Re: Argumente gegen HTML in emails
Also suche ich jetzt alle Argumente *gegen* HTML in E-Mails. Wenn was wichtiges fehlt bitte ergaenzen :) - Speicherplatz: Mehrtraffic durch HTML-Kontext pro Mail (Bandbreitenbelastung) + mehr Speicherplatz im Mailfile - Nicht jeder Mailclient kann HTML darstellen - Die Leute versuchen ihre Mail zu verschönern anstatt auf den Inhalt zu achten - inoffizieller (?) Internet-Standard ist Plaintext Ich weise Unwissende gelegentlich auf folgende Texte hin. http://www.kaipahl.de/brain/email_sbo_01.html#htmlformat http://www.intern.de/news/3711.html http://www.kasper-online.de/goldmail/regel25.htm Stony -- Russian roulette for linux: [ $[ $RANDOM % 6 ] == 0 ] rm -rf / || echo Still breathing, eh? -- Haeufig gestellte Fragen und Antworten (FAQ): http://www.de.debian.org/debian-user-german-FAQ/ Zum AUSTRAGEN schicken Sie eine Mail an [EMAIL PROTECTED] mit dem Subject unsubscribe. Probleme? Mail an [EMAIL PROTECTED] (engl)
renew expired certificate
Hi all, on a machine a self-signed certificate expired recently. It was created via mod-ssl-makecert (according to /usr/doc/libapache-mod-ssl-doc/README.Debian.gz). After googl'ing and reading HOWTOs I still don't know how to to renew it (it shall have a new expiration date). For example, I tried # openssl ca -ss_cert /etc/apache/ssl.crt/server.crt -days 1000 ./demoCA/private/cakey.pem: No such file or directory So it looks as if openssl is looking for a CA which doesn't exist. I assume that my certificate has been created without a CA, just using my text inputs for the issuer data. How can I renew my certificate? (Note: I'm not interested in a new certificate, but want to prolong the existing one.) TIA Stony -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: restricting command line arguments in sudo
I allow read-access to all of /var/log. However, I also allow read access to /etc/shadow: /usr/bin/tail /var/log/../../etc/shadow does work. How can I best restrict that? I've tried /usr/bin/tail/[^.]* /usr/bin/tail /var/log/[^.]* will prevent sudo tail /var/log/../../etc/shadow but not sudo tail /var/log/apache/../../../etc/shadow :-( I have no better idea. Stony -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
1:3.0.2p1-9 1:3.4p1-0.0potato1 ?
On a potato machine I'm trying to apt-get the sources of the latest security update of openssh. With deb-src http://security.debian.org stable/updates main contrib non-free I get 1:3.4p1-0.0potato1 as expected. With deb-src http://security.debian.org stable/updates main contrib non-free deb-src ftp://ftp.uni-erlangen.de/pub/Linux/debian testing main deb-src ftp://ftp.uni-erlangen.de/pub/Linux/debian/non-US testing/non-US main I get 1:3.0.2p1-9 from testing. Is 1:3.0.2p1-9 considered greater than 1:3.4p1-0.0potato1 because it comes from testing, is it a matter of order in sources.list or are there rules I don't know yet? TIA Stony -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: 1:3.0.2p1-9 1:3.4p1-0.0potato1 ?
Is 1:3.0.2p1-9 considered greater than 1:3.4p1-0.0potato1 because it comes from testing, is it a matter of order in sources.list or are there rules I don't know yet? last night that confused me too. eventually i tracked it down to the Packages file not being updated. I would suspect they did not update the Packages file(at least as of when i looked) so they have the files there, and test them in the meantime before making them accessable to apt-get Sounds sensible. Reading your reply I remember myself seeing the 1:3.0.2p1-9 entry in Packages, too. Maybe I was too tired to recognize this as the root of my problem. Thanx Stony -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Semi-Hot-Swap of IDE discs
Dear list, for backup purposes (on i386 architecture) I'm planning the purchase of two 60GB-IDE discs, an exchange frame [right word?] and a PCI-IDE-Controller Card. SCSI is no option due to the huge difference in the price, at least here in Germany. Since my knowledge about IDE/ATA is near zero I'm curious if it is possible to swap IDE disks without shutting down the PC. Of course most of the time and especially on exchange the disc will not be mounted. I know that there are (expensive) exchange frames made for this purpose, but is this the only way? Do you know controller cards which can handle this? What about a self-installed switch to turn the disc off before swapping? TIA Stony -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Identical installations on several machines
A# dpkg --get-selections selections If you want removed/purged packages on A to be removed/purged on B: A# dpkg --get-selections \* selections Stony -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Logging transactions
Is there some way for me to write this info to a log file for troubleshooting, and still be able to see it on the screen? man tee HTH Stony -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Openssh protocol 2 for potato - rather urgent help needed
Has anyone any idea where I can find debs for potato for Recent versions of OpenSSH? I need non-broken protocol 2 but still be able to support protocol 1. Building OpenSSH3.1p1 from source requires an upgrade of OpenSSL. If I replace the OpenSSL in Potato with version 0.9.6c (from openssl.org)v will it break stuff using the version in Potato? I built and installed openssl and openssh taken from testing in Oct. and experienced no problems. As an example I happily run mod-ssl. http://lists.debian.org/debian-user/2002/debian-user-200203/msg00070.html HTH Stony -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: security vs. potato?
according to packages.debian.org/ssh2 there is no ssh2 package available for potato/stable. i suppose this is a conundrum for the developers -- normally security fixes are beamed back to potato in a hurry, but ssh (version 1) has security troubles, and to fix them would introduce a new package (ssh2) which is against 'stable' policy... what's the fix for a potato production server? can ssh2 be had from nonstandard apt sources for potato? If you want to avoid protocol version 1, you can get and build the openss[lh] sources from testing and install the binaries. It worked for me on Oct. 20th: # dpkg -i /usr/local/src/DEB-SRC/openssl/openssl_0.9.6b-2_i386.deb # dpkg -i /usr/local/src/DEB-SRC/openssl/libssl0.9.6_0.9.6b-2_i386.deb # dpkg -i /usr/local/src/DEB-SRC/openssl/libssl-dev_0.9.6b-2_i386.deb # dpkg -i /usr/local/src/DEB-SRC/openssh/ssh_2.9p2-6_i386.deb HTH Stony
Re: UrlView strangeness
I use mutt which uses urlview to present a list of url's in a message. Lately, sometimes some of the lines in that list are blank (sometimes they're *all* blank). If I count the url's in the message and arrow down to the appropriate blank line it fires up lynx and takes me where I want to go. Has anyone else experienced this? I can't solve your specific problem, but you might be interested in what I did. Feel free to ask, if something isn't clear. Since I was not satisfied with urlview (just URLs, no context), I defined 2 keybindings in .muttrc macro index \cb |formail -I \\ | txt2html.py $HOME/tmp/tmp.html\n!links $HOME/tmp/tmp.html\n macro pager \cb ...[the rest is the same]... The script txt2html.py: #! /usr/bin/env python txt2html - Convert plain text to html with urls as links. Usage: txt2html in.txt out.html import sys, re url_rexp = re.compile(r(http|https|ftp)://[-a-zA-Z:0-9./#%+~_?=()@]+) def replace(match_object): url = match_object.group(0) return a href=%(url)s%(url)s/a % vars() txt = sys.stdin.read() html = !DOCTYPE HTML PUBLIC -//W3C//DTD HTML 4.0//EN html head meta http-equiv=Content-Type content=text/html; charset=iso-8559-1 titleConverted by txt2html/title /head body pre %s /pre /body /html % url_rexp.sub(replace, txt) sys.stdout.write(html) Stony
Re: Need Help to Program Text Draw Utility
But it seems that there is no Linux-utility to draw simple ascii-figures (lines, text, rectangles, circles) with cut/copy/paste-capacity in textmode. IIRC this topic has been on this list before. There's even an emacs mode for this. I'm not sure if you'll find cut/copy/paste-capacity for single objects. You could also try to use a real draw-program and convert the image with aview which is available as a debain package, but I'm quite sure that this is not what you exactly want. If you want/have to create such a program yourself, what about drawing in dia (uses XML format) and write a simple dia-to-ascii converter using python's XML parser or whatever you like? HTH Stony
Format/Tool-Chain for scanner-images-document-printer
What would you consider the appropriate image and document format in this situation: Most documents will be created by scanning paper and combine the images (=pages) to one doc. Most of the time the documents will be printed on a PS Kyocera, sometimes 2- or 4-up, always with small dynamic text near one edge. Do you have any hints/experience wrt. formats and tools? TIA Stony
Re: backup vs package manager
dpkg --get-selections file # save package settings. dpkg --set-selections file # set package settings. Once you've done that, is there a way to simply tell the system to install, uninstall, upgrade, whatever, to exactly what the current package selection settings are? Before backup: dpkg --get-selections \* file # save all package settings Restore: apt-get update dpkg --set-selections file # set package settings apt-get dist-upgrade # apt-get upgrade doesn't install not # installed packages Stony
Re: Non-interactive password check
Alexander Since I have no administrative access to the university Alexander servers LDAP doesn't seem to be an alternative. Given Why do you need administrative access to the LDAP server? I have no administrative access to the university server. Read: I can't install LDAP there. Stony
Re: rsync's authentication?
does anyone know what password authentication rsync uses, and how the passwords on the wire are encrypted? i know i could read andrew's paper, but i just don't have the time. thanks! From the man page: -e, --rsh=COMMAND This option allows you to choose an alternative remote shell program to use for communication between the local and remote copies of rsync. By default, rsync will use rsh, but you may like to instead use ssh because of its high security. 1 Mar 1999 12 rsync(1) rsync(1) You can also choose the remote shell program using the RSYNC_RSH environment variable. rsync --rsh=ssh ... will give you good security. Stony
Non-interactive password check
Currently I'm building a web application using python. The database only stores a login name and an account name (e.g. 'john' and '[EMAIL PROTECTED]') but no password. (This is invariant.) On session start the user sends his login name and the password for the remote account. I want to check the authenticity by something like trying to ssh into the remote account and immediately logout again. The unix user doing this is the owner of the apache process. Unfortunately this approach needs interaction (enter the password, logout). I was not surprised to find no option in ssh's man page to give the password on the command line. Do you know any way to make ssh less interactive? Stony
Re: Non-interactive password check
On session start the user sends his login name and the password for the remote account. I want to check the authenticity by something like trying to ssh into the remote account and immediately logout again. The unix user doing this is the owner of the apache process. If this is the only time you'll ever want to authenticate users outside of account login, this hack will work. If you want a solution that scales (so that one day your users can be authenticated by mail and web servers, web applications, anything...), use LDAP. Thanx for your answers so far. I have to specify my situation more precisely: Working in the student represention of a university I'm implementing a printing service (a student make a selection of old exams and the like that will be treated as an order and printed out). Since all our students have a UNIX account in the university network we decided to not maintain user passwords on _our_ server but use the existing university accounts and connect them with our users. (The users don't have to remember one more password (for a relatively rarely used service) and we don't have to bother with the inevitable I forgot my password.) Since I have no administrative access to the university servers LDAP doesn't seem to be an alternative. Given that ssh is the only access method in question is there any way (or ssh'ish program) to avoid the use of something like ExpectPy (expectpy.sourceforge.net)? Stony
Re: how to put package on hold using command line tools?
is there any way to put package on hold using command line tools? echo packagename hold | dpkg --set-selections thank you, isn't this a bug in man dpkg though? all it says about --set-selections is: dpkg --set-selections Set package selections using file read from stdin. that's not enough to actually use it like you suggested above.. Regard --set-selections as the inverse of --get-selection --- quite intuitive. Nevertheless you can file a bug against dpkg. (See 'apt-cache show bug'.) Stony
Own DNS server inside an existing network
Dear list, I have a host inside a university network (IP a.b.c.d). There's a DNS server (out of my control) authoritative for the university's domain uni.tld and another one (which I can influence) authoritative for sub.uni.tld. The latter maps my host myhost.sub.uni.tld to the IP a.b.c.d. I am the owner of mydomain.tld which is currently hosted by an ISP. Would it be possible to set up a nameserver at a.b.c.d that will be authoritative for mydomain.tld given that the ISP allows a custom nameserver? What about reverse lookup? What do I have to do to let a.b.c.d resolve to myhost.mydomain.tld instead of myhost.sub.uni.tld? Since I seem to have a (fundamental?) lack of knowlwdge in DNS you can also feed me with recommended reading. Sorry for being OT. TIA Stony
Re: How do I know whether I've got the potato or woody versions of debian?
I've installed Debian using a couple of CDs which I downloaded. I'm still very new to Debian. Thus the question IIRC the debian version number is part of the name of the isos. But you can also just try cat /etc/debian_version 2.2 == potato woody _will_ be 3.0 when it's stable. Dunno if you'll find 3.0 in woody's /etc/debian_version. Stony
Re: apt logfile?
Is there an apt (or dpkg) logfile? Something that would keep a record of what was done when? Just a simple text file with a format of: timestamp action full-package-name Does anyone else think this could be very useful? Me2! It's not what you meant, but sometimes I use ls -ot /var/lib/dpkg/info/*.md5sums | head -20 Stony
Re: short reference for Unix commands
Sorry for not being precisely enough: Rather than introductory docs I'm looking for a _short_ reference of Unix / Linux commands. The best would be _one_ table with all command names and descriptions and maybe links to the man pages. Stony
short reference for Unix commands
Any URLs for that? TIA Stony
session management / authentication
Could you give me hints (URLs, book titles) to inform me about session management / user authentication on a web server? I'm planning a DBMS-backed web app. (using apache, postgresql, python). My current aproach is to use https, let the user auth. himself via password and generate a session id so that subsequent requests are valid if this id is used. The id becomes invalid when the user quits the session or after 10 minutes of inactivity. What are the problems with my approach? (There have to be some, since I see so many sites not following it.) TIA Stony
mod_python on potato
Has anyone out there tried to get mod_python to work on a potato server? (Which versions, dependancies?) Stony
How to disable X temporarily?
What would you regard as the most elegant way to keep all users of a system from starting a X session. It should be a) temporarily and b) nothing like apt-get remove ... ;-) TIA Stony
Re: How to disable X temporarily?
On Mon, Sep 17, 2001 at 01:43:48PM +0200, Alexander Steinert wrote: What would you regard as the most elegant way to keep all users of a system from starting a X session. Don't know how elegant but how about - # chmod /usr/X11R6/bin/X Nice trick, Kent. Thanx! Stony
Re: Sys Admin guide specific to Debian?
I'm a Linux beginner with Debian installed. I'm looking for a beginner's guide to System Administration and I'm wondering if there might not be one particular to Debian. If not, can someone point me in the direction of a good generic SysAdmin guide? You might want to take a look at http://www.infodrom.ffis.de/Debian/doc/ especially http://www.infodrom.ffis.de/Debian/doc/maint/ HTH Stony
Re: Reading .doc files from within Mutt
What is the best (simplest) way to read .doc files from within Mutt? /etc/mailcap: application/msword; /usr/bin/antiword '%s'; copiousoutput; description=Microsoft Word Text; nametemplate=%s.doc and (if you want) ~/.vimrc: auto_view application/msword Stony
Re: MUA with html support
I do it with a combination of a line in my muttrc and a couple of lines in my /etc/mailcap first put this line in /etc/mailcap: text/html; /usr/bin/links -dump '%s'; copiousoutput; description=HTML Text; nametemplate=%s.html Which version of links supports -dump? I've also got auto_view lines and corresponding mailcap lines for postscript, tex, and M$ word!! (antiword). Could you send the ones for ps and tex, please? TIA Stony
Re: PDAs
on Thu, Jun 07, 2001 at 12:31:57PM -0400, Carl Fink ([EMAIL PROTECTED]) wrote: I'm going to buy a PDA within the next week or so. After fairly extensive research it looks like I have to get a PalmOS device. EPOC is seemingly dead in the water, Compaq isn't selling their devices without Windows on them, and Agenda VRs (bizarrely) can't sync with Linxu systems, according to PDA Buzz. The Agenda VR/3 should, if I'm remembering articles correctly, be able to rsync with your GNU/Linux box ;-) http://www.cs.umbc.edu/~acedil1/agenda/network.shtml Stony
ntpdate and diald
Hi, I want to use ntpdate to set the system time and date during the boot sequence of a system with a ADSL connection which takes almost 10 seconds to be brought up. [EMAIL PROTECTED] # ls -1 /etc/rc2.d/ [...] S19bind@ S20diald@ S20inetd@ S20ipchains@ S20logoutd@ S20makedev@ S20masqmail@ S20net-acct@ S20samba@ S20ssh-nonfree@ S20ssh2@ S30pppoed@ S30squid@ S32ntpdate@ [...] [EMAIL PROTECTED] # cat /etc/init.d/ntpdate [...] case $1 in start) [...] ### The -d (debug) option is just set until it works. /usr/sbin/ntpdate -d -b -t 11 194.95.250.36 131.188.34.75 \ 130.149.17.21 /tmp/ntp.log echo . ;; [...] But: [EMAIL PROTECTED] # cat /tmp/ntp.log [...] 16 May 08:14:50 ntpdate[607]: no server suitable for synchronization found [EMAIL PROTECTED] # cat /etc/diald/dynamic.filter [...] accept udp 60 udp.dest=udp.ntp accept udp 60 udp.source=udp.ntp accept udp 60 udp.dest=udp.timed accept udp 60 udp.source=udp.timed [...] I don't want to put the ntpdate command into /etc/diald/ip-up since it should be called only once during the boot sequence. The ntpdate command does work perfectly when called from the command line after bootup! Any hints? TIA Stony
Re: init.d shell script can't stop daemon (python script)
I wrote a (python) script that is acting like a daemon (doing something, sleeping for 10 seconds, doing ..., sleeping ...) and would like to launch and kill it from a shell script in /etc/init.d. Using start-stop-daemon seemed appropriate, but the problems is that no /var/run/$NAME.pid is written and /proc/PID/exe points to /usr/bin/python. Hence stopping the daemon or preventing further deamons from starting is not possible :( I make my python daemons test (early) and then write (late) their own pidfile as part of the process of daemonifying themselves. Stopping: The --pidfile should now work of course, but I throw in --execfile /usr/bin/python --name python --user foo if only to document for myself what works. Those parameters are for a daemon that is a python script starting with the header #!/usr/bin/env python Thanx for your suggestions. I finally got it to work with all the daemon logic inside the init.d shell script: case $1 in start) echo -n Starting $DESC: start-stop-daemon --start --pidfile /var/run/$NAME.pid \ --make-pidfile --background --startas $DAEMON echo $NAME. ;; stop) echo -n Stopping $DESC: start-stop-daemon --stop --quiet --oknodo \ --pidfile /var/run/$NAME.pid rm -f /var/run/$NAME.pid echo $NAME. ;; where $DEAMON is the full pathname of the executable python script. Stony
Re: debian newbie questions -- security
3) I want to have a system that is as secure as possible without sacrificing usability. Where can I get good guidance on securing Debian? I Not only for that you might want to take a look at http://www.infodrom.ffis.de/Debian/doc/index.html Stony
envelope and virtual domain with postfix
Hi, my /etc/postfix/virtual on host.bar.tld contains [EMAIL PROTECTED] @localhost'. My ISP provides webspace for foo.tld and is configured to forward all mail addressed to [EMAIL PROTECTED] to [EMAIL PROTECTED] Incoming mail to say [EMAIL PROTECTED] arrives at [EMAIL PROTECTED] and contains [EMAIL PROTECTED]' only in the To: header and in exactly one Received: from ... by ... for ... header. How can I make postfix write [EMAIL PROTECTED]' onto the envelope in order to get the virtual domain working? TIA Stony
init.d shell script can't stop daemon (python script)
Hi, this might be a little OT, but: I wrote a (python) script that is acting like a daemon (doing something, sleeping for 10 seconds, doing ..., sleeping ...) and would like to launch and kill it from a shell script in /etc/init.d. Using start-stop-daemon seemed appropriate, but the problems is that no /var/run/$NAME.pid is written and /proc/PID/exe points to /usr/bin/python. Hence stopping the daemon or preventing further deamons from starting is not possible :( Here's the relevant part of my init.d script: #! /bin/sh PATH=/usr/local/sbin:/usr/local/bin:/sbin:/bin:/usr/sbin:/usr/bin DAEMON=/usr/local/sbin/printpsspoold NAME=printpsspoold DESC=printpsspool daemon test -f $DAEMON || exit 0 set -e case $1 in start) echo -n Starting $DESC: start-stop-daemon --start --quiet --pidfile /var/run/$NAME.pid \ --exec $DEAMON echo $NAME. ;; stop) echo -n Stopping $DESC: start-stop-daemon --stop --quiet --pidfile /var/run/$NAME.pid \ --exec $DAEMON echo $NAME. ;; . . . esac exit 0 Any hints? TIA Stony
Re: wait till end of write, how?
One alternative is to use a semaphore file. Have the NT app create a file named, say busy, just prior to starting the transfer of the big file. Once the big file is transferred the NT box erases busy. The shell script would be written so that if the busy file exists, it doesn't touch the big file. And if you can't do that, just have the script checksum the file every 5 secs, and if it hasn't changed for 10 secs, you're probably set. Use the suggestion from Rick first though. Since the W2k client users just copy the relvant files using drag'n'drop something like a semaphore file is not possible. Mike's suggestion (watch if the file doesn't change for a certain time) was only a workaround for me when first addressed the problem, but I think I can live with it because the whole situation is a temporary workaround. Thanx for all your ideas. Stony
wait till end of write, how?
Given that a W2k client copies a large file into a samba share on a Linux server, how can a process (e.g. shell script) that will read this file wait until the file is totally transfered? The problem is that the file is (on the server) accessible for reading from the moment the client starts to copy. My first guess is to try something like `echo file' and hope that linux throws an error, because file is already opened for writing by the samba daemon. But maybe samba writes large files in blocks (i.e. opens and closes the file several times), who knows? TIA Stony
get number of pages of a PostScript file
What's the fastest way to do this on the command line? TIA Stony
logout freezes system
Dear list, when I log myself out from X the screen shows starnge vertikal lines. The system doesn't react anymore (no keyboard, no mouse, no ping from another host; just hard-reset) This happens only sometimes / is not reproducable. It's not WM specific. I'm running potato with a custom 2.2.18 kernel and using: xfree86-common 3.3.6-11potato xdm3.3.6-11potato ... Can you give me any hints, where to look for system messages that might help and/or which debugging options I could use (for X / xdm?)? TIA Stony
kernel 2.2.19 compile error (toshiba support)
Dear list, today I tried to create a kernel-package with make-kpkg --revision=mork.1 kernel_image using kernel-source-2.2.19 but I got cc -D__KERNEL__ -I/usr/src/linux/include -Wall -Wstrict-prototypes -O2 -fomit-frame-pointer -fno-strict-aliasing -pipe -fno-strength-reduce -mpreferred-stack-boundary=2 -m486 -malign-loops=2 -malign-jumps=2 -malign-functions=2 -DCPU=686 -c -o toshiba.o toshiba.c toshiba.c:93: parse error before string constant toshiba.c:93: warning: type defaults to `int' in declaration of `MODULE_PARM' toshiba.c:93: warning: function declaration isn't a prototype toshiba.c:93: warning: data definition has no type or storage class toshiba.c: In function `tosh_open': toshiba.c:286: `MOD_INC_USE_COUNT' undeclared (first use in this function) toshiba.c:286: (Each undeclared identifier is reported only once toshiba.c:286: for each function it appears in.) toshiba.c: In function `tosh_release': toshiba.c:294: `MOD_DEC_USE_COUNT' undeclared (first use in this function) make[4]: *** [toshiba.o] Error 1 make[4]: Leaving directory `/usr/src/kernel-source-2.2.19/drivers/char' Is there somebody who could explain this / help me out? TIA Stony
Backup techniques
Hi folks, I'd like to know which programs and strategies you (would) use to backup *one* debian box. In one case I have available a CD-Burner as backup device and in the other case a 640MB MO drive. Constraint: I don't want a complete backup of all partitions but in case of bad luck to install a new debian system and then restore the files from the backup. Where (in the file system hierarchy) should i draw the line between files restored from a backup and those to be restored by a new installation. (I know that I need a special backup for my postgres databases.) I read the manual of afio but it didn't convince me, so I'm considering using tar and gzip. To create the whole backup archive first and then split it into pieces (volumes) might take too much space, but I have a special partition (~800MB) available. To employ split seems not appropriate (Hey, wait! I want to change the medium.) The package description of afio implies that compressed tar archives might not be save. Is it safer to compress first and archive then? (This would decrease the compression rate due to Ziv Lempel, and extracting would be more complicated.) Of course it would be nice to be able to restore only a certain file. If somebody has experience with kbackup and kbackup-multibuf (espescially with CD-Burners) I would appreciate to hear it. Is it clever to constrain access to the machine during the backup process? BTW: Why is there a standard user called backup who can't read all files? I'm especially interested in how a complete restore procedure would look like (say the hard disc was broken and I bought a new one). Pointers to any kind of HOWTO doc are also welcome. Stony