partially. This is a bug in the dependencies of sysv-rc-conf, it should depend on both sysvinit and sysv-rc so that sysvinit gets upgraded properly .. >Is it safe to overwrite >this file with the version distributed by sysv-rc, or is this a >temporary fix? Probably not, you should install sysvinit first, that will make sure sysvinit, sysv-rc and initscripts are installed in the proper order. I've filed a bugreport against sysv-rc. Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: where are debian boot-messages logged?
In article <[EMAIL PROTECTED]>, Bram Mertens <[EMAIL PROTECTED]> wrote: >On my desktop (running SuSE 8.2) the messages printed to tty1 when the >machine is booting are stored in /var/log/boot.msg. On my laptop I >can't find any file in /var/log/ that contains these messages. Suse uses bootlogd for that. If you're running testing/unstable you can edit /etc/default/bootlogd and enable bootlogd. The output will be in /var/log/boot. The reason that this is off by default is that it makes some systems unbootable, esp non-i386 machines with serial consoles. Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: autofs or am-utils?
In article <[EMAIL PROTECTED]>, Ron Johnson <[EMAIL PROTECTED]> wrote: >If I just want my USB pen drive to be automounted & dismounted, >it sounds like autofs would be the one to use, then. Or supermount. That's a seperate kernel patch, though. http://supermount-ng.sourceforge.net/ http://ck.kolivas.org/patches/2.6/2.6.9/2.6.9-ck2/patches/supermount-ng207.diff Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: BIG filesystems, Big Files, Transparent Compression?
In article <[EMAIL PROTECTED]>, Ben Russo <[EMAIL PROTECTED]> wrote: >(see below for long story background ) > >The last time I created a large HW RAID5 volume (1.6 TB) the kernel was >unable to see all of it... If I create several smaller block devices >(like 400GB each) can LVM bind them together into a larger single >filesystem? ( I am aiming for 4-6 TB ) You need to use a 2.6 kernel and compile the kernel with support for large block devices. In "make menuconfig", "device drivers -> block devices -> Support for Large Block Devices". Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Where is the key for the password files stored?
In article <[EMAIL PROTECTED]>, Micha Feigin <[EMAIL PROTECTED]> wrote: >Where is the key used to decrypt the passwords (shadow) files stored? >I want to reinstall an old redhat system with debian, but I wish to >preserve old users and passwords to ease the transition process. You can't decrypt passwords - because they are not crypted at all. The algorithm is just a one-way hash. When you login, the password you enter is simply encrypted (one-way- hashed, really) again and compared with the one in the shadow file. The password in the shadow file is never decrypted as such. Why not simply copy the relevant parts of the password and shadow files over, why do you want to decrypt first ? Mike. -- "In times of universal deceit, telling the truth becomes a revolutionary act." -- George Orwell. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Can't umount / at shutdown
In article <[EMAIL PROTECTED]>, David Baron <[EMAIL PROTECTED]> wrote: >On Friday 15 October 2004 15:03, [EMAIL PROTECTED] >wrote: >>init 1 root 10u FIFO3,6176334 /dev/initctl > >It is this "init" process, apparently. I have had the same problem with recent >2.6.* kernels. Meanwhile, I have changed the umountfs script to use an -l >"lazy" umount. This get's rid of the error messages. The offended filesystem >was fine until the last two kernel upgrades. Now, it recovers the journal >(ext3) on startup but all is then clean. You have /dev on a seperate filesystem ? Mike. -- "In times of universal deceit, telling the truth becomes a revolutionary act." -- George Orwell. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Can't umount / at shutdown
In article <[EMAIL PROTECTED]>, Pau Capdevila <[EMAIL PROTECTED]> wrote: >I've added a probe in the umount rc script to see what's opened just >before trying to umount filesystems. I can see the following > >COMMANDPID USER FD TYPE DEVICESIZE NODE NAME >lsof 2401 root1w REG3,6 33130315 /var/log/reboot.log > >What is preventing the umount? The lsof itself is the only thing that has a file open for writing. Mike. -- "In times of universal deceit, telling the truth becomes a revolutionary act." -- George Orwell. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: NFS slow/lockups
In article <[EMAIL PROTECTED]>, Tim Kelley <[EMAIL PROTECTED]> wrote: >On Wed, Oct 13, 2004 at 09:05:21PM +0100, Alan Chandler wrote: >> Just to show its not networking problems, I have resorted to copying >the files >> via scp - even encrypted I am getting about 0.5Mb/sec > >That's horrible! Is this a 10Mb network or dialup or what? Surely you >mean .5 MBytes/sec (which also isn't too hot for a 10Mbit ethernet). > >> Any idea whats wrong and how to improve performance? > >Some basic things: > >1. If you have a switch, make sure it your interfaces are at full >duplex But do not force the switch interface to full duplex. At least on Cisco's, if you configure the switch at full duplex (as opposed to "auto"), then it will not negotiate full/half duplex anymore. And since almost all Linux ethernet device drivers default to half-duplex, you get a guaranteed duplex/half-duplex mismatch. And .5 Mbytes/sec on a 100 mbit network is what you end up with. Mike. -- "In times of universal deceit, telling the truth becomes a revolutionary act." -- George Orwell. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: change /dev/console owner to user logged on console?
In article <[EMAIL PROTECTED]>, Marc D Ronell <[EMAIL PROTECTED]> wrote: >The following message is a courtesy copy of an article >that has been posted to gmane.linux.debian.user as well. > >>> "Miquel" == Miquel van Smoorenburg <[EMAIL PROTECTED]> writes: > > > In article <[EMAIL PROTECTED]>, > > Marc D Ronell <[EMAIL PROTECTED]> wrote: > >> Under Debian, is there a method which allows any user directly > >> logged onto the host and using the host's keyboard and monitor to > >> take ownership of some that host's devices? > > > pam_console > >So I am using Debian Sarge 3.1, and I believe pam is installed, but I >do not see either a directory or file called pam_console when I search >using an up-to-date locate check: Ah. I've never used it either, but thanks to Google I'be developed a keyword-memory (using Google as external-storage memory) which said "pam_console". It appears that the current pam module that does something similar is called pam_devperm: Package: libpam-devperm Version: 1.5-2 Depends: libc6 (>= 2.3.1-1) Description: PAM module to change device ownership on login This PAM module allows you to change device ownership and permissions when a user logs in. . This is functionally equivalent with SunOS logindevperm(4). Mike. -- "In times of universal deceit, telling the truth becomes a revolutionary act." -- George Orwell. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: change /dev/console owner to user logged on console?
In article <[EMAIL PROTECTED]>, Marc D Ronell <[EMAIL PROTECTED]> wrote: >Under Debian, is there a method which allows any user directly logged >onto the host and using the host's keyboard and monitor to take >ownership of some that host's devices? pam_console Mike. -- "In times of universal deceit, telling the truth becomes a revolutionary act." -- George Orwell. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: More on umount problems on shutdown
In article <[EMAIL PROTECTED]>,
David Baron <[EMAIL PROTECTED]> wrote:
>Several posters, including myself, have cited problems umounting the file
>systems on shutdown with recent 2.6.8 kernels. This may be caused by timing
>problems leaving the file systems busy. I got rid of the error messages and
>got almost all clean file systems on restart by using umount -l ("lazy
>umount") and a sleep to let things settle down.
>
>One of the file systems is ALWAYS recovering its journal on restart (mounts
>to /usr/share). It is found clean and everything is OK, but ...
>
>I tried a sync before umount. Did not help. In the umountfs script, it might
>be too late since all processes have been killed off. The journal service for
>each FS is a process, is it not?
No.
>So I put the sync, followed by a sleep, in
>the script before the killall's. Did not help there either.
>
>Any other ideas?
Yes, try to put a "fuser -vm /usr/share; sleep 10" in the script,
after the killalls. That will perhaps tell you what is keeping
the mountpoint busy.
Mike.
--
"In times of universal deceit, telling the truth becomes
a revolutionary act." -- George Orwell.
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: sysvinit/gnome/display bug report
In article <[EMAIL PROTECTED]>, Arjen Dragt <[EMAIL PROTECTED]> wrote: >I am trying to determine the best category under which >to file a Debian bug report. >The symptoms (completely repeatable) are that trying >to change to runlevel S (via telinit or init) does not >work, insteat causes my screen (console - Gnome >session) to go really wierd.command line starts >scrolling very fast (blank lines), any mouse clicks >initiate multiple windows (of whatever was clicked >on). It looks a lot like a loop gone bad (endless >loop). Well, that's probably because the system is doing what you're TELLING it to do, not what you WANT it to do :) You don't use 'init s' to go to single user mode. You need to go through runlevel 1 so things get shut down properly. So 'init 1' or 'telinit 1' will do what you want. If fact you should probably simply use the standard way to go to single user mode, which is simply 'shutdown now'. Mike. -- "In times of universal deceit, telling the truth becomes a revolutionary act." -- George Orwell. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: running fsck on root
In article <[EMAIL PROTECTED]>, Rob Benton <[EMAIL PROTECTED]> wrote: >I've my drive setup like this: > >/dev/hdb1 /boot dos >/dev/hdb2 / xfs >/dev/hdb3 /usr/local xfs >/dev/hdb5 /u01 xfs >/dev/hdb6 swap > > >I've tried shutting down to single-user mode and then remounting / as ro >but all the xfs disk check utilities complain. Are there any tricks to >getting the xfs utilities to check mounted filesystems? Or maybe >there's a better way to approach this I just haven't thought of? # man xfs_repair -d Repair dangerously. Allow xfs_repair to repair an XFS filesystem mounted read only. This is typically done on a root fileystem from single user mode, immediately followed by a reboot. # mount /dev/hda3 on /mnt type xfs (ro) # xfs_repair /dev/hda3 xfs_repair: /dev/hda3 contains a mounted filesystem fatal error -- couldn't initialize XFS library # xfs_repair -d /dev/hda3 Phase 1 - find and verify superblock... Mike. -- "In times of universal deceit, telling the truth becomes a revolutionary act." -- George Orwell. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: `who -d' Dead processes
In article <[EMAIL PROTECTED]>, Floris Bruynooghe <[EMAIL PROTECTED]> wrote: >I had a computer (old PC) running at my student house over the holidays >and was using it (as I am now) to ssh into and do my normal work etc. >Just by accident I discovered the `who -d' command and saw I had a >couple of 100 processes listed that way. Also they where using up my >pty's. Coz I was scared of running out of pty's I rebooted the system >since then. But I get again lots of dead processes. Are you sure those dead processes actually exist .. who -d doesn't list dead processes. who -d lists empty slots in the /var/run/utmp file. Mike. -- "In times of universal deceit, telling the truth becomes a revolutionary act." -- George Orwell. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: root partition full and du and df do not agree
In article <[EMAIL PROTECTED]>, John Harrold <[EMAIL PROTECTED]> wrote: >I have a delima. My root partition is about 24Gb and it's full. The problem >is that I cannot figure out what is taking up the space. If I goto '/' and >run 'du -x' it says that about 500Mb are being used. This is about what I >would expect. Now If I run 'df' it says the partition is 100% full. Is >there any way to check for unlinked files which are taking up space or >something like that? Sure. Try ls -ld /proc/*/fd/* 2>&1 | fgrep '(deleted)' .. that will show you all deleted files still held open by running processes. Mike. -- "In times of universal deceit, telling the truth becomes a revolutionary act." -- George Orwell. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Orderly dismount at shutdown?
In article <[EMAIL PROTECTED]>, David Baron <[EMAIL PROTECTED]> wrote: >Now that I have my multi-partition file system up and running: One of the >partitions (mounted /usr/share) does not dismount in an orderly manner on >shutdown. I get messages like: > Illegal seek >...hdb6 not mounted ... It's kind of hard to diagnose a problem like this without the exact message.. >File system is OK on restart. (The first time, I needed ext3 journal recovery >but this has not recurred.) You can't unmount a filesystem if it's busy (something still running with files open on that filesystem), or if there's something else mounted on it. Perhaps that's your problem. Mike. -- "In times of universal deceit, telling the truth becomes a revolutionary act." -- George Orwell. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Exim4 return-path header
In article <[EMAIL PROTECTED]>, Eddy <[EMAIL PROTECTED]> wrote: >Hi, > >I use Exim4 with Sarge and have a little configuration problem regarding >mail headers. > >What should I do in order to have "Return-path" and "Sender" headers added >by Exim to be exactly the same as the "From" header created by my mail >client ? Make sure your mail client calls exim with the correct options to set this header. I'm not sure if kmail can do that, the right incantation for mutt is 'set envelope_from'. Alternatively, let exim listen on port 25 on localhost, and tell kmail to use SMTP server localhost:25 to deliver mail. Probably kmail will use the same address for MAIL FROM: and From: and that will solve your problem nicely. Mike. -- The question is, what is a "manamanap". The question is, who cares ? -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: /dev/pts/0: Operation not permitted
In article <[EMAIL PROTECTED]>, Johann Spies <[EMAIL PROTECTED]> wrote: >When I do a su - sysadm on one server with Woody I get the error >message in the subject line. That happens only when su to sysadm is >being used. > >Apart from the error message everything seems work normally: > >14:40:38 [EMAIL PROTECTED]:/var/log/exim4$ sudo su - sysadm >/dev/pts/0: Operation not permitted >14:40:42 [EMAIL PROTECTED]:~$ > >What is causing this? Probably a 'mesg n' in /etc/profile or somesuch. After a su, /dev/pts/0 is still owned by the previous userid. Mike. -- The question is, what is a "manamanap". The question is, who cares ? -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Where's routed?
In article <[EMAIL PROTECTED]>, Jon <[EMAIL PROTECTED]> wrote: >Check out the package zebra. Replaces routed and ripd. >http://www.zebra.org/ Zebra has been superseded by quagga, http://www.quagga.net/ Mike. -- The question is, what is a "manamanap". The question is, who cares ? -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: domain name of debian
In article <[EMAIL PROTECTED]>, John Summerfield <[EMAIL PROTECTED]> wrote: >Rus Foster wrote: > >> On Mon, 23 Aug 2004, Shu Hung (Koala) wrote: >> >>> -BEGIN PGP SIGNED MESSAGE- >>> Hash: SHA1 >>> >>> where does the domain name of a debain stored? >>> is there anyway to change it? >> >> >> Try edit /etc/domainname or /etc/hostname depending on what you want >> to acheive >> >> Rus > >dnsdomainname is in /etc/resolv.conf >[EMAIL PROTECTED]:~$ cat /etc/resolv.conf >search demo.room >nameserver 192.168.9.4 >nameserver 192.168.8.1 >[EMAIL PROTECTED]:~$ hostname -f >Dolphin.demo.room >[EMAIL PROTECTED]:~$ Well, only to the point that you need to register the full hostname aka FQDN in either DNS or /etc/hosts as well. >NIS domain name may be in /etc/defaultdomain (to my surprise). More >usually, it's in /etc/yp.conf It's _never_ in /etc/yp.conf - always in /etc/defaultdomain. The name is an inheritance from Sun, they invented NIS and on Solaris the NIS domainname is stored in /etc/defaultdomein. Mike. -- The question is, what is a "manamanap". The question is, who cares ? -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: OT: Which tool, and how, to get partial string from file?
In article <[EMAIL PROTECTED]>, Kent West <[EMAIL PROTECTED]> wrote: >>> On Fri, Aug 20, 2004 at 09:46:52PM -0500, Kent West wrote: How can I test to see if the word "tuber" is in the /etc/passwd file, reliably, and take an action if it is, and take a different action if it's not, in a bash shell script? > >How do I test to make sure the name is in two files? Like so? At this point, I suggest you start studying shell script programming. Get a book, or read one of the many online tutorials such as http://www.tldp.org/LDP/abs/html/ Mike. -- The question is, what is a "manamanap". The question is, who cares ? -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: OT: Which tool, and how, to get partial string from file?
In article <[EMAIL PROTECTED]>, Kent West <[EMAIL PROTECTED]> wrote: >Miquel van Smoorenburg wrote: > >>In article <[EMAIL PROTECTED]>, >>Kent West <[EMAIL PROTECTED]> wrote: >> >> >>>A programmer would know this . . , but not me ;-) >>> >>>I'm using a script to build a file by concatenating portions of two >>>other files. Then end result needs to be checked to make sure a certain >>>word shows up in a line. >>> >>>I know that "grep programm " would return the line: >>> >>>A programmer would know this . . , but not me ;-) >>> >>>but what I want returned is just the word "programmer". >>> >>> >> >>sed -ne 's/^.*\(word_here\).*$/\1/p' < file >> >> >Here's my test script: > >> #!/bin/bash >> >> if [ sed -ne 's/^.*\(icewm\).*$/\1/p' < /home.local/snert/.xinitrc ] ; >> then >> echo "Yep" >> else >> echo "Nope" >> fi That's the complete wrong way to use '[' ! You want something like: WORD=`sed -ne 's/^.*\(icewm\).*$/\1/p' < /home.local/snert/.xinitrc` if [ -n "$WORD" ] then For a primer on how to use '[', do "man test". >>BTW, why not simply >> >> if grep -q word file >> then >> # word was present in file >> bla >> bla >> fi >> >> >> >My test script: > >> #!/bin/bash >> >> if [ `grep -q icewm /home.local/snert/.xinitrc` ] ; then >> echo "Yep" I wrote if grep -q word file Why change it to if [ `grep -q icewm /home.local/snert/.xinitrc` ] ? That means something completely different. "sh" scripts are a programming language. They do what you say, not what you mean ;) After 'if' comes a command. The exit status of the command is used as return value: 0 is true, 1 (or any non-zero value) is false. The '[' thing is just a command. Look: $ ls -l /usr/bin/[ -rwxr-xr-x 1 root root 23928 Jul 16 13:37 /usr/bin/[* It's also known as 'test'. See "man test". But you can put other commands there, like grep. "grep -q" exits with exit-status 0 (true) on a match. So that's why I said if grep -q word file and that's why it's completely different from if [ grep ... Mike. -- The question is, what is a "manamanap". The question is, who cares ? -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: OT: Which tool, and how, to get partial string from file?
In article <[EMAIL PROTECTED]>, Kent West <[EMAIL PROTECTED]> wrote: >A programmer would know this . . , but not me ;-) > >I'm using a script to build a file by concatenating portions of two >other files. Then end result needs to be checked to make sure a certain >word shows up in a line. > >I know that "grep programm " would return the line: > > A programmer would know this . . , but not me ;-) > >but what I want returned is just the word "programmer". sed -ne 's/^.*\(word_here\).*$/\1/p' < file BTW, why not simply if grep -q word file then # word was present in file bla bla fi Or "if ! grep -q word file" if you want to do something if word wasn't present in file. Mike. -- The question is, what is a "manamanap". The question is, who cares ? -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Sarge RAID install
In article <[EMAIL PROTECTED]>, Paul Gear <[EMAIL PROTECTED]> wrote: >WA9ALS - John wrote: >> Newbie installing latest Sarge would like to configure RAID 1. Would >> someone please show me what the partition table should look like for the >> simple case of / and swap partitions, configured with RAID 1? Thanks! - > >It's not supported on / or /boot - see the thread "Rant about installer >features" for more discussion and some links. Oh but it is. Just do a normal install, then convert it to RAID1- see http://www.miquels.cistron.nl/raid/ Mike. -- The question is, what is a "manamanap". The question is, who cares ? -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: bad shutdown
In article <[EMAIL PROTECTED]>, Silvan <[EMAIL PROTECTED]> wrote: >I think I have a CD-ROM going south. I started the KDE CD ripper flummy, and >then immediately stopped it. I wound up with a kaudiocreator process hung up >eating system CPU cycles and making my hard disk do unpleasant sounding >things filling the log up with "hdd lost interrupt" messages. The system was >stable, but I didn't want to leave it doing that. > >So I closed everything out, logged off, waited a bit, and then rebooted. The >last thing I saw before the reboot was a warning that /home was busy, and so >was not umounted. Makes sense. You probably hit a hardware or kernel bug. As a result some processes (kaudiocreator) got stuck in kernel mode and kept all filesystems they were using busy. At that point there is no way to umount those filesystems. >I booted into unclean/check forced mode on home and / both. Fortunately, >there were no serious problems. I never see serious problems even after a hard reboot. Ofcourse, I use journalling filesystems everywhere (ext3 or xfs). >Sure, I was having weird kernel interrupt problems, but it seems poorly >conceived to me for the init scripts to take one shot at something as >important as umounting everything, and then just say OK, to hell with it, >let's reboot anyway. Well, there's really nothing else to be done. What can be done? Repair the hardware by magic and somehow re-initialize the kernel? Rebooting is the safest way out. Note that the standard shutdown sequence _does_ call sync() just before halt/reboot. >I'm not sure what could have been done at that stage of the game, but I'm >miffed I didn't even get the chance to try. Because there's nothing to be done at that stage. Be glad that the system did reboot (imagine if it's in some remote location). >Anyone have any suggestions what I should do about this to increase the odds >that this won't happen again. (Other than getting rid of the CD ripper >flummy and trying to remember not to try to use the damn CD-ROM until I get >it replaced, I mean.) If you're using ext2, switch to ext3 (make sure your kernel supports ext3, then tune2fs -j /dev/hdaX for all your filesystems, change ext2 to ext3 in /etc/fstab, shutdown -rF now). Mike. -- The question is, what is a "manamanap". The question is, who cares ? -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: kill process by name
In article <[EMAIL PROTECTED]>,
Magnus Therning <[EMAIL PROTECTED]> wrote:
>On Tue, Aug 03, 2004 at 07:58:02PM +0100, Oliver Elphick wrote:
>>On Tue, 2004-08-03 at 19:29, Muratorio, Diego wrote:
>>> Hello Rick, I am looking for the same, did you find something about
>>> kill process by name?
>>>
>>
>>You want killall, from the psmisc package.
>
>Or 'pkill'. I've recently fallen in love with it, and it's cousin
>'pgrep'. Both are in the package procps.
Very true. "pkill" will work as expected on other OSes if it is
available. On the other hand, the killall command does something
*very* different on non-Linux OSes.
If you do "killall something" on Solaris as root, the system
goes down. *Hard*. Better not to get used to something dangerous
like that ("killall" should never have been called that under
Linux, but it's way too late to fix that).
Mike.
--
The question is, what is a "manamanap".
The question is, who cares ?
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: exim? imap?
In article <[EMAIL PROTECTED]>, Hendrik Boom <[EMAIL PROTECTED]> wrote: >I'm running a debian woody server, using exim (I believe -- actually, >I have whatever Debian installs as default, which I believe is >exim), and I have peen persuaded that I should change over to imap. You don't change over to imap from exim - you simply add an imap server, which gives imap access to the existing mailboxes. >I'm trying to get some kind of orientation on the whole process. >Do I have to convet to maildir format? Depends on what IMAP server you choose. >What happend to existing messages in mbox format? If you choose an IMAP server that supports mbox, well, nothing .. >What software do I need to get? Which do you recommend? Dovecot. It's not in woody, but there's a apt-gettable woody version at http://backports.org/ >My clients will be on several other machines (and will no longer have >to ssh to the server to read mail -- that's the whole point). >What client software should they have on Linux? Mozilla or mozilla-thunderbird, or Evolution, or Balsa. >on Windows? Outlook Express does IMAP. You can also install mozilla (-thunderbird). >on computers >likely to be found in public libraries run by idiots? Webmail. For example, install apache + openwebmail. There's a debian package for woody at http://www.apt-get.org/ . There's lot of choice though - search for "webmail" at freshmeat.net to get an idea. "Inside Systems Mail" looks very nice. Mike. -- The question is, what is a "manamanap". The question is, who cares ? -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Exim "From " header
In article <[EMAIL PROTECTED]>,
Master_PE <[EMAIL PROTECTED]> wrote:
>Op za 24-07-2004, om 13:17 schreef Vincent Lefevre:
>> On 2004-07-24 11:00:59 +0200, Master_PE wrote:
>> > I already touth that my Pop3 client is broken. But thats nothing i can
>> > do about. I use Maildir as a format.
>>
>> Then I don't know why you get a "From " line; is it really present
>> in the messages contained in your maildir directory? If yes, this
>> may be because some problems in your config file. Do you have the
>> latest version (possibly modified)? You may try to do a backup of
>> your config files, purge the exim packages (making sure that
>> /etc/exim4 no longer exist) and reinstall the
>
>OK, I use exim 3.35-1woody3. And i have patched ths version with mysql
>support. But when i remove this exim version and instal the official
>version of woody i have the same problem.
>
It's a local configuration issue. Look:
>##
># TRANSPORTS CONFIGURATION #
>##
># ORDER DOES NOT MATTER#
># Only one appropriate transport is called for each delivery.#
>##
>
># This transport is used for local delivery to user mailboxes. On debian
># systems group mail is used so we can write to the /var/spool/mail
># directory. (The alternative, which most other unixes use, is to
>deliver
># as the user's own group, into a sticky-bitted directory)
>
>Virtual_delivery_SPAM:
> driver = appendfile
> group = mail
> user = ${lookup mysql{select uid from TblMailadressen where
>username='$local_part'}}
> mode = 0660
> maildir_format
> mode_fail_narrower = false
> envelope_to_add = true
> return_path_add = true
> directory = /var/domeinen/${lookup mysql{select maildir from TblMailadressen where
> username='$local_part'}}/.SPAM/
A transport for a maildir _must_ also set:
no_from_hack
prefix = ""
suffix = ""
If you add that you won't have a From_ line anymore, and also lines
that start with "From " won't be changed to ">From ".
Mike.
--
The question is, what is a "manamanap".
The question is, who cares ?
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: nfs locking daemon?
In article <[EMAIL PROTECTED]>, Christian Schwarz <[EMAIL PROTECTED]> wrote: >Is there a special "nfs locking daemon" for NFS-clients? > >I'm running a server with nfs-kernel-server and on the server there is a >"lockd" process. If I try to log into GNOME 2.2 on the client (which has >nfs-mounted home directories), GNOME complains about "a problem with your >operating system configuration": "A common cause of this error is that the >"nfslock" service has been disabled." Unfortunately, I can't find a >"nfslock" package in Debian. It's called "nfs-common". Mike. -- The question is, what is a "manamanap". The question is, who cares ? -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: migration of Maildir to new system. Also advice wanted on performance
In article <[EMAIL PROTECTED]>, Pim Bliek <[EMAIL PROTECTED]> wrote: >Unfortunately I didn't get a reply earlier, so maybe my questions were >too vague or mail-guru's are on holiday :). Anyway, I will try to make >it more clear which way I am thinking. > >I did some hours of research and for me the choice comes down to these >alternatives for IMAP: > >Courier-IMAP <--> Cyrus-IMAP You forgot dovecot. It's comparable to Courier-IMAP only it has more features, supports mbox too, and is way faster. Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: moving an installed Debian system onto RAID-1
In article <[EMAIL PROTECTED]>, Robert Waldner <[EMAIL PROTECTED]> wrote: >On Tue, 06 Jul 2004 18:05:20 -0000, "Miquel van Smoorenburg" writes: >>>I've got an installed Debian system which I want to move to RAID-1. > >>Try this: >> >>http://www.miquels.cistron.nl/raid/ > >From what I gather, I need to tell LILO about the exact geometry not > only of the disk, but of the partition. Correct? The geometry of the disk, and the offset of the boot partition. >Can't move to a 2.6 kernel or Sarge/Sid, as this is going to be a > production box. But I'll give it a try tomorrow. It won't work with a 2.4 kernel. You need recent 2.6. Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: moving an installed Debian system onto RAID-1
In article <[EMAIL PROTECTED]>, Robert Waldner <[EMAIL PROTECTED]> wrote: >I've got an installed Debian system which I want to move to RAID-1. Try this: http://www.miquels.cistron.nl/raid/ Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: pppoa package available?
In article <[EMAIL PROTECTED]>, Russell D Cook <[EMAIL PROTECTED]> wrote: > DSL recently became available in my rural area, and I subscribed. I >use a linux stable machine as a firewall and >gateway for my home lan, accessing the dsl modem via an ethernet port. >My provider advises me that they will >be changing over to ATM soon, and will be providing me a new modem. What kind of modem ? Ethernet or USB (actually I mean, how does it connect to your computer). If it's an ethernet modem, then all the PPPoA stuff will be done in the ADSL modem/router itself. Your PC will just speak ethernet. If it's an USB modem, then you have a bigger problem - AFAIK, there's support for only one kind of USB ADSL modem under Linux and not even in the standard kernel - The SpeedTouch USB. See http://linux-usb.sourceforge.net/SpeedTouch/ . Any other modem will likely not work, and getting the speedtouch driver to work is not trivial. Just make sure they provide you with an ethernet ADSL modem/router, or just get one from the store if they don't (zyxel, e-tech, linksys, etc). Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: 2.4.26 upgrade => no auto power off
In article <[EMAIL PROTECTED]>, Glenn Meehan <[EMAIL PROTECTED]> wrote: >I just upgraded my kernel from 2.4.18 to 2.4.26 and now my system won't >auto power down. >I am running testing/unstable. What version of sysvinit are you running? 2.85-16 had a bug that caused this. Has been fixed in the current version. Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: handling runlevels independently with update-rc.d
In article <[EMAIL PROTECTED]>, Sebastian Kügler <[EMAIL PROTECTED]> wrote: >> It's probably easier to use `ln -s` than to get update-rc.d working as >> you would like. > >I am now using symlinking rather than update-rc.d, and it seems to work fine. Sure, until policy changes, the internal implementation of sysv-rc changes, or the user has file-rc installed, in which case things will break (perhaps even spectacularly). Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: kernel 2.6.6 - why shutting down my hard drives while rebooting?
In article <[EMAIL PROTECTED]>, LeVA <[EMAIL PROTECTED]> wrote: >Roberto Sanchez 2004. május 13. 01:51 dátummal ezt írta: >> Thomas Adam wrote: >> > --- LeVA <[EMAIL PROTECTED]> wrote: >> >>Hello! >> >> >> >> >> >>well. I think this is something with a misconfigured kernel, >> >> because this isn't happening with my 2.4.26, and 2.6.5 kernels. I >> >> think I just compiled-in something "wrong" into my kernel. >> > >> > Sounds like you're using APIC. If you want to inhibit this: >> > >> > linux noapic >> > >> > from the lilo prompt. >> >> Actually, this is a result of some changes to the IDE flush code >> that have made it unable to differentiae between a reboot and a >> shutdown. It is being worked by the kernel devs. > >Eh, this is bad :-\ >Can I turn this off? :)) Sure. Go back to a 2.6.5 kernel, then wait for 2.6.7 to be released. Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: bootlogd: ioctl (/dev/ttyzf, TIOCCONS): Bad file descriptor
In article <[EMAIL PROTECTED]>, Niels L. Ellegaard <[EMAIL PROTECTED]> wrote: >I just looked at my system (Sarge/sid/linux-image-2.6.5-386), and it >seems that my bootlogd is not running. It does not create a >/var/log/bootlogd and at some point during the boot process I get the >following error. > > bootlogd: ioctl (/dev/ttyzf, TIOCCONS): Bad file descriptor Looks like something is not well with /dev/ttyzf. Is it present? Are you using 'udev' ? >I pressume that I need to examine the setup file for bootlogd bootlogd doesn't have a config file. >and tell >it to stop trying to listen to /dev/ttyzf, but don't know how to do >this. That's because you can't. Wouldn't make sense; bootlogd *needs* the pseudo terminal or it won't work at all. > I would be grateful for a hint or an enlightening flame :) Bootlogd is disabled by default because on many systems, it simply doesn't appear to work :/ You added yet more proof of that. Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: RAID performance
In article <[EMAIL PROTECTED]>, Alvin Oga <[EMAIL PROTECTED]> wrote: >> As for filesystems, I personally prefer ReiserFS, but that's just >> personal preference. If you're comfortable with ext3, then by all >> means use ext3. > >ditto, reiserfs is a better choice than ext3 > > see the reiserfs vs ext3 benchmarks at namesys.com With many small files reiserfs might be faster. But in all the bonnie-like pure disk I/O tests I ran on fast disk arrays, ext3 is much much faster than reiserfs. And if you're using mainly large files, XFS is faster still. Run your own benchmarks, don't trust others ... Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: software raid and lvm
In article <[EMAIL PROTECTED]>, Greg Sidelinger <[EMAIL PROTECTED]> wrote: >I'm trying to find some documentation how installing testing on a system >using root on software raid1 and lmv. I've done some searching around >and most of the stuff I have seen is for the 3.0 installer and not the >new debian installer. So if anyone knows where I might find some >documentation on this could you please let me know. I'm not exactly sure what your goals are, but if you want to use LVM just to create "partitions" on the RAID1 device, there is another way. Recent 2.6 kernels have support for partitionable raid. There's a MICRO-HOWTO at http://www.miquels.cistron.nl/raid/ Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: /tmp clean at boot time
In article <[EMAIL PROTECTED]>, Umar Draz <[EMAIL PROTECTED]> wrote: >hi dears > > i have woody with kernel 2.4.25 (now i have install mysql through >source) its working . > > now when i reboot my machine. during boot up /tmp directory is clean >and also mysql.sock delete thats why mysql server not run during boot. > >how i can solve this problem that my /tmp directory not clean at boot time? I think your initial problem is that the mysql server doesn't run, and you think it has something to do with the cleaning of /tmp, right? It probably doesn't. If the mysql server needs a socket in /tmp at startup, it will create it itself. Now the Debian mysqld puts that socket in /var/run/mysqld, so it sounds like you're not using the Debian package. Are you perhaps starting the mysql daemon way way too early at bootup, and then bootmisc.sh comes along and removes /tmp/mysql.sock ? Well, that's your problem right there. You should probably start mysql at /etc/init.d/S60mysql or something, not earlier. Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: sid: udev dependency problem
In article <[EMAIL PROTECTED]>, Martin Lorenz <[EMAIL PROTECTED]> wrote: >-=-=-=-=-=- > >The following occurs when I try to install a new version of udev... >seems to be caused by 'initscripts', which is - I have no idea why - not >installed on my system... > >not sure if I should file a bug. seems too trivial a problem to me?! .. large dependency chain .. >[EMAIL PROTECTED]:/etc/apt# apt-get -u -m install mount >Reading Package Lists... Done >Building Dependency Tree... Done >The following packages will be REMOVED: > mount-aes >The following NEW packages will be installed: > mount >WARNING: The following essential packages will be removed >This should NOT be done unless you know exactly what you are doing! > mount-aes >0 upgraded, 1 newly installed, 1 to remove and 10 not upgraded. >Need to get 134kB of archives. >After unpacking 12.3kB disk space will be freed. >You are about to do something potentially harmful >To continue type in the phrase 'Yes, do as I say!' > ?] no >Abort. That's the problem. That "mount-aes" package. It's not in Debian AFAICS, I have no idea where it came from. Initscripts Depends: on a minimal version of mount, you don't have mount installed but mount-aes instead which is not an official Debian package and now you have dependency problems. If you don't need mount-aes just install "mount" and do enter 'Yes, do as I say!' in the above dialog. Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: libc6 - initscripts conflict
In article <[EMAIL PROTECTED]>, Axel <[EMAIL PROTECTED]> wrote: >Hello, >I am trying to upgrade libc6 2.3.2.ds1-10 to libc6_2.3.2.ds1-11 in Sid, >but I get this error: >dpkg: error processing /var/cache/apt/archives/libc6_2.3.2.ds1-11_i386.deb >(--unpack): > trying to overwrite `/etc/default/devpts', which is also in package initscripts > >The installed version of initscripts is 2.85-11. > >Could someone give me a hint on how to solve this? You can't, really. It's a bug in libc6. See http://bugs.debian.org/cgi-bin/pkgreport.cgi?pkg=libc6 Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: /dev/pts/X not getting reused
In article <[EMAIL PROTECTED]>, Adeodato Simó <[EMAIL PROTECTED]> wrote: >Hi, since some days ago, I've noticed that pts devices are not being >reused. I mean, the usual behavior was: open a xterm, which gets e.g. >pts/7, close it, open another one, which will get pts/7 again. Now it'd >get pts/8, and then pts/9, and so on. Not that I care *that* much but I >just prefer the old behavior, where pts get recycled. > >Has anybody here heard of something similar? I've searched the archives >and googled around but I couldn't find a thing. It's a feature of the 2.6 kernel. Not much you can do about it except going back to 2.4 (or hacking the kernel yourself). >[Please CC me on replies, M-F-T set accordingly.] Sorry, reading this (and replying) through a usenet gateway. Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: What's the mistake in this init script?
In article <[EMAIL PROTECTED]>, Adam Aube <[EMAIL PROTECTED]> wrote: >Adam Funk wrote: > >> Sometimes /etc/init.d/exim restart stops but doesn't restart >> the exim daemon. Does anyone see anything wrong? > >Maybe the sleep interval is too short, and Exim doesn't finish exiting >before the script tries to start it again. > >Try changing the "sleep 2" to "sleep 5" and see if that helps. Or change --stop into --stop --retry SIGTERM/10, that sends SIGTERM and then waits to a maximum of 10 seconds for exim to die. Adjust to taste ... see "man start-stop-daemon". Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: how to do/repair a raid1 missing disk install (was: Re: lilo + raid = disaster (again))
In article <[EMAIL PROTECTED]>, Henrique de Moraes Holschuh <[EMAIL PROTECTED]> wrote: >On Fri, 26 Mar 2004, Antony Gelberg wrote: >> sector, and I'm tired of it. If you don't have time to make it into a >> howto, let me know and I'll do it. > >Please do it :-) I lack the time to write (and maintain) such a howto. Note that from the 2.6.5-rc1 kernel on, you can RAID1 2 complete disks and partition the MD device. It's more natural, and performance will be better. The kernel RAID1 code tries to do smart things with head locations / seeks, but it doesn't know about multiple MD devices so if you're using more than one MD device on a diskset, that optimization is gone. Mike. -- Netu, v qba'g yvxr gur cynvagrkg :) -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Promise or 3Ware? - booting
On Fri, 26 Mar 2004 01:08:50, Alvin Oga wrote: > On Thu, 25 Mar 2004, Miquel van Smoorenburg wrote: > > > The problem with software raid is that if the first (boot) disk > > goes south, the system won't boot anymore. If you're using RAID1 > > just for data integrity and don't care if you have to open up the > > machine and replace the disk to get it going again, then that's > > no problem. If you want to maximize uptime, use hardware RAID1/RAID5. > > i've not seen sw raid1 preventing a system boot if either/any disks > goes down ... I have. > - just need to config it correctly .. and test it by pulling > the ide cable off of the disk under test That's if the disk is _dead_. What if you have 2 disks in RAID1 mode, you boot from the first disk, and while the boot sector is OK, the rest of the disk doesn't respond (bad sectors or whatever). The kernel won't be started. There, you're dead in the water. You need to go to the system, and swap the disks physically, before you can boot again. No data loss, presumably, but downtime. The only way to fix this is to use either hardware raid, or a form of software raid supported by the system BIOS. The latter isn't supported under Linux with 2.6 kernels, so that's not an option. > - a properly config'd raid setup should still keep working > even if its in degraded mode Well even if the first (boot) disk is dead completely, if the BIOS doesn't support booting from the 2nd disk automatically (90% of the BIOSes out there don't) you're _still_ dead in the water. > - raid is NOT a reliable data backup > raid1 is the worst for data backup.. > ( "rm filename.foo" and its gone from the other disk too > ( in a matter of seconds You can't say it's not reliable .. it does _exactly_ what it's supposed to do. You could use LVM snapshots I guess. > > I think I should convert our internal dutch documentation to an > > english MINI-HOWTO one of these days. > > whats the url of the dutch version that can be externally accessible ?? It's on an internal webserver that isn't accessible from the outside. Hang on .. http://www.miquels.cistron.nl/bootable-raid1.html Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Promise or 3Ware?
In article <[EMAIL PROTECTED]>, Adrian 'Dagurashibanipal' von Bidder <[EMAIL PROTECTED]> wrote: >On Wednesday 24 March 2004 18.58, Derrick 'dman' Hudson wrote: >> I now need to buy some hardware to set up a 2-disk RAID-1 array for a >> server. The server will run debian with kernel 2.6. The cost >> (money-wise) is not really an issue. The disks are ATA/133 (already >> purchased). Which would you recommend, and why? >> >> Promise FastTrak TX2000 controller >> with 2 Promise SuperSwap 1000 host-swappable drive bays >> >> 3Ware Escalade 7006-2 controller (or 7506-4LP model) >> with 1 3Ware RDC-400 hot-swappable drive cage > >For RAID 1 with only 2 disks, neither. > >The driver problems with promise were mentioned, and the 3ware probably is >overkill. Have you thought about using software RAID? Yes, it uses a bit more >CPU and doubles the PCI bus load, compared with the 3ware (dunno about >Promise - I'm not sure if it does RAID1 in hardware). The problem with software raid is that if the first (boot) disk goes south, the system won't boot anymore. If you're using RAID1 just for data integrity and don't care if you have to open up the machine and replace the disk to get it going again, then that's no problem. If you want to maximize uptime, use hardware RAID1/RAID5. But if you wanted to minimize downtime, you'd have bought hot-swappable SATA disks anyway, I guess. If you're going to do software RAID1 - the latest 2.6.5-rc kernel supports partitionable RAID1 so you can just RAID1 2 entire disks instead of just the partitions. I've updated sysvinit to work correctly on such a system (root partition with dynamic major device). I think I should convert our internal dutch documentation to an english MINI-HOWTO one of these days. Mike. -- Netu, v qba'g yvxr gur cynvagrkg :) -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Concurrent connections on file server
In article <[EMAIL PROTECTED]>, Paul Johnson <[EMAIL PROTECTED]> wrote: >Short answer: A lot of connections, not really. > >Long answer: Well, TCP/IP only allows for something like 65535 >concurrent connections... .. per remote host. Mike. -- Netu, v qba'g yvxr gur cynvagrkg :) -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Severe network slowdown
In article <[EMAIL PROTECTED]>, F.L. Tak <[EMAIL PROTECTED]> wrote: >I'm having very strage network problems with Woody. A seemingly random >amount of time after the machine has booted (I've seen it happen after a >few days, but also after a few hours), all network traffic slows to >almost a halt. If I then reboot, everything is back to normal, so I >guess it't not a problem in the network. Also another machine, which is >on the same local switch, has no such problems. >The only thing I did to the standard Woody installation is build a >2.4.21 kernel with make-kpkg configured with the 2.4.18-bf2.4 config >(since I have a HPT272 IDE controller). Other than that, I have no idea >what might cause this. Even if it is the kernel, I have no idea why. >Also I don't even know where to look. >Can anybody give me some pointers as to what might be the problem or >where I should start looking? Also, are there any tools I might use to >help diagnose the problem? I've heard of this before where people had installed some (firewall?) package that automatically/dynamically added iptables rules. If you end up with thousands of rules, the network becomes dead slow. Mike. -- Netu, v qba'g yvxr gur cynvagrkg :) -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Missing /var/log/wtmp
In article <[EMAIL PROTECTED]>, Cheryl Homiak <[EMAIL PROTECTED]> wrote: >am running debian unstable. I seem to be missing /var/log/wtmp >from my system. I have searched debian >and found no file by that name in the package contents searches. I >searched google; I did >find some correspondence on a cygwin list about it being missing. Their >solution was to insert something in one of the postinstall scripts. Is >there a more straightforward way of re-creating it on my system? I'm >getting daily reports about logrotate not being able to find it. >TIA. # touch /var/log/wtmp Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: /etc/init.d/ - add/remove services
In article <[EMAIL PROTECTED]>, Thomas Hood <[EMAIL PROTECTED]> wrote: >"Miquel van Smoorenburg" wrote in part: >> A thought just hit me. >> What if we added a "update-rc.d enable|disable" command? > >This has already been wished for. See sysv-rc wish #214757 No, that's not what I mean. That is a runlevel editor. I mean a generic disable-startup-at-boottime and enable-startup-at-boottime option, which is what is >and sysvinit wish #67095. Yes, that's what I mean, but renaming symlinks looks ugly to me, while pointing them to /dev/null seems more elegant. Maybe it's just me. >> That has never been done because the implementation would >> be awkward and wouldn't fit into the sysv-rc design. >> But what if we used the destination of the symlink ? > >Suppose you have a service foo enabled in runlevel 2 and disabled >in runlevel 3. On moving from 2 to 3 you want foo to be stopped. >However, if foo is disabled in runlevel 3 by having its S entry >symlinked to /bin/true then it won't be stopped. That's another issue. That should be handled by a runlevel editor. >The right thing to do is to write a simple runlevel editing tool >and include it in the sysv-rc package. This tool would rename >Snn symlinks to K(100-nn) symlinks and vice versa K(100-nn) .. hmm, http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=216028 >and would keep >records of what it had done so that reversion was easy. The sysv >update-rc.d program would be rewritten to work through the runlevel >editing tool; commands coming from update-rc.d would determine what >the tool considered to be the "default" setting. >Other init systems would implement this tool differently, of course. Ouch, complicated ... Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: /etc/init.d/ - add/remove services
In article <[EMAIL PROTECTED]>, Thomas Hood <[EMAIL PROTECTED]> wrote: >Note that the expectation of the System V init system is >that every service have either an S or a K symlink in each >runlevel. If there is no symlink for a service in a particular >runlevel then the behavior of sysv's invoke-rc.d is undefined >for that service in that runlevel. Methods of shutting off >services must take that into account. > >Suppose you have service foo that is S20 in runlevel 2 and >you want to shut it off. You should do something like: > >/etc/rc2.d/$ mv S20foo K80foo_originallyS20 > >As someone has already pointed out, update-rc.d is intended for >use by maintainer scripts. It allows maintainer scripts to >work independently of which init system is in use, whether it >is the System V init system, file-rc or some other. The >local administrator need not use update-rc.d to change sysv >symlinks. A thought just hit me. What if we added a "update-rc.d enable|disable" command? That has never been done because the implementation would be awkward and wouldn't fit into the sysv-rc design. But what if we used the destination of the symlink ? The "disable" command would simply walk over all symlinks in /etc/rc?.d of the form ^S\d+name$ and point them to /bin/true. The "enable" command would do the reverse, point the symlinks back to /etc/init.d/name. That would keep the essence of any local config, and would make it trivial to switch services on and off. Anyone looking at it would also intuitively understand what was going on. Ofcourse invoke-rc.d should know about this, and file-rc needs to be updated (it can easily store this state in its config file I think) but otherwise implementation should be trivial. Only thing is - should the links of the disabled service point to /bin/true or to some special, non-existing location that is easy to detect with test -e ? Or perhaps use /dev/null, which is detectable with test -e file && ! test -s file ? I see that currently /bin/true is a small ELF binary, not a shell script anymore, so it would't be too much overhead to run it. Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: postfix [was Re: Sendmail vs Exim vs Others]
In article <[EMAIL PROTECTED]>, Nano Nano <[EMAIL PROTECTED]> wrote: >On Sat, Jan 31, 2004 at 01:21:46AM +0100, Vincent Lefevre wrote: >> On 2004-01-30 14:57:37 -0800, Nano Nano wrote: >> > Aha, that explains why the 2nd message worked: I have a mutt rule that >> > adds the correct From for list-replies. I guess I'll have to make sure >> > Mutt adds a valid From or Sender in all cases. >> > >> > I'll have to make sure all mail-generating programs do that: is there a >> > facility in Debian to assist with that, or will I need to administer >> > Identities in each MUA individually? >> >> Isn't postfix able to do email rewriting? > >I guess but it apparently doesn't do it OOB. I never liked how Exim >leaked my AccountFullName+EtcEmailAddressesEmail identity in the Sender >when I was using an alternate valid From email and nickname (like here). You can configure exim not to do that if you don't like it. Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Exim vs Procmail (was: Sendmail vs Exim vs Others)
In article <[EMAIL PROTECTED]>, Vincent Lefevre <[EMAIL PROTECTED]> wrote: >On 2004-01-30 18:34:17 +0000, Miquel van Smoorenburg wrote: >> See http://www.exim.org/ . Click on "Documentation and FAQs". > >There are several things I don't like: You're probably right on most of those, exim filtering isn't the end-all be-all of mail filtering, but: >* Exim uses the .forward file which is also used by other MTAs. > This may lead to clashes and lost mail if something changes at > the system level (e.g. another MTA is used). A bit dangerous. > A .eximrc would have been a better idea. This is configurable. Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Sendmail vs Exim vs Others
In article <[EMAIL PROTECTED]>, Vincent Lefevre <[EMAIL PROTECTED]> wrote: >On 2004-01-30 09:03:28 -0800, Steve Lamb wrote: >> Exim does not need an MDA and has its own user-level filtering. > >But the man page is far from being clear and incomplete (compared to >the procmail man pages). > >First, what is the user configuration file? > >How can I interface it with getmail so that exim returns with no error >only when the mail has been successfully stored to the user's mailbox >(or discarded)? > >Can it add a "Lines:" header? See http://www.exim.org/ . Click on "Documentation and FAQs". The same documentation is available as a text file in /usr/share/doc/exim (spec.txt and filter.txt). Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: postfix [was Re: Sendmail vs Exim vs Others]
In article <[EMAIL PROTECTED]>, Nano Nano <[EMAIL PROTECTED]> wrote: >On Thu, Jan 29, 2004 at 09:53:43PM -0800, Nano Nano wrote: >> >> Configuration: Internet with smarthost >> Append .domain? No >> Smtp relay host? My ISPs smtp server >> Final destination domains? >> --->Force synchronous updates on mail queue? Yes >> >> Is that it? Just drop in and go? I did "force syncronous updates" so I >> wouldn't ever lose mail. > >My first test message to the outside world bounced with: > ><[EMAIL PROTECTED]>: host smtp.comcast.net[216.148.227.125] said: >550 >[PERMFAIL] comcast.net requires valid sender (in reply to RCPT TO >command) > >exim always added my "Sender" header for me. > >What to do? It's not the Sender: header. If it refused your message at RCPT TO, it didn't even see the body yet. I bet postfix used an invalid MAIL FROM. Make sure your machine has a valid, existing domain name or that postfix is configured to use a valid, existing domain name as its primary domain (in exim, that's primary_hostname/ qualify_domain - I don't know the postfix equivalent). Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: [OT] Bruce Perens talks to BBC
In article <[EMAIL PROTECTED]>, Nano Nano <[EMAIL PROTECTED]> wrote: >On Sat, Jan 24, 2004 at 11:29:02PM +0000, Miquel van Smoorenburg wrote: >> In article <[EMAIL PROTECTED]>, >> Paul Morgan <[EMAIL PROTECTED]> wrote: >> >Not to mention the fact that the US is following more than one thread by >> >being by far the largest donor of aid to poorer nations >> >> Google for "foreign aid usa denmark netherlands" and you'll >> find things like http://www.just1world.org/development-aid.htm >> or http://www.vexen.co.uk/USA/foreign_aid.html which show that >> this is untrue to say the least. > >Lies, damn lies, and statistics. Amen. >Fact is we feed more people than them: TRUE. Well, may I refer you to the first line of your posting. Thanks. Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: [OT] Bruce Perens talks to BBC
In article <[EMAIL PROTECTED]>, Paul Morgan <[EMAIL PROTECTED]> wrote: >Not to mention the fact that the US is following more than one thread by >being by far the largest donor of aid to poorer nations Google for "foreign aid usa denmark netherlands" and you'll find things like http://www.just1world.org/development-aid.htm or http://www.vexen.co.uk/USA/foreign_aid.html which show that this is untrue to say the least. Wrt Iraq: Search Google News for "David Kay". Google is one of the best American companies. Use their tools to your advantage. Mike. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Fwd: xfs_force_shutdown(md(9,0),0x8) called from line 1070 of file xfs_trans.c.
In article <[EMAIL PROTECTED]>,
alberto <[EMAIL PROTECTED]> wrote:
>we're having nasty problems an our new raid on a linux box:
>When performing hevy i/o (in present case was scp -r), the disks
>get disconnected with this kernel message:
>
>-
>Dec 3 15:55:34 machine sshd(pam_unix)[1791]: session opened for
>user by (uid=500)
>Dec 3 15:56:08 machine kernel: 0x0: 58 41 47 46 00 00 00 01 00 00 00
>04 00 10 00 00
>Dec 3 15:56:08 machine kernel: xfs_force_shutdown(md(9,0),0x8)
>called from line 1070 of file xfs_trans.c. Return address =
>0xf8a23aa8
>Dec 3 15:56:08 machine kernel: Filesystem "md(9,0)": Corruption of
>in-memory data detected. Shutting down filesystem: md(9,0)
>Dec 3 15:56:08 machine kernel: Please umount the filesystem, and
>rectify the problem(s)
>--
>
>What is the nature of this problem? Kernel drivers? Hardware
>failure? Filesystem inconsistency?
I've seen this message ("Corruption of in-memory data detected")
a number of times on the linux-kernel mailinglist, and (worse)
I've seen it myself on a machine with an XFS filesystem.
The machine I used had both software- and hardware raid, and
I can't remember if that was on the MD array or on the hardware
raid5 array ..
There's probably either a bug in XFS or a bug in the combination
of XFS and MD.
>What can be a possible solution?
Debug the XFS filesystem and the MD/raid layer, find the
bug, fix it, and submit it to the kernel developers ...
Mike.
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
