Outlook more efficient in storing mails?
Hello! I've been reported that Outlook on Windows is more efficient in storing mails with attachments, as it stores them in unencoded 8-bit format while the various Unix tools[1] store them as they've been received, so a big attachment would be stored mime-encoded taking 1/3 more disk size. Does someone know of work-arounds to this with Thunderbird or Evolution? Ciao, Enrico [1] except of course mutt storing to compressed mailboxes, but then they can't be big -- GPG key: 1024D/797EBFAB 2000-12-05 Enrico Zini [EMAIL PROTECTED] signature.asc Description: Digital signature
Re: [OT] Sensible Dual/Quad w/ Lots of Ram System
On Sun, Aug 15, 2004 at 05:40:42AM -0700, William Ballard wrote: I currently have P43.2 1GB, and it's overclocked. But video encoding still takes hours and mozilla compile still takes a while. The expert group that appears to be behind me in this moment suggests you just use some faster form of mass storage. Which, if you think about it, makes sense :) Ciao, Enrico -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
offlineimap is slow during the initial sync
Hello. I'm setting up offlineimap to synchronize two mail databases archived in mbox and accessed using dovecot-imapd. At the first run, it's copying a lot of messages around creating duplicates, and it's extremely slow (it's processing a 30Mb mbox and it's been eworkking for more than 2 hours on that). I imagine that creating duplicates is somehow part of it's working, but I don't know about the slowness: maybe it's because of the mbox format? If that is the case, are there tools for converting back and forth from mbox to maildir? Would they help? Ciao, Enrico -- GPG key: 1024D/797EBFAB 2000-12-05 Enrico Zini [EMAIL PROTECTED] -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: what's the diff between 'aptitude upgrade' and 'aptitude dist-upgrade'?
On Mon, Jul 05, 2004 at 02:22:40AM -0500, dircha wrote: An equally adequate reference is available as: /usr/share/doc/aptitude/README ...which is not too hard to paste: aptitude dist-upgrade This command will also attempt to upgrade packages, but it is more aggressive about solving dependency problems: it will install and remove packages until all dependencies are satisfied. Because of the nature of this command, it is possible that it will do undesirable things, and so you should be careful when using it. Ciao, Enrico -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: accessing a fuji digital camera
On Mon, Jun 14, 2004 at 05:46:37PM -0800, Greg Madden wrote: And after I try to mount it (mount /dev/sda /mnt/fuji): mount -t vfat /dev/sda1 /mnt/fuji You must use sda1, not sda: the data is on the first (and only) partition of the disk. Ciao, Enrico -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Survey on Debian usage
* Enrico's survey on Debian usage - April 2004 This is an exploratory survey on Debian usage. It is intended to gain some insights on Debian usage experience. The goal of this survey is to understand the purposes for which Debian is being used and the ways in which people use Debian to achieve (or try to achieve) these purposes. While this survey is wholly voluntary, the data you give here can be an essential building block for constructing useful documentation to provide Debian Developers more insight on how the distribution is being used. Filling this survey shall take around 30 minutes. Please try and take the time to respond. To fill in the survey, simply reply quoting this message and add your answers near the A: marks. There's no limit to the length of answers, so, if you are in doubt, don't hesitate to be verbose. English answers are preferred, but you can reply in Spanish, French or Italian if you feel more comfortable doing so. All responses will be anonymised before publishing the data. Replies will be published without headers and signature, while the original messages will only be seen and kept by me (Enrico Zini [EMAIL PROTECTED]), and of course the administrators of some mail servers on the way to here. You can use encryption if you prefer: my key ID is 797EBFAB. Results will be published in my Debian pages (http://people.debian.org/~enrico/), and announced on the same lists where I posted this survey. Thank you for your time! * Survey - Q: Are you a: (note: you may choose more than one) A: [ ] Debian developer, or otherwise involved in contributing to Debian [ ] Free software developer [ ] User, using Debian for: _ [ ] Other: __ - Q: Could you please provide one or two typical episodes (whatever you feel like) of your life with Debian? A: - Q: Could you please provide one or two fantasy episodes of your life with Debian, if you were to make a movie or a comic strip about yourself as a Debian protagonist? A: - Q: What Debian distribution are you using? A: - Q: What do you think people use Debian for? A: - Q: What is the thing you most frequently use Debian for? A: - Q: What do you think is the most INTERESTING aspect or feature of Debian? A: - Q: Are there some aspects or features of Debian that do not satisfy you? A: - Q: What do you think is the most USEFUL aspect or feature of Debian? A: - Q: Is there something you especially like doing with Debian? What is it? A: - Q: Is there something you DO NOT like doing with Debian? What is it? A: - Q: What is the MOST SUCCESSFUL deed you have achieved with Debian? A: - Please take a moment to recall what is the WORST INCIDENT you had with Debian. Q: What was it? A: Q: How did you get into it, and out of it? A: - Imagine you have a crystal ball allowing you to see 10 years in the future, and have a look inside. Q: Will you still be using Debian? A: Q: If yes, what will you be doing with it? A: - Q: Finally, please feel free to leave any feedback to Debian Developers A: Thank you again for your time and help in making improvements to Debian! Enrico -- GPG key: 1024D/797EBFAB 2000-12-05 Enrico Zini [EMAIL PROTECTED] signature.asc Description: Digital signature
Undoing the 'l' command in mutt
Hello, I use mutt, unchanged keybindings. After I type 'l' to see only the messages matching a given pattern, I'd like to get back to seeing the whole mailbox. Is there a way of removing the 'l' filter besides reopening the mailbox? The simpler way would be 'l'+Enter, but it does not work. I had a look in the '?' list of all keybindings, but found nothing. Ciao, Enrico -- GPG key: 1024D/797EBFAB 2000-12-05 Enrico Zini [EMAIL PROTECTED] -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Undoing the 'l' command in mutt
On Tue, Apr 13, 2004 at 08:40:36PM +0200, Philipp Weis wrote: I use mutt, unchanged keybindings. After I type 'l' to see only the messages matching a given pattern, I'd like to get back to seeing the whole mailbox. Is there a way of removing the 'l' filter besides reopening the mailbox? The simpler way would be 'l'+Enter, but it does not work. I had a look in the '?' list of all keybindings, but found nothing. You should have a look at the patterns section of the mutt manual. ~A matches all messages, so l~A shows you all messages again. Ok, so there is no direct remove filter command, and a kludge is needed. Totally non-intuitive, but fair enough. Wishlist bug has been filed. Thank you, Enrico -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Setting up a Taiwanese KDE environment
Hello, I'm trying to setup a proper Taiwanese Sarge. Just installing the correct localization files, KDE and OpenOffice have Big5 menus and everything, Konsole is able to view Big5, applications talk Taiwanese. Cool! Now, it comes the turn of being able to input Taiwanese with the Zhu-Yin input method. I've tried cpanel, which tries to regenerate locales and then hangs. Note: I already have all correct Taiwanese locales generated, but it overwrites my locales.gen file and takes ages to regenerate things I don't use. Then I've tried running xcin and xcin2.3. xcin2.3 shows a nice input method selection thingie, which I can click and change, but the keyboard always inputs english. I've had a bit more luck with chinput: running chinput, exporting the right env variable and starting konsole, I can Ctrl+space in it and type characters, and with Ctrl+Shift I can change the input method, but I haven't been able to find Zhu-Yin. So, since most of the documentation of these tools is in some Chinese, can someone please tell me the shortest path to have Zhu-Yin Taiwanese input method enabled as everywhere as possible? Ciao, Enrico -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: apt-get install leads to much trouble
On Sat, Feb 21, 2004 at 10:49:30PM +0100, Moritz Beller wrote: Isn't there any possibility of getting my apt-get work again? Thanks for all of you who have any clue and tell me. I can't recall the precise details, however I solved this issue by removing libxcursor-dev and xlibs-dev, including depending packages. In the end, it was only -dev libraries. I used dselect to do it, to be able to fiddle swiftly with the dependencies. Ciao, Enrico -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: New kernel crashing - how to test without corrupting disks?
On Sat, Jan 17, 2004 at 11:46:14PM -0600, Ramasubramanian Ramesh wrote: problem. I like to try a couple of combination. But like to minimize the risk of disk corruption. How do I prepare before issuing an kernel-killing command? Perhaps go into single user mode and unmount most of the partitions? Are are more things that can be done to prepate well for a freeze test? If you've compiled the magic sysrq support among the kernel hacking functions, Alt+SysRq+U mounts all filesystems read-only, and from there you're safe. Else, you can unmount filesystems or mount -o remount,ro them. In the latter case, you might need to go into single-user-mode to be able to remount your root file system read-only. However, if your system is not doing anything else, journaling file systems and a sync issued before the experiment should be just fine (mounting read-only is the sure bet, though). Ciao, Enrico -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: debian pwd
On Fri, Jan 16, 2004 at 08:37:57PM +0100, Arkel wrote: does anybody know how to get super user privilege when a normal user not supposed to If you find out how, please report it as a bug with severity critical against the involved package. Thank you. :) Ciao, Enrico -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: More 2.6.0 woes
On Sat, Jan 10, 2004 at 03:14:36PM -0800, Paul Johnson wrote: You might want to keep an eye on http://bugs.debian.org/cgi-bin/pkgreport.cgi?pkg=kernel-image-2.6.0-1-686-smp and possibly subscribe to the report when it arrives. Could you keep an eye out for it and post the URL when it does? It finally did: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=227391 On Tue, Jan 13, 2004 at 09:26:35PM +0100, Johannes Raspe wrote: since kernel 2.6.0 the modules evbug tsdev evdev are loaded with every boot. In case of evbug, which writes every keystroke in a logfile, this is very annoying. How can I stop the system from loading these modules ? You can have a look at the thread with subject More 2.6.0 woes for some tips, and at the bug report #227391 (http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=227391). So far, moving the module elsewhere seems to be the quickest work-around. Ciao, Enrico -- GPG key: 1024D/797EBFAB 2000-12-05 Enrico Zini [EMAIL PROTECTED] -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: More 2.6.0 woes
On Sat, Jan 10, 2004 at 03:14:36PM -0800, Paul Johnson wrote: http://bugs.debian.org/cgi-bin/pkgreport.cgi?pkg=kernel-image-2.6.0-1-686-smp and possibly subscribe to the report when it arrives. Could you keep an eye out for it and post the URL when it does? Sure. In the meantime, the bug has still not being processed, and I've sent the report directly to the maintainer to make him aware of the problem. The problem is quite serious: the keystrokes are viewable by default with dmesg by any user. Ciao, Enrico -- GPG key: 1024D/797EBFAB 2000-12-05 Enrico Zini [EMAIL PROTECTED] signature.asc Description: Digital signature
Re: 3COM Gig ethernet driver compile woes (source from 3COM/ASUS)
On Sun, Jan 11, 2004 at 01:07:49PM +0100, Robert Epprecht wrote: I got the source for 3COM gig ethernet card from Asus website and I am having compile issue. I have compiled it without real problems on Debian woody... I know, this will not help you much, but maybe it's good to know that it *can* work... If it's the same as the one I've just bought, kernels 2.6.x have native support for it as module sk98lin. It may be that older kernels support it, too. Ciao, Enrico -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: More 2.6.0 woes
On Sat, Jan 10, 2004 at 09:41:43AM +, Alan Chandler wrote: rmmod evbug stops that; to make it permanent, I removed evbug from the /lib/modules/version/modules.inputmap but I don't know if this is The Right Way. If it's not The Right Way, I'd like to be updated. I tried that, but its not permenant - I think depmod, run during boot recreates the modules.inputmap file. Another option I can see is probably renaming /lib/modules/version/kernel/drivers/input/evbug.ko into something else. This should always work :) Last night I've reported a bug about the evbug issue to kernel-image-2.6.0-1-686-smp[1] but it still hasn't appeared in the BTS. You might want to keep an eye on http://bugs.debian.org/cgi-bin/pkgreport.cgi?pkg=kernel-image-2.6.0-1-686-smp and possibly subscribe to the report when it arrives. Ciao, Enrico [1] This is what I use, I'm sorry I couldn't come up with better packages to report to -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: More 2.6.0 woes
On Sat, Jan 10, 2004 at 12:14:17AM +, Alan Chandler wrote: I am now getting loads of messages into syslog like below. The previous kernel 2.4.22 did report one spurious serial interupt but that was all. Any idea how to stop this flow of messages into syslog (as you can see, several a second) Jan 9 23:38:28 kanger kernel: evbug.c: Event. Dev: isa0060/serio0/input0, Type: 0, Code: 0, Value: 0 rmmod evbug stops that; to make it permanent, I removed evbug from the /lib/modules/version/modules.inputmap but I don't know if this is The Right Way. If it's not The Right Way, I'd like to be updated. In the meantime, since those log lines you posted could very well have been your keyboard events, be aware that you might have just exposed a password you typed :( Someone could tell me why make-kpkg created kernels load evbug at boot by default if it's compiled? That doesn't really seem a good idea to me. Not at all. Ciao, Enrico -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
What about some user analysis?
Publishing --- Yours truly, Enrico -- GPG key: 1024D/797EBFAB 2000-12-05 Enrico Zini [EMAIL PROTECTED] pgp0.pgp Description: PGP signature
Building a Xinerama kiosk
Hello, I've been asked to build a nice kiosk with a dual-head machine, and I'm having difficulties with a couple of issues. The kiosk: - One head, the normal monitor, will be used by users to browse a website. - The other head, a TV screen, will continuously show a fancy display of updating data, composed by a Galeon fullscreen window with a complex webpage full of java applets, reloading pages and stuff. The system so far is based on an updated Debian testing distribution. It works, dual head works, window positioning with --geometry usually works, we've prepared everything to run in a diskless system from a live CD, everybody is happy and excited, except for a couple of issues we can't solve cleanly: - I need to place a full screen Galeon window in the second head. Galeon ignores positioning with the --geometry switch (it only obeys window size indications). So far, the only way I've found to make it go there is to write some sawfish lisp code. Is there a better way to do that? - I need to restrict mouse cursor movement so that the user can never move it out of the first head and mess with the TV screen. Even better could be to restrict the mouse cursor so that it never moves out of the maximized galeon window in the first head. How do I do that? - I've managed to disable X killing trough C-A-Backspace. How do I disable vt switching (C-A-Fn)? Bye, Enrico -- GPG key: 1024D/797EBFAB 2000-12-05 Enrico Zini [EMAIL PROTECTED] -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: 2.4.18 netfilter NAT problem
On Wed, Feb 05, 2003 at 10:50:24AM +0100, Enrico Zini wrote: The server is a smal woody box with kernel 2.4.18 and an iptables configuration for firewall and NATting the other two home computers to access the net thru my ADSL connection. The problem is that a natted connection hangs after some data has been transmitted: if I ssh to an external host from the server, the session works fine for hours; if I ssh to an external host from a box behind the server, the connection hangs after I performed a few operations. I've solved the issue and I'm replying to this mail to add the solution to the list archives. The problem was solved lowering the MTU in PPPOE from the default value to 1440. My ISP probably changed some setting while I was away. Bye, Enrico -- GPG key: 1024D/797EBFAB 2000-12-05 Enrico Zini [EMAIL PROTECTED] -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
2.4.18 netfilter NAT problem
Hello, When I came back from vacation I've found that my small home server started having a problem with NAT/Masquerading. The server is a smal woody box with kernel 2.4.18 and an iptables configuration for firewall and NATting the other two home computers to access the net thru my ADSL connection. The problem is that a natted connection hangs after some data has been transmitted: if I ssh to an external host from the server, the session works fine for hours; if I ssh to an external host from a box behind the server, the connection hangs after I performed a few operations. For example, I can start mutt, but the connection hangs between when it writes scanning messages and when it displays the messages; I can't use reportbug, nor jabber. I've tried rebooting the server (maybe in this last month the netfilter code hit some race condition?), but with no luck. Where should I search now? I've attached my firewall script. Bye, Enrico #!/bin/sh DATAFILE=/etc/ppp/netdata case $1 in start) if [ ! -r $DATAFILE ] then echo $DATAFILE not found: failsafe stop 2 $0 stop exit 1 fi . $DATAFILE if [ -z $OUT_IP -o -z $OUT_IFACE ] then echo $DATAFILE did not export IP and interface: failsafe stop 2 $0 stop exit 1 fi # Example data read from $DATAFILE: # OUT_IFACE=ppp0 # OUT_IP=80.116.79.148 # OUT_PEER=192.168.100.1 # OUT_TAG= # Moduli speciali modprobe ip_conntrack modprobe ip_conntrack_ftp modprobe ip_conntrack_irc # Configurazione sysctl echo 1 /proc/sys/net/ipv4/conf/all/rp_filter iptables -P INPUT DROP iptables -P OUTPUT ACCEPT iptables -P FORWARD DROP iptables -F iptables -t mangle -F tc qdisc del dev ppp0 root # Apre loopback iptables -A INPUT -i lo -j ACCEPT # Apre eth0 iptables -A INPUT -i eth0 -j ACCEPT # Apre eth1 iptables -A INPUT -i eth1 -j ACCEPT # Attiva il masquerading iptables -t nat -A POSTROUTING -o $OUT_IFACE -j MASQUERADE ## Apre le porte locali # Deny totale degli spaccaballe for i in 65.116.32.194 do iptables -A INPUT -d $OUT_IP -i $OUT_IFACE -s $i -j DROP done # Deny degli spaccaballe che pingano a lungo for i in 165.91.1.102 193.204.5.62 do iptables -A INPUT -d $OUT_IP -i $OUT_IFACE -s $i --protocol icmp -j DROP done # Apre UDP per DDT #iptables -A INPUT -d $OUT_IP -i $OUT_IFACE --protocol udp --dport 1052 -j LOG --log-prefix Accept DDT: -m limit --limit 5/minute iptables -A INPUT -d $OUT_IP -i $OUT_IFACE --protocol udp --dport 1052 -j ACCEPT # Apre le risposte alle connessioni dall'interno all'esterno iptables -A INPUT -d $OUT_IP -i $OUT_IFACE -m state --state ESTABLISHED,RELATED -j ACCEPT # Apre le connessioni dall'esterno verso le porte locali che servono for PORT in 22 80 443 11412 do iptables -A INPUT -d $OUT_IP -i $OUT_IFACE -m state --state NEW --protocol tcp --dport $PORT -j LOG --log-prefix Accept TCP:$PORT: iptables -A INPUT -d $OUT_IP -i $OUT_IFACE -m state --state NEW --protocol tcp --dport $PORT -j ACCEPT done # Stessa cosa, ma loggando poco for PORT in 113 do iptables -A INPUT -d $OUT_IP -i $OUT_IFACE -m state --state NEW --protocol tcp --dport $PORT -j LOG --log-prefix Accept TCP:$PORT: -m limit --limit 5/minute iptables -A INPUT -d $OUT_IP -i $OUT_IFACE -m state --state NEW --protocol tcp --dport $PORT -j ACCEPT done # Stessa cosa, ma senza loggare for PORT in 6346 do iptables -A INPUT -d $OUT_IP -i $OUT_IFACE -m state --state NEW --protocol tcp --dport $PORT -j ACCEPT done # Accetta i multicast del peer iptables -A INPUT -i $OUT_IFACE -s $OUT_PEER -d 224.0.0.1 -j ACCEPT ## Apre il forward # Accetta dalle interfacce locali iptables -A FORWARD -i eth0 -j ACCEPT iptables -A
Re: best way to have gcc-3.2 on woody?
On Thu, Oct 31, 2002 at 11:34:03AM -0200, Klaus Imgrund wrote: That has nothing to do the compiler as far as I can see - mencoder works for me compiled with gcc 2.96 - no problem.The guys at the mplayer lists Fine for that. But suppose I'm doing some C++ development in woody, and that I've written some piece of legal code using some advanced C++ feature (like throwing exceptions created using multiple inheritance) that only g++3.2 gets right. In that case, I'd like to have g++3.2 available, but I wouldn't want to go through the assle to run an unstable system. What would then be the best way to have gcc-3.2 on wooy? Bye, Enrico -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Mustek MDC 3000
Hello, I've been proposed a discount offer for a Mustek MDC 3000 digital camera, but I haven't found any mention about Linux support for it, neither in Mustek web site nor in some searches in Google. The MDC 800 model seems to be supported, but for the MDC 3000 there is no mention. Is there somebody with this camera that can tell me how well it is supported, and if it is supported at all? If there is no support, I'll be happy to let them keep it. Bye, Enrico -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Apache and php segfaulting
On Tue, Nov 27, 2001 at 10:20:22AM +0100, [EMAIL PROTECTED] wrote: used also for php development, and since the bug has already been reported (more or less in bugs 119688, 121096, 109454, 116856, 120492), first I tried to recompile apache and php from source, but I had no luck; then I wanted to just downgrade to the previous version, but I don't have the packages anymore and I can't find them in ftp.debian.org nor any mirrors I've tried. I've been able to trace the problem to php4-recode. Removing it from the system solved the problem. A bug was already filed against it (114298 and 114826), but no mention of it was found neither in apache nor in php4, making it impossible to find it since the symptom was not related to the responsible package. Should I now post additional informations to all these 5 bugs mentioning it? Bye, Enrico
Re: Random mysql index corruption on Dell Poweredge 2450
On Mon, Oct 29, 2001 at 05:01:36PM -0500, Atul Mukker. wrote: I would recommend you to immediately move to 1.18 megaraid driver. Do you still see the issue? No: it seems to be fixed now, thank you very, very much. I waited a little to post the answer to give the machine a little time to stress itself. Since when I compiled in the 1.18 megaraid driver, everything has been working 24/24 with no problems, for at least a week. I'm posting this to the lists even if some time has passed, to avoid my problem to remain unanswered, for the curious and the list archives. Bye, and thanks again! Enrico -- GPG key: 1024D/797EBFAB 2000-12-05 Enrico Zini (Unibo) [EMAIL PROTECTED]
Random mysql index corruption on Dell Poweredge 2450
Hello! I'm posting this to three lists because I can't track down the problem to a single cause. We are setting up a Dell Poweredge 2450 to run the central MySQL database of our crowded web site. If we raise the load on the database by moving some services from the current server to the new one, we start experiencing random database indexes corruption. Sadly, we've been unable so far to track down the issue to a single query or some reproducible sequence of events. I'll try to include here all the details we have. The MySQL errors are all 127 Record-file is crashed or 134 Record was already deleted (or record file crashed). Here are some examples: Got error 134 from table handler executing query UPDATE Delayed2 SET [...] WHERE ID=1 AND Tipo=1 (err: 1030) Got error 127 from table handler executing query SELECT Simbolo, Prezzo, UNIX_TIMESTAMP(Ora), TotVol, NT FROM Realtime WHERE ID=23 AND Tipo = 5 (err: 1030) This happens both with the Debian MySQL version 3.23.36-6 (from testing) and with the 3.23.43 precompiled binaries downloaded from the www.mysql.com web site. We can fix the tables with myisamchk, but after some time (ranging from a couple of hours to a couple of days) the database get corrupted again. The system is a Debian testing with kernel 2.4.12 (an upgrade to 2.4.13 is planned). File system is ext2. Since I don't know if it's a MySQL fault or a hardware fault, I also include hardware and driver details: dmesg log of AIC7XXX and megaraid initialization: SCSI subsystem driver Revision: 1.00 scsi0 : Adaptec AIC7XXX EISA/VLB/PCI SCSI HBA DRIVER, Rev 6.2.1 Adaptec aic7899 Ultra160 SCSI adapter aic7899: Ultra160 Wide Channel A, SCSI Id=7, 32/255 SCBs scsi1 : Adaptec AIC7XXX EISA/VLB/PCI SCSI HBA DRIVER, Rev 6.2.1 Adaptec aic7899 Ultra160 SCSI adapter aic7899: Ultra160 Wide Channel B, SCSI Id=7, 32/255 SCBs megaraid: v1.17a (Release Date: Fri Jul 13 18:44:01 EDT 2001) megaraid: found 0x8086:0x1960:idx 0:bus 0:slot 2:func 1 scsi2 : Found a MegaRAID controller at 0xe0808000, IRQ: 20 megaraid: [1.01:1p00] detected 1 logical drives megaraid: channel[1] is raid. megaraid: channel[2] is raid. scsi2 : AMI MegaRAID 1.01 254 commands 16 targs 2 chans 8 luns scsi2: scanning channel 1 for devices. Vendor: DELL Model: 1x4 U2W SCSI BP Rev: 1.16 Type: Processor ANSI SCSI revision: 02 scsi2: scanning channel 2 for devices. scsi2: scanning virtual channel for logical drives. Vendor: MegaRAID Model: LD0 RAID5 17136R Rev: 1.01 Type: Direct-Access ANSI SCSI revision: 02 Attached scsi disk sda at scsi2, channel 2, id 0, lun 0 SCSI device sda: 35094528 512-byte hdwr sectors (17968 MB) Partition check: /dev/scsi/host2/bus2/target0/lun0: p1 p2 p3 p5 p6 Some /proc stats: service:~# cat /proc/megaraid/0/config Controller Type: 438/466/467/471/493 Base = e0808000, Irq = 20, Logical Drives = 1, Channels = 2 Version =1.01:1p00, DRAM = 128Mb Controller Queue Depth = 254, Driver Queue Depth = 126 service:~# cat /proc/megaraid/0/stat Statistical Information for this controller Interrupts Collected = 2065925 Logical Drive 0: Reads Issued = 136738, Writes Issued = 1929155 Sectors Read = 2611796, Sectors Written = 25004064 service:~# cat /proc/megaraid/0/mailbox Contents of Mail Box Structure Fw Command = 0x02 Cmd Sequence = 0x66 No of Sectors= 0008 LBA = 0x16c5c8a DTA = 0x0230f000 Logical Drive= 0x00 No of SG Elmt= 0x00 Busy = 0 Status = 0x00 service:~# cat /proc/megaraid/0/status TBD Sadly, I don't know if I should upgrade some firmware, nor I know what releases our firmware are, since the machine is hosted in a farm ~100Km from here and it's hard for me to track boot messages and run boot floppies. Is there a way to know that from Linux? Do you have any hints for me to try to solve this problem? Bye, Enrico -- GPG key: 1024D/797EBFAB 2000-12-05 Enrico Zini (Unibo) [EMAIL PROTECTED]
Adaptec problems with 2.4.6?
Hello! I'm having some problems with kernel 2.4.6 and an Adaptec AIC-7890 controller. The first one is that it shares the interrupt with the sound card. This should not be an issue, since they are PCI devices, and it never gave any problems, except lately, when if I play mp3 and make some heavy activity on the SCSI bus (scan a photo, write a CD, havily use the hard disk) while playing MP3s, the computer freezes hard. Here are the interrupt assignments: CPU0 CPU1 0: 89648 80663IO-APIC-edge timer 1: 4836 4120IO-APIC-edge keyboard 2: 0 0 XT-PIC cascade 3: 3760 3292IO-APIC-edge serial 5: 0 0IO-APIC-edge SoundBlaster 8: 0 1IO-APIC-edge rtc 9: 0 0IO-APIC-edge acpi 10: 25675 25669 IO-APIC-level aic7xxx, EMU10K1 11: 97137 96693 IO-APIC-level usb-uhci, eth0 12: 1443 1110IO-APIC-edge PS/2 Mouse 14:742507IO-APIC-edge ide0 15: 480721 609965IO-APIC-edge ide1 NMI: 0 0 LOC: 170231 170248 ERR: 0 MIS: 0 The other problem is when I make heavy use of my Plextor CD-R drive: here's what happens from time to time: Jul 16 23:06:06 marvin kernel: scsi0:0:0:0: Attempting to queue an ABORT message Jul 16 23:06:06 marvin kernel: scsi0:0:0:0: Device is active, asserting ATN Jul 16 23:06:06 marvin kernel: Recovery code sleeping Jul 16 23:06:11 marvin kernel: Recovery code awake Jul 16 23:06:11 marvin kernel: Timer Expired Jul 16 23:06:11 marvin kernel: aic7xxx_abort returns 8195 Jul 16 23:06:11 marvin kernel: scsi0:0:6:0: Attempting to queue an ABORT message Jul 16 23:06:11 marvin kernel: scsi0:0:6:0: Cmd aborted from QINFIFO Jul 16 23:06:11 marvin kernel: aic7xxx_abort returns 8194 Jul 16 23:06:21 marvin kernel: scsi0:0:6:0: Attempting to queue an ABORT message Jul 16 23:06:21 marvin kernel: scsi0:0:6:0: Cmd aborted from QINFIFO Jul 16 23:06:21 marvin kernel: aic7xxx_abort returns 8194 Jul 16 23:06:21 marvin kernel: scsi0:0:0:0: Attempting to queue a TARGET RESET message Jul 16 23:06:27 marvin kernel: aic7xxx_dev_reset returns 8195 Jul 16 23:06:27 marvin kernel: scsi0:0:6:0: Attempting to queue a TARGET RESET message Jul 16 23:06:27 marvin kernel: aic7xxx_dev_reset returns 8195 Jul 16 23:06:27 marvin kernel: Recovery SCB completes Jul 16 23:06:27 marvin kernel: Device 0b:00 not ready. Jul 16 23:06:27 marvin kernel: I/O error: dev 0b:00, sector 984096 Jul 16 23:06:27 marvin kernel: Device not ready. Make sure there is a disc in the drive. Jul 16 23:06:27 marvin kernel: Device 0b:00 not ready. Jul 16 23:06:27 marvin kernel: I/O error: dev 0b:00, sector 984096 Jul 16 23:06:27 marvin kernel: Device not ready. Make sure there is a disc in the drive. With the kernel 2.4.5 I had problems, too, but the system just chose to hang forever all disk activity instead of reporting errors. Two questions: 1) Do some people know something of similar problems? 2) Since this is not so debian-specific, is there a more appropriate mailing list to report this? TIA, Enrico -- GPG key: 1024D/797EBFAB 2000-12-05 Enrico Zini (Unibo) [EMAIL PROTECTED]
Marking bad blocks
Hello! Using badblocks I've found some bad blocks on one of my hard disks, and I'd like to mark them bad so that Linux will avoid to use them. The number of bad blocks I've found is low, but they are scattered on many disk partitions: some of them are formatted ext2, some reiserfs and one is used as swap. For the ext2 partition the problem is easily solved, since e2fsck has the -l switch to be used exactly for that purpose. The problem are reiserfs and the swap partition. The swap partition could be solved, too, by running mkswap with the -c option. The only problem with that is that I'll have to wait for another full badblock check, and since I already have the bad block list, I'd like to just pass that list to mkswap to make it quicker. If there isn't any way of doing it, I'll use mkswap -c and be fine. We remain with reierfs. Reiserfsck has no options regarding bad blocks, and I can't remember any other reiserfs utilities that could help with that. Is there a way of doing it? Now I'll just wait for news here, while trying to figure out why an IBM-DJNA-352500 three months old which passes all ide-smart internal OnLine and OffLine tests should turn out to have bad blocks at all. And doing backups, of course. Bye, Enrico -- GPG public key available on finger -l [EMAIL PROTECTED]
Putting computer on standby or sleep
Hello, I would like to know if (and how) I can make my computer suspend so that all (most of) is powered down but I can resume it and have it in the same state it had before the suspend. Something like portables do, but with a workstation. The system is a dual PII-450 on a Supermicro P6DGU 440GX motherboard on an ATX power supply, and runs Debian unstable updated every night with kernel release 2.4.1. APM is disabled on multiprocessor systems, ACPI works. Thanks in advance, Enrico -- GPG public key available on finger -l [EMAIL PROTECTED]
Upgrade to g++ 2.95.3 in testing
Hello! I upgraded to gcc and g++ 2.95.3 in my testing system and I cannot compile c++ anymore. The problem is it can't find libstdc++. This also made autoconf configure scripts stop to run stating that the c++ compiler cannot create executables. I solved removing the old g++ and gcc packages (keeping only g++-2.95 and gcc-2.95. Funny, the old g++ and gcc packages were also version 2.95 (2.95-2)), and making two symlinks in /usr/local/bin pointing gcc to gcc-2.95 and g++ to g++-2.95. Right now, the only gcc and g++ packages I have are the -2.95 ones. IMHO the links to gcc and g++ should have been created automagically; should I file a bug report? Is there something wrong on my system? The system is an up-to-the-last-minute Debian testing with mysql and php4 from unstable. TIA, Enrico
On bug reporting
Hello! I would like to know if there are guidelines for bug reporting: I use debian unstable, and I find bugs, but I don't report most of them for many reasons. I've never discussed these reasons with anyone. Now I would like to. I identified until now three kind of such unreported bugs: - Evident bugs I work mainly offline, so I don't have constant access to bugs.debian.org, and usually can't know when a bug has already been reported. When I find a bug in a package that is likely to already have been reported, I don't do it to avoid the possibility of flooding the bug database with reduntant reports. An example of this is when tar changed behaviour of the 'I' switch, or when a package can't be upgraded due to missing dependancies (like glademm, erlang and wordtrans are now), or when a package keeps redoing the same debconf questions on most updates. I only report these bugs when I see the problem persisting after a week or two. Should I always report it, instead? - Pedantic bugs Sometimes I notice something that should be different, but I don't know if it is to be considered as a bug, and I don't report it. For example, packages like skipstone, or powermanga or lxdoom appear in the Debian menus but not in the Gnome menus (is there a policy for Gnome menus?), or alsa doesn't load oss-emulation modules on demand with kernel 2.4.0+devfs (are packages supposed to correctly cope with devfs?) Should I be on the safe side and risk reporting a non-bug, or should I be on the other safe side and risk non-reporting a bug? - Non-Debian small bugs Sometimes I find small bugs that are clearly to be reported upstream, like wrong i18n translations (many, many, many of them), usability quirks or even usability suggestions, that I don't report upstream because I can't quickly find the address of the right person, and there's not an handy tool like `bug' for them, or there is, but is online, or there is, offline, but I don't know how to launch it, because every program has its own. It would be very handy to file a `whishlist' bug to the Debian BTS, knowing they are eventually reported upstream, but I don't do it to avoid flooding the BTS with non-debian-related items. Could the Debian BTS be used as a proxy service to upstream mantainers (considering that with debian-native packages it already is supposed to collect these kind of reports anyway) or we should wait for the development of a different common bug reporting system for such little issues? I would like to hear your opinion on this behaviours, to get out of Doubt into The Right Way (TM) to report bugs. Bye! Enrico -- GPG public key available on finger -l [EMAIL PROTECTED]
Debian potato on an EGA display
Hello! I'm running Debian on a Celeron system with an EGA video board. Everything is fine except the cursor is displayed 2 characters to the right of where it should be. For example, if I type 'foo', I get foo _ but if I press backspace I get fo _ To solve this inconvenient I found a valid work-around with the soft cursor feature of the Linux kernel, that allows to use a software cursor instead of the hardware one. You can see Documentation/VGA-softcursor.txt in the Linux kernel for more informations. The software cursor looks ok. To enable the software cursor I simply run the command echo -e \033[?17;0;64c Now, with slink everything was fine, but, after upgrading to potato, vim started restoring the hardware cursor on program startup, when coming back from a shell, when coming back from suspend and leaving the hardware cursor on exit. I have no idea on how to stop this; changing editor is not an option, so I would like to know if someone can help me fixing vim behaviour or fixing the hardware cursor under EGA, whichever is easier. Thanks in advance, Enrico
Re: Using Mylex RAID ctrl with debian 2.2
On Tue, 1 Feb 2000, Olivier CARRERE wrote: I tried to compile a new 2.2.14 kernel (the file named 'linux' I suppose ;)) with Mylex and ramdisk support and replace the one included in the standard install, and all seemed to work fine *but* at the stage of installing kernel and modules, it couldn't access rescue.bin; I tried to access it via floppy, with no result. Is there something to change in rescue.bin also, or somewhere else? Hello! I installed a slink system on a Mylex AxxeleRAID 250 only sustem just a couple of months ago, and I can help; in this case, however, it just seems like you didn't compile the minix file system in your 2.2.14 kernel. Feel free to ask for more informations. Read you soon! Enrico
Matrox Rainbow Runner
Hello! I was planning to buy a new computer, and include some tv-tuner/grabber card; I saw that Matrox makes an edition of the G400 (called Marvel or something like that) with a TV-tuner and video acquisition/output; it seems quite a nice thing to buy, and the price is not that bad. I didn't find any mention of it in the Kernel v.2.2.14 video4linux documentation, though, and I would like to know if this card is somewhat supported on Linux or not. Read you soon! Enrico -- GPG public key available on finger -l [EMAIL PROTECTED]
Building an emergency repair CD
Hello! I'm building an emergency repair CD for the servers we sell. The idea is that if the system ever goes so crazy it can't even boot, you put the CD in, reset the computer and the CD boots and overwrites anything with a copy of the working system. I already managed to build the boot environment, with an initrd image whose purpose is to mount a bigger media (eg. CD-Rom) where it could find a minix image with the rest of the boot system, and mount it with loopback on /usr; all of this seem to work fine. Now I just have to write the barbarian overwriting bulldozer. I would have liked to do a zcat image.gz /dev/sda, but the image should be on a CD-Rom, that isn't big enough to store a complete disk image. Until now, I always kept the base Linux system (without /var and /home) in a single partition of about 600Mb, so that it can always fit on a CD, but to restore the image of a single partition I have to restore the partition table first. How can I backup and restore the partition table? I was thinking about using dd if=/dev/sda of=saved.partition count=1 and vice versa, but i wouldn't save logical partitions. I then thought about using sfdisk -d saved.partition and sfdisk saved.partition, but I need to be sure that: 1) It recreates a partition table exactly equals to the original 2) It supports Mylex AcceleRAID logical disks ...so the two questions of this post are: 1) Does sfdisk -d | sfdisk _always_ rewrite _exactly_ what was before? 2) Does sfdisk support Mylex AcceleRAID logical disks? As a side note, this system could evolve in a boot system to make partitions dumps and restores, be network aware, be able to apply patches to the restored partition (one of the goals is to have a single common CD dump of your Windows 95 clients that you stick in, type the name of the PC, and it restores the clean system and appropriately sets the client IP address and system name). Anybody who is interested in working on such a system please contact me; GNU Public License, of course. Read you soon! Enrico -- GPG public key available on finger -l [EMAIL PROTECTED]
Considering PAM [Was: How can I change a password from a script?]
On Fri, Dec 03, 1999 at 04:55:27PM -0500, Joe Block wrote: Now we have shadow passwords, MD5 hashes, NIS, LDAP, PAM... wow! It's fantastic, but I need something that knows how to change passwords on my system, because I don't. Check out chpasswd - you can pass it a list of username:newpassword pairs. Does it integrate well with PAM? I would not want to have to drop it because I learned to use PAM and changed the password database to something different than passwd. Yesterday evening I also had a look at PAM and it seemed not possible to me to change password with it in a non-interactive way, and I also had the impression it is not possible to use web pages to allow a user to change password, since PAM chatting can't span different CGI requests. The telnet link, however is a good idea I wasn't thinking about, though I would like to present to users something more attractive than a monochrome terminal window with english messages. Changing the shell to /usr/bin/passwd, would also require them to type their old password twice, which is a burden it would be hard for me to explain to them. Thank you! Read you soon! Enrico -- GPG public key available on finger -l [EMAIL PROTECTED]
Re: How can I change a password from a script?
On Fri, Dec 03, 1999 at 04:57:51PM -0500, Nick Cabatoff wrote: How do you prevent people from cracking passwords via your web page? I'm still looking for a secure way to accept passwords via HTML - even with SSL, from what I understand the available authentication stuff isn't suitable for use with /etc/passwd. It's too easy for someone to write a brute-force password scanner that won't leave traces. I don't have problems like this because it's an intranet server and all connections from outside are blocked by ipchains rules, since they haven't any reason to be; you could, however, use SSL just to have an encrypted channel, and provide a page to authenticate the user, set a SSL-only cookie with an authorization token and use that for the following session. Password scanners leave traces in your web logs, and you could block too many consecutive attempts from the same IP, while SSL is protecting user data and authorization tokens. I don't see any problems with this; if you do, however, let me know, because that's what I would use, and I'm probably going to have a need for it in a couple of months. :-) Read you soon! Enrico -- GPG public key available on finger -l [EMAIL PROTECTED]
What hardware is good for Debian servers?
Hello! I had many problems with Asus, and it was one of the brands people said to be good. During my quest to solve many of these problems, I discovered that Asus is likely to have problems, or just glitches, that make you loose time where you shouldn't. Now, since there are many servers to build and configure awaiting for us, I would like to build some knowledge on what to look for and what to try to avoid. So, suppose I ask you to build a fast and reliable Debian server, something that does file serving with samba, dials internet on demand via diald, spools a couple of printers, serves faxes via Hylafax, processes e-mail from, to and within the intranet and performs backups of what's on it. What brand and type of motherboard, raid controller, CPU, modem and backup system would you use, to be able to install Debian on it without risking to go mad after some hardware flaw? What are good brands of motherboards? Asus? Intel? Soyo? Tyan? MSI? VIA? Others? What are the ones to avoid like death? I'd also would like to find multiprocessor motherboards without expensive (and Debian unfriendly) Adaptec cards integrated on them, since high end servers are likely to mount a custom raid controller like the Mylex AcceleRAID and don't have a need for them. Is there some internet site where to look for this kind of informations? I would like to avoid marketing hype and read about tests and real experiences, possibly taking Linux and Debian into account. Read you soon! Enrico -- GPG public key available on finger -l [EMAIL PROTECTED]
How can I change a password from a script?
Hello! I'm root, I'm on a Debian Slink or on a Debian Potato, and I would like to present my intranet users a web page to change their passwords. It would be easy to do, if I just had to work with the good old /etc/passwd database: read the old password, verify it, encrypt the new one and change it in passwd. Now we have shadow passwords, MD5 hashes, NIS, LDAP, PAM... wow! It's fantastic, but I need something that knows how to change passwords on my system, because I don't. I would like to call passwd from my setuid root CGI (in which all security precautions would have been taken), feed him the new password and let him to whatever it pleases, but it could complain about passwords being too weak. I don't need those checks: I could call a password checker from the CGI to complain to the user in a web page in case I needed to, but I want a way to say set bob's password to '42bob69' and have it done even if bob's password is 'a', or an empty string, in whatever way the system is configured to do it. passwd had a switch (-o, if I can recall it well) for root telling it not to complain about weak passwords, but now it's gone. I used it to add a password to my home user account when connecting to Internet, and removing it on disconnect. How can I do it cleanly on a Debian system? Is there a PAM call for it? Is it possible at all? It's also a problem changing passwords with Samba, since it uses a chat script with passwd to do the job, but has problems reporting if and why the password was or was not updated. Thanks in advance. Read you soon! Enrico -- GPG public key available on finger -l [EMAIL PROTECTED]
Problems with Asus P2B-DS/350 and double PIII 450
Hello! I have a dual Pentium III 450 system with an Asus P2B-DS/350 main board, in which I removed everything but the floppy drive and the IDE CD-Rom, master on the second channel. The standard Debian boot disk hangs probing the Adaptec SCSI card bundled in the main board; I tried making custom boot kernels with 2.2.13 and everything hangs asking if I have a color monitor. I'm trying to build the smallest possible 2.2.13 boot kernel for that server, but I have no idea what could be the cause. Is there some boot problem HOWTO or some other source of information about problems booting linux or hardware problems with Linux in general? Windows NT boots, installs and seems to work fine. Is there some known problem with Linux and this Asus mainboard, or with double Pentium III? TIA, Enrico
Re: quickie.
On Tue, Nov 23, 1999 at 10:16:26AM +1100, Marc-Adrian Napoli wrote: sure someone will know the answer to this. whoever does, could you please reply to me off the list. i'm planning on building a new server. i want to raid 5 the hard drives for security, reliability etc. can i mount a scsi raid 5 system onto the / partition? will kernel 2.2 do it? Yes to all, but it won't perhaps be an automatic install: there's a custom LILO that boots from a Mylex DAC960 logical (raid) disk in the Linux-Mylex pages at http://www.dandelion.com/Linux/. I'm trying to do exactly that, but I'm having other problems and I can't still tell you how to do it and how great it works. Read you soon! Enrico -- GPG public key available on finger -l [EMAIL PROTECTED]
Botting a Mylex DAC960 RAID controller
Hello! I'm trying to install Debian slink on a double pentium 2 server with a Mylex DAC960 RAID controller with 3 IBM UW2 9Gb disks. After compiling a custom kernel for the boot floppy and making it boot and see the DAC960, I found that the Debian installation hangs: it gets stuck asking me if I have a color display (the cursor stops moving). The server has 256Mb RAM, but I tried to boot with Linux mem=256 and it dumped registers. Syslinux, instead, complains that low memory is lower than about 600Kb, and asks me to boot it holding Ctrl. Is there some custom boot floppy to try with? Has someone installed Linux in a DAC960-only environment? Read you soon! Enrico
Re: Botting a Mylex DAC960 RAID controller
On Mon, 22 Nov 1999, Nathan E Norman wrote: Try `mem=256M' instead. Please let me now how the Mylex performs ... No luck: the installation keeps hanging asking about colors. I can't see why syslinux reports having less than 608Kb of base memory. Well, I *can* see why, since the base memory is in fact less than 608Kb (I tried to also boot with a dos floppy, and the results are the same), but I don't know who's eating that ram. I'm trying with a RedHat 6 (The server will need Debian, anyhow), and I had more luck: it didn't require a custom kernel and had a module for the DAC960. It seems to hang, however, when formatting the drives, when a disk is reported dead. Perhaps we have also a problem with the disks, we'll see. Redhat installation hangs when formatting/mounting. I'd need more data on that base memory requirement complained by syslinux. Anyway, DAC960 seems to be a good controller, and Linux support seems to be the good Linux support I always found with Mylex devices. I'll be able to tell more that impressions when I'll be able to have it running. Read you soon! Enrico
Re: Botting a Mylex DAC960 RAID controller
On Mon, 22 Nov 1999, Jean-Yves BARBIER wrote: Try `mem=256M' instead. Please let me now how the Mylex performs ... No luck: the installation keeps hanging asking about colors. Try `mem=255M' instead. Fine, I'll give it a try tomorrow. Even if it starts, anyway, the problem remains: raid disks on DAC960 aren't neither EIDE nor SCSI disks, and /dev/rd* devices are not created in the normal boot floppies; a quick hack I thought of is to change the major numbers of /dev/sda* devices, so that Debian installation will believe it's coping with normal scsi devices; I'm not satisfied with it, however, since normal devices for it should be named something like /dev/rd/c0d0*. No, that's not on a Sun. BTW RedHat 6.0 used that c0d0* convention right, so their boot disks can be worth a glance for `ideas' O:-) Another problem is the Asus mainboard has an Adaptec UW SCSI controller built-in that has been disabled on the BIOS configuration. Sadly, just the SCSI bios has been disabled from that board, and the Linux Adaptec driver hangs when it finds its controller. The only way I found to avoid Debian boot disks try to load Adaptec drivers, is to make a new kernel. No use were the adaptec boot disks from http://www.debian.org/~adric/aic7xxx/, and even RH6 hung when trying to initialize it: RH6 can only be installed in 'expert' mode, on that machine. Not that I'm afraid of expert modes, anyway. Read you soon! Enrico
Re: find | egrep
On Mon, 11 Oct 1999 [EMAIL PROTECTED] wrote:
Not a debian question, moreso a generic Unix question.. I need to be able
to use find and egrep to scan a directory which has more than 3000 files in
it. I read in the manual ie man egrep that the lines are limited to 2048
bytes. The error i get is /usr/bin/find: 0403-027 The parameter list is
too long. Is there any way i can work around this?
There's more than one way to do it(tm):
find options | xargs egrep pattern
find options | while read a; do egrep pattern $a; done
[...]
find options -exec egrep pattern {} \;
The latter being the preferred one.
Bye! Enrico
Mount as many partitions as possible read only (what do I do with /etc?)
Hello! In order to improve reliability, reduce fsck times, increase foolproofness, preserve the wildlife and so on (on a Debian Slink server with a Very Interruptible Power Supply (tm)), I wish to mount read only as many partitions as I can. My goal would be to have everything read only except /var, /home and /tmp. It's trivial to have /usr read-only, but I have problems with the root partition: - It has to contain /etc since it's required at early boot time to run rc scripts - /etc has to be read-write since you have to be able to change passwords, add and remove users, configure samba shares via swat, update adjtime, mtab and so on. What's The Right Way (tm) to have / mounted read-only? TIA, Enrico -- GPG public key available on finger -l [EMAIL PROTECTED]
Re: Mount as many partitions as possible read only (what do I do with /etc?)
On Wed, Oct 06, 1999 at 08:15:14PM +1000, Brian May wrote: What's The Right Way (tm) to have / mounted read-only? My diskless package (I just uploaded 0.3.2 to master today) does what you want (but for diskless systems). Sadly, my system is the network server: the one and only system that cannot be diskless. It diverts /sbin/init to /sbin/init.orig, and installs a shell script in place of /sbin/init. This shell script mounts /etc from an NFS server. This can prove to be a good idea, mounting /etc from a rw partition. It will indeed require some work to make it a robust script, since the etc partition will have to be fscked, action should be taken if fsck fails, etc. ( :) ), just like in checkroot.sh I thought this was an old question with an obvious solution, but it's turning out to be nontrivial. We have a policy allowing /usr to be mounted read-only, have we nothing for / ? The quick-and-easy solution seems to be the sync mount option suggested by Peter, and if this is a new issue I'll file a wishlist bug agains debian-policy. Bye and thanks! Enrico -- GPG public key available on finger -l [EMAIL PROTECTED]
Re: Mount as many partitions as possible read only (what do I do with /etc?)
On Wed, Oct 06, 1999 at 11:07:33AM +0200, Peter Palfrader wrote: I did mount / with the sync option on a server of mine. So the danger of data loss is reduced a bit. Since / is not that big a fsck is quite fast. Thanks! This seems the best available solution for my system. Bye! Enrico -- GPG public key available on finger -l [EMAIL PROTECTED]
Re: Modem connection speed
On 8 Sep 1999, John Hasler wrote: This is my chatscript (I added ECHO and REPORT to the original pppconfig generated script, but nothing changed): connect /usr/sbin/chat -v -f /etc/chatscripts/mclink2 Change that to connect /usr/sbin/chat -v -r report_file -f /etc/chatscripts/mclink2 where 'report_file' is the file you want the speed written to. Opening /var/log/connection_speeds... chat: Sep 09 16:23:04 CONNECT 50666/ARQ/V90/LAPM/V42BIS Closing /var/log/connection_speeds. Opening /var/log/connection_speeds... chat: Sep 09 10:55:38 CONNECT 50666/ARQ/V90/LAPM/V42BIS Closing /var/log/connection_speeds. Fine. Now, what if I wanted it to be logged through syslogd, possibly without those Opening and Closing lines? (and, BTW, another thing I would like to know is why doesn't it CONNECT 56000 :( ) TYA! Enrico
Monitor various system informations
Hello! I am looking for a way to monitor various system informations for the debian servers we install. An example of informations to monitor is: - Network throughput - Uptime - Mail queue size - If I'm sending mail, what's the sending progress (1) - If I'm retrieving mail (fetchmail), what's the progress (1) - Is the modem connected? At what speed? What's its current throughput? - User specific disk and mail stats (requires user autentication) ...and the like. The goal is to have various ways to read the stats (web page on a server's http server, gnome application, win32 application). The first step is to write ad application who's able to gather all the required information and to share it in a consistent way. That said, the questions are: 1) SNMP seems to be a good way to share these informations, but I have to learn it: before starting, is it the right way to go? 2) LDAP instead? Other services? 3) Is there a way to interrogate fetchmail about how's going? If not, is someone working on it? Who? I'll give help. If nobody's working on it, I will. 4) Is there a way to interrogate exim and pppd about how's going? Luckily diald already has a way to gather informations. TIA! Enrico (1) Progress is something like: sending message 3/7, 23Kb/85Kb (27%), 104Kb/541Kb (19%) total -- GPG public key available on finger -l [EMAIL PROTECTED]
Can't install gnome-apt and console-apt
Here's the screen dump: marvin:~# apt-get install gnome-apt console-apt Reading Package Lists... Done Building Dependency Tree... Done Some packages could not be installed. This may mean that you have requested an impossible situation or if you are using the unstable distribution that some required packages have not yet been created or been moved out of Incoming. The following information may help to resolve the situation: Sorry, but the following packages have unmet dependencies: gnome-apt: Depends: libapt-pkg2.5 but it is not installable E: Sorry, broken packages marvin:~# That's true: libapt-pkg2.5 isn't there. Apt is right, something is wrong. Sigh. Ciao! Enrico -- GPG public key available on finger -l [EMAIL PROTECTED]
Modem connection speed
Hello! I would like to see the modem connection speed after the CONNECT message in my logs; I use ppp 2.3.5-2 on a slink box. * This is my chatscript (I added ECHO and REPORT to the original pppconfig generated script, but nothing changed): ABORT BUSY ABORT NO CARRIER ABORT VOICE ABORT NO DIAL TONE ABORT NO ANSWER ATZ OK ATP1M1X4 OK ATDP0,T147580001 ECHO ON REPORT CONNECT CONNECT \d\c * This is the chat invocation line (generated by pppconfig): connect /usr/sbin/chat -v -f /etc/chatscripts/mclink2 #pppconfig_connect * This is the log: Sep 8 19:13:42 eddie pppd[259]: pppd 2.3.5 started by root, uid 0 Sep 8 19:13:43 eddie chat[260]: abort on (BUSY) Sep 8 19:13:43 eddie chat[260]: abort on (NO CARRIER) Sep 8 19:13:43 eddie chat[260]: abort on (VOICE) Sep 8 19:13:43 eddie chat[260]: abort on (NO DIAL TONE) Sep 8 19:13:43 eddie chat[260]: abort on (NO ANSWER) Sep 8 19:13:43 eddie chat[260]: send (ATZ^M) Sep 8 19:13:43 eddie chat[260]: expect (OK) Sep 8 19:13:43 eddie chat[260]: ATZ^M^M Sep 8 19:13:43 eddie chat[260]: OK Sep 8 19:13:43 eddie chat[260]: -- got it Sep 8 19:13:43 eddie chat[260]: send (ATP1M1X4^M) Sep 8 19:13:44 eddie chat[260]: expect (OK) Sep 8 19:13:44 eddie chat[260]: ^M Sep 8 19:13:44 eddie chat[260]: ATP1M1X4^M^M Sep 8 19:13:44 eddie chat[260]: OK Sep 8 19:13:44 eddie chat[260]: -- got it Sep 8 19:13:44 eddie chat[260]: send (ATDP0,T147580001^M) Sep 8 19:13:44 eddie chat[260]: report (CONNECT) Sep 8 19:13:44 eddie chat[260]: expect (CONNECT) Sep 8 19:13:44 eddie chat[260]: ^M Sep 8 19:14:19 eddie chat[260]: ATDP0,T147580001^M^M Sep 8 19:14:19 eddie chat[260]: CONNECT Sep 8 19:14:19 eddie chat[260]: -- got it Sep 8 19:14:19 eddie chat[260]: send (\d) Sep 8 19:14:20 eddie pppd[259]: Serial connection established. What's wrong? TYA! Enrico -- GPG public key available on finger -l [EMAIL PROTECTED]
Re: Modem connection speed
On Wed, 8 Sep 1999, Hasso Tepper wrote: I would like to see the modem connection speed after the CONNECT message in my logs; I use ppp 2.3.5-2 on a slink box. Depends which modem do you use. Look at your modems manual to see how to enable connection speed reporting. ATX4, and as you can see I added it to the modem init string in the chatscript. To be even more pedantic, I tried the same commands via minicom, and the connection speed _does_ appear after CONNECT. TIA, Enrico
printing on a pcl printer is slow!
Hello! I have a Debian box that, among other things, shares an HP LaserJet 5 connected to a parallel (ECP/EPP) port for a mixed win95/linux environment. If I print from win95 everything is fine, but printing from Linux (that involves using GhostScript to convert PostScript to PCL to feed the printer) the printer is always stopping waiting and processing data. I discovered that PCL generated from windows is small and compact, where PCL generated by GhostScript is _huge_. As an example, a LaTeX-made 5 page 78Kb PostScript with no images becomes 610Kb when converted to PCL and ready to print, and I've seen greather growth. I guess the slowdown is caused by the limited throughput of the parallel interface. I am using gs-aladdin 5.50-3 on a plain Slink system, input is filtered by /etc/magicfilter/ljet4-filter (since there was no ljet5-filter), and magicfilter is version 1.2-28. And now, the questions: - Why is PCL generated from GhostScript so big? - Are there plans to optimize it? - Are there plans to make libraries and utilities for printing directly PCL? (like a dvipcl or direct pcl output for Gnome and KDE printing systems) - Are there plans to make GhostScript interact with the printer to query for installed fonts, or to make an utility to query for installed fonts and generate a configuration file for GhostScript? - Are there plans to make some printing system able to exploit printer specific options? For example I know that the HP LaserJet 5 supports selection of resolution, toner saving, printing multiple copies and such, but I have to resort to the printer's panel to control these settings. - Is there some way to turn a Linux box into an high performance PostScript module for a printer? - How can I print my documents at 12ppm from Linux? TYA, Enrico -- GPG public key available on finger -l [EMAIL PROTECTED]
Understanding apm on an intranet server
Hello! Is there some walkthrough to configure apm to do everything to lower power consumption (i.e. turn off hard drives, set the cpu to low power mode, switch off the monitor, whatever else) even if I'm not using a laptop? It could be interesting to have a server (e.g. a print server, or a local diald masquerading connection manager, or an intranet web server, or both) that needs not to be switched off after work time, but can use the least possible power when inactive until maintenance time overnight, and sit sleeping again until the first connection the next morning: what are all the exact things I need to do for this? How can I configure apm (I found no configuration files for it, except the two empty suspend.d/resume.d directories in /etc/apm)? Where do I disable writing --MARK-- lines in log files (that cause hard disks activity and turn them on)? What are slink's standard periodical events (e.g. mail checks) that I should disable, or set not to be run beyond work time, or better, not to be be run when the system is in low-power mode? Is this last thing possible, and if it is, how? And now, some more specific troubleshooting: When going back and forth from X I get this message on console: apm: set display ready: Unrecognized device ID while, instead of a monitor switchoff after the usual 10 (?) minutes being idle, I get this: apm: set display standby: Unrecognized device ID Why? What do they mean? Why isn't my monitor switched off and how can I get it to do it? OTOH X's dpmi is working well. Bye and TYA, Enrico -- GPG public key available on finger -l [EMAIL PROTECTED]
Flushing the slink question cache
Hello! I've come up with a bunch of practical questions around the Debian Slink distribution: 1) There isn't a group 'shutdown' to whom add people allowed to shutdown, while I often find a need for it in many environments: is there another way (i.e. The Debian Right Way) to do it? 2) /cdrom isn't given an entry in /etc/passwd after a fresh install, even if it has been used during all the installation and there is a /cdrom directory for it: this is a note for the next installation procedure 3) make-kpkg rebuilds all of the kernel after every single change: if I just compiled a kernel and wanted to add, say, a module, everything gets recompiled instead of just the few required files: is there a clean way to allow this? Can I just set do_clean := NO in debian/rules? 4) It has been a pain for me to install in a SCSI only environment with an Adaptec 2940UW Pro host adapter: no way to see it until I found reference to the alternative install disk for the 2940 searching debian-user. Installing from that disk and the plain official slink cd's also didn't install on the hard disk the patched kernel, so I needed to boot the system with rescue=/dev/sda1, install the kernel from the floppy and rerun lilo. Is there some distribution specific troubleshooting page on the Debian web server to provide help for troubles that came up after the official release, like pointers for alternative boot disks like the Adaptec one or the Gnome slink staging area, tips, specific faq and such? 5) On a server I installed (hamm upgraded to slink), the wwwoffle daily maintenance script happen to kill the wwwoffle daemon; I didn't happen to discover why is it doing this, and I can't just work around it putting /etc/init.d/wwwoffle start as the last line in the maintenance script because I can't guess what state wwwoffle was before being killed (e.g. online, offline, autodial) 6) SSH came with unencrypted communcation disabled at compile time; I run a network in a secure environment and want to use ssh for its cryptographical authentication, but allow it to operate unencrypted when connecting to local hosts, since there's no risk of network snooping and I want to avoid unnecessary computation. Couldn't the option be disallowed by default in a system wide configuration file instead of compile time? This isn't an important issue, however, because one can suppose that if I'm smart enough to tell when an environment is secure, then I'm smart enough to download the source package and recompile it. I just wanted to point it out. 7) apmd: I often get these two errors printed on console, the first one when going back and forth from X and the latter instead of a monitor switchoff after the usual 10 (?) minutes of no operation on console: apm: set display ready: Unrecognized device ID apm: set display standby: Unrecognized device ID Why? What do they mean? Why isn't it switching off my monitor and how can I get it to do it? X's dpmi is working well, instead. And, while I'm on apm: is there some walkthrough to configure apm to do everything to lower power consumption (i.e. turn off hard drives, set the cpu to low power mode, switch off the monitor, whatever else) even if I'm not using a laptop? It could be interesting to have a server (e.g. a print server, or a local diald masquerading connection manager, or an intranet web server, or both) that needs not to be switched off after work time, but can use the least possible power when inactive until maintenance time overnight and sit sleeping again until the first connection the next morning: what are all the exact things I need to do for this? How can I configure apm (I found no configuration files for it, except the two empty suspend.d/resume.d directories in /etc/apm)? Where do I disable --MARK-- lines in log files that turn on hard disks? What are slink's standard periodical events (e.g. mail checks) that I should disable, or set not to be run beyond work time, or better, not to be be run when the system is in low-power mode? Is this last thing possible, and if it is, how? 8) I've never been able to use the multi_cd access method of dselect: I always happen to see only the contents of the first cd, while with apt I need every cd mounted to see them both. For the former, I know it's my fault, for the latter I understood I have to wait for potato to be able to use apt with multiple cd: am I right? And for multi_cd, what's the right way of operation? I have never been asked for the second CD. 9) Less Debian specific, but very important: :) how can I configure a key combination in Enlightenment to behave like WindowMaker's Raise/Lower action? (i.e. raise a window or send it to back if it's already on top: it allows you to switch between your applications like
Another two questions...
Hello! I forgotten two more questions to ask: 1) How's life going for visual impaired people? The only package specific to blind users is emacspeak, but I know there's more: is BRLTTY still alive? Last time I checked was some two years ago: has something new come out since then? 2) I need a way to shut down a debian box in an easy way with a keyboard but no monitor and no user login: can I configure an alternative sequence like ctrl+alt+del to do a shutdown instead of a reboot? Can I have the PC's speaker beep something or better have some samples feed to /dev/dsp when the system can be safely turned off? Can I just do it in a script /etc/rc0.d/S89saygoodnight ? Bye and TYA Again, Enrico -- GPG public key available on finger -l [EMAIL PROTECTED]
Managing a hybrid slink/potato Debian installation
Is there a way to say apt that the system should be slink except for some packages (eg. wmaker, x11amp, gnome, wine) that should come from potato, so that I can safely issue an apt-get upgrade without being asked to download 63.1Mb of packages? TYA, Enrico -- PGP key available on finger -l [EMAIL PROTECTED]
apt-get weirdness
Hello! Could someone explain me this? marvin:~# apt-get install libgnome-dev Updating package status cache...done Checking system integrity...ok The following extra packages will be installed: libtiff3g-dev libglib-dev libungif3g imlib-progs libungif3g-dev gdk-imlib-dev libjpeg62-dev libgtk1.1-dev libpng2-dev The following packages will be REMOVED: libgtkmm-dev giflib3g libjpegg-dev libgtk-dev The following NEW packages will be installed: libtiff3g-dev libglib-dev libungif3g imlib-progs libungif3g-dev gdk-imlib-dev libjpeg62-dev libgtk1.1-dev libpng2-dev libgnome-dev 0 packages upgraded, 10 newly installed, 4 to remove and 2 not upgraded. 1 packages not fully installed or removed. Need to get 0b/2043k of archives. After unpacking 769k will be used. Do you want to continue? [Y/n] The system should be plain official i386 slink (from a pre-release slink I put the two official CDs in two cd readers, pointed sources.list to them and did apt-get update; apt-get dist-upgrade) TYA, Enrico -- PGP key available on finger -l [EMAIL PROTECTED]
Wishes and ideas...
Hello! Here are some wishes and new ideas I've piled up in last days: Many people periodically upgrade their systems using apt as I do. One thing I would like to have is a log file showing what changed on my system after each apt-get run, extracting the relative portions of the changelogs of the updated packages and grouping them together. Another nice things that could be added to apt are the ability to upgrade only those pachages who shifted only by upstream release, so to limit the download charge; the best thing would be to keep old .deb's and download just patches to them, but I feel this would complicate things a lot. Changing of scope, an ambitious idea I had was to extend the concept of themes to the global system configuration. I noticed that themes are used whenever an applications supports a configuration system both powerful, comprehensive and full of options, and that Linux servers and applications are exactly that. I wondered if it could be possible to make some webserver or graphic workstation theme that could be installed on a Linux box and provide appropriate configuration for most packages to specific needs. As of now, if one manages to fine tune his box for some specific purpose, it is unpratical to make its work available to people interested in the same purpose. The whole Linux concept provides an unmatched set of extremely powerful, full-featured and versatile applications, and there must be a way to bind them to do a specific job without working on a number of different configuration files each time. Something similar is already achieved in a very simple way by scripts like eximconfig, but it could be extended to a coherent framework to support for more cases. Think of a web server: you can have a web server in hour home Linux box, a little intranet web server, a big virtual hosting web server, each that could choose whether or not to provide, for example, suid cgi, ssi, php3, accounting and squid caching. Why can't we have a web server theme or configurator that can cope with this set of options and provide and entry-level configuration for apache, cgiwrap, php and whatever needs it, so that the sysadmin needs only to tune it up (add the server name and very site-specific features) and not to redo all the work from scratch? I think it's time to think about something that could do to configuration files what dpkg does to the file system. Bye, Enrico -- PGP key available on finger -l [EMAIL PROTECTED]
Why is lpd making this dns query?
Hello!
I have set up a print server and a print client with magicfilter and everything
would be fine, were it not for lpd on the server side making a dns query (and
waiting a long timeout) for every remote connection.
My system is not always connecting to the internet, and a query for a host not
in local network results in a timeout looking for forwarders, so I looked if in
/etc/hosts.lpd there were host names that required forwarding to resolve. No,
they are all local, and the client box is both local and in /etc/hosts.lpd.
So I configured the DNS to log incoming queries, and I fond this:
Dec 8 14:08:37 eddie named[549]: XX /127.0.0.1/./A
Dec 8 14:08:55 eddie last message repeated 5 times
So why is lpd so insistently looking for a '.' dot?
As a temporary solution I've set the dns to avoid recursion when not on line
through a couple /etc/ppp/ip-{up|down}.d scripts that change a symbolic link to
two different /etc/named.conf.
The question is: what is lpd looking for?
Ciao! Enrico
--
PGP key available on finger -l [EMAIL PROTECTED]
Upgrading with APT: which packages is it going to download?
Hello! As my modem internet connection is slow and (being in Italy) expensive, I would like to use apt to upgrade my system, but being myself the connection using a Zip disk to carry .deb files from our debian mirror at the university to home. Is there some way to know what is apt going to download, so that I can go, fetch them and plug them in? If I had a list like: main/binary-i386/base/libc6_2.0.7t-1.deb main/binary-i386/admin/svgatextmode_1.8-5.2.deb [...] I could even generate a script on the zip drive that will do the downloading form me; if I could also know file size I could have an idea of what downloading is going to be. Ciao! Enrico
wwwoffle breaks file uploads
It appears that when using wwwoffle (Version: 2.1c-1), Netscape (Navigator 4.04, the only I tried) is not able to do file uploads (input type=file in a form enctype=multipart/form-data) anymore. If I disable proxies from Netscape everything starts going fine. Clues? Ciao! Enrico
What about a Linux key?
Hello PC users! What about putting a penguin sticker on top of those Window 95 keys on all new keyboards ? They could come handy as new modifiers, for example to switch virtual consoles, leaving Alt free and providing a good alternative for Ctrl-alt(gr)-Fn in X and DosEmu; more, one could use them to control X window managers leaving usual keys for applications (e.g. Alt-Tab used by Midnight Commander is usually bound to some sort of cycling through windows), and we could use a single-key compose. Their scancodes are already seen by Linux, but I could not find any way to bind them to a modifier (it seems there are no extra modifiers to use). Also, I'm not good enough to fiddle with xmodmap. Anyway I have been successfull in binding the menu-like thing to compose: echo keycode 127 = Compose | loadkeys. Is there a way to have a new modifier or we need to hack kernel? How is Linux doing on a Sun, since they have plenty of extra keys and modifiers? Maybe we could build a new keyboard mapping table? Bye, Enrico S -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word unsubscribe to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .
What about a Linux key?
Hello PC users! What about putting a penguin sticker on top of those Window 95 keys on all new keyboards ? They could come handy as new modifiers, for example to switch virtual consoles, leaving Alt free and providing a good alternative for Ctrl-alt(gr)-Fn in X and DosEmu; more, one could use them to control X window managers leaving usual keys for applications (e.g. Alt-Tab used by Midnight Commander is usually bound to some sort of cycling through windows), and we could use a single-key compose. Their scancodes are already seen by Linux, but I could not find any way to bind them to a modifier (it seems there are no extra modifiers to use). Also, I'm not good enough to fiddle with xmodmap. Anyway I have been successfull in binding the menu-like thing to compose: echo keycode 127 = Compose | loadkeys. Is there a way to have a new modifier or we need to hack kernel? How is Linux doing on a Sun, since they have plenty of extra keys and modifiers? Maybe we could build a new keyboard mapping table? Bye, Enrico -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word unsubscribe to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .
