Re: LILO-rific
On Mon, Sep 04, 2000 at 01:15:56PM -0700, Tal Danzig wrote Hello, Set boot to /dev/hda (the MBR) and your setup should work just fine. Tal If you're really paranoid, you can use it as-is; you just have to use fdisk (linux or windows) to make /dev/hda3 (and only /dev/hda3) active/bootable after installing LILO. That tells the DOS MBR which partition boot record to load. On Mon, 4 Sep 2000 16:08:55 EDT, [EMAIL PROTECTED] said: : Okay, I pretty much get LILO... Here's my deal: Windows in on hda1, Linux is : on hda3 (hda2 is a compaq backup, and hda4 is Linux Swap). What I want to : have happen is that when I start my comp without any boot floppies or CDs, I : have the choice to load Win or Lin. I would like Win as default, and I would : like a password protecting Win, although that's not so important. What I've : got as my lilo.conf at this point (which I have not yet loaded with the lilo : command) is: : : boot=/dev/hda3 : root=/dev/hda3 : install=/boot/boot.b : map=/boot/map : vga=normal : delay=20 : other=/dev/hda1 : label=Win : table=/dev/hda : password=secret : image=/vmlinuz : label=Linux : read-only : : I'm a little confused about boot being set, since shouldn't that maybe be : hda1, since windows MBR, er... I really don't get that part so well. The : last question is: As of now I'm using a floppy to boot to Linux. Otherwise : the system boots to Windows no questions asked, since I haven't touched the : MBR. If I load lilo into the MBR, and I decide it's not working out for me, : is there a quick way to pull it back out? Much thanks, bye! : : -Chris : : : -- : Unsubscribe? mail -s unsubscribe [EMAIL PROTECTED] /dev/null : : -- | Tal Danzig | Join #libranet on the | | [EMAIL PROTECTED] | openprojects IRC network | | http://www.libranet.com| Tal Danzig | | The TOP Desktop! | [EMAIL PROTECTED]| -- Unsubscribe? mail -s unsubscribe [EMAIL PROTECTED] /dev/null -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: Netscape 4.75 problems
On Tue, Sep 05, 2000 at 06:20:04PM +0100, Anthony Campbell wrote On 05 Sep 2000, Dale L . Morris wrote: [snip] I am currently running 4.73 Netscape from potato and it works fine. But when I do the apt-get upgrade it ignores the netscape packages for 4.75. Anyone know why? [snip] How do you get to 4.73? I got 4.06 when I upgraded slink to potato. I have upgraded subsequently via proposed updates but this has not yielded 4.73. Like the kernel, Netscape packages are version-specific; that is to say that different versions are treated as separate packages, not as different versions of the same package. To install the latest statically-linked version of Netscape Communicator, use something like # apt-get install communicator-smotif-475 If you want to track the latest release, try # apt-get install communicator Communicator is a meta-package that depends on the latest statically-linked release; that way if Communicator 4.76 makes it out the door and gets packaged, you should get it along with everything else when you go # apt-get update; apt-get upgrade John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: continuing upgrade after losing connection
On Sat, Sep 02, 2000 at 02:46:31AM -0800, [EMAIL PROTECTED] wrote hi, I'm in the process of upgrading from slink to potato. I'm at the step where i'm doing apt-get --feixbroken --show-upgraded dist-upgrade. My isp is set up such that I only have a connection for up to 8 hours at a time. I ran the apt command overnight last night and got a lot of packages but it wasn't long enought to get everything. Is there way to indicate to apt-get which packages I now have in /var/cache/apt/archives so that it doesn't try to re-download them and starts with the files it doesn't already have? (please keep in mind that if any of the answer to this is in the man pages for apt-get, sources.list, or apt.conf that for whatever reason, I've never gotten those pages in my current installation and I can't find what package might have included them that I might e missing. That's one thing I'm really hoping will change with this upgrade, it's very frustrating not to be able to read parts of the FM for myself) Just my understanding, but I believe that apt is smart enough to check its cache before downloading, so if it's already got a bunch of stuff it won't get those again. here's a thought if I did a dpkg --get-selections filename and went in and editted it so that everything I already have is set to 'hold' and then did a dpkg --set selections filename would that non-dangerously do what I want to do, or does it have the potential to seriously F anything up? -Alice (come on potato... big upgrades... no whammies... no whammies...) Your main danger is of running out of space in /var/cache/apt/archives; if it looks like you will and you can't arrange for more space try upgrading a few large thing incrementally first, like maybe # agt-get install emacs20 # apt-get install task-tex # apt-get upgrade # apt-get dist-upgrade John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
ftp.debian.org Packages files
Hello! I notice that the Packages files for potato/main, potato/contrib and potato/non-free on ftp.debian.org were replaced recently; the new files don't work for me. In place of Filename: filename these copies have FileName: filename which doesn't work (at least, dpkg-deb doesn't recognize it). Is this the result of a buggy maintenance script? Has ftp.debian.org been cracked? I could simply fix/replace my copies of these, but for reasons which should be obvious I'd like to check that nothing untoward is going on. Does anyone on the list with knowledge of ftp.debian.org know what the story is? John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: Verifying my debs / suggestion for ISO downloads
On Wed, Aug 30, 2000 at 02:03:18AM +0800, [EMAIL PROTECTED] wrote So I'm now on the point of downloading my potato. What's the Debian equivalent of a Redhat rpm --checksig *.rpm? This command is supposed to verify the package signatures (md5, pgp, gpg -- but I've gone only as far as the md5). How do I know if my download is all right? I have a more or less working Storm Linux installation, so I guess I can dpkg it. BTW: I downloaded my Storm thru Linuxberg. The iso image came in 45 mb chunks, with an MD5 sum check list for all. A cool idea. Why hasn't anybody else thought of this (or have I been looking in the wrong places)? A simple split command can break up that 650 ton blue whale into more manageable kittens. For the clueless downloader, we can provide a simple cat script to piece together the bits (as Linuxberg and/or Storm Linux has done.). The debian website contains a file 'md5sums.gz', which contains md5sums for the files you will be downloading (and more besides) but AFAICT there is no 'automated' way to check individual packages. A better solution IMHO is to use rsync to download the packages, which effectively verifies the files as they are downloaded; it doesn't protect you against packages that are corrupt on Debian's site, but that's only happened a couple of times in the last few years across all of the supported architectures. Rsync also makes it much easier and faster to maintain your mirror, to (e.g.) ensure that you always have the latest security updates. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: What is up with Debian 2.2 Potato ??
On Fri, Aug 25, 2000 at 08:45:34PM -0700, Nate Amsden wrote Joey Hess wrote: Nate Amsden wrote: I brought up a new virtual terminal and sure enough a uname -a shows kernel 2.2.12, but there is no /lib/modules/2.2.12 There is however a /lib/modules/2.2.17. sounds like an oreilly cd ?? Impossible, O'Reilly is not selling any CD's labeled Debian 2.2. It looks to me like he didn't wipe out entirely his old install, which had the 2.2.12 kernel. even if he did not wipe out his installer the reinstall should of wiped it out for him, i have never had to wipe out an install before reinstalling. If he installed LILO in the MBR first time around and hasn't replaced his MBR since (e.g. he hasn't installed LILO this time around, or installed it in a partition boot record without replacing his MBR) then he may still boot his old kernel, if the space that it occupies on the disk hasn't been overwritten in the meantime. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: no ppp compression even though pppstats says VJCOMP; new in potato
On Fri, Aug 25, 2000 at 11:55:44PM -0400, Daniel Barclay wrote BCC: [EMAIL PROTECTED] References: [EMAIL PROTECTED] --text follows this line-- I don't seem to be getting any compression in ppp. Is there something extra I need to do in potato's ppp to fully enable compression? pppstats lists most packets in the VJCOMP column, seemingly indicating that it thinks packets are getting compressed. However, the maximum bytes per second I get when downloading a very compressible large web page is no larger than the bytes per second for uncompressible data. (I get about 3850 bytes per second raw speed (e.g., downloading a .gz file). The compressible web page used to download at around 11000 bytes per second on my previous slink system, but now it downloads at only 3850 bytes per second.) Look elsewhere. VJ compression applies to headers only, and its impact will even in the best case be much less noticeable than the numbers you quote here (at least, for packets containing more than a few bytes of data). This is much more likely to be an issue with your serial port or modem setup, or perhaps your MRU (man pppd for details). Good luck, John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: your mail
On Sun, Aug 27, 2000 at 12:02:48PM +1000, W. J. Simson wrote Could you please tell me when daylight saving was reintroduced in Australia Regards Jeanette Simson The eastern states went 1 hour ahead at 2am today (Sunday Aug 27), presumably to maximise solar energy collections and sunblock sales during the Olympic Games. More sensible states (WA (Perth), SA (Adelaide)) resisted the urge, and go ahead in a month or two. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: Procmail
On Fri, Aug 25, 2000 at 03:59:10AM -0500, Will Trillich wrote On Thu, Aug 24, 2000 at 07:35:07PM -0700, Dale L . Morris wrote: I'm setting up procmail on my system. I've done this in the past successfully using Redhat 6.2 and sendmail, but I have a couple of questions about Debian and exim. 1. Do I need a .forward file? Or is .procmailrc enough? imagine needing to kill a mosquito, and having someone hand you an atomic-powered, planet-smashing axe. that's how i think of what you can do with procmail recipes. i'm pretty sure [y'all stomp on me if i'm off, here] that a .forward file would be a tepid, drab subset of the things you can do in .procmailrc file. Most mta's don't, by default, run procmail as an MDA. Depending on your MTA and how it's configured, you may need a .forward file to get procmail actually run; if you're using Exim, a .forward file containing just the line |/usr/bin/procmail should be enough. If you're running something else, check your MTA's (or procmail's) documentation to see if you need a .forward file, amd what you'ld put in it. 2. Do I need to have mh installed? well, i don't. draw your own conclusions. :) Not unless you want to read your mail using mh, or use a mailer that relies on mh. Even then, you can use nmh instead. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: windowmaker
On Fri, Aug 25, 2000 at 08:33:16PM +0800, Goeman Stefan wrote Hello, Probably a stupid question. I am running WindowMaker under Gnome. I want to have the wmitime icom appear on my desktop when I log in. Putting a line like: exec wmitime in my .xsession file does not work. How should this be done? Maybe a stupid answer... Most Xsession files end with a command like exec wmanager to start a window manager or some other persistent app; your session ends when that app ends, and bash stops executing your script at that point, as the exec command replaces the bash process with your wmanager (or whatever). If you just add exec wmitime to the end of an existing, working Xsession, the command will never be executed; add a command like wmitime to your Xsession before the exec command that was probably there when you started and it should work. For a little more info on exec, man bash and search for ' exec ' HTH, John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: Is it just me or.......
On Fri, Aug 25, 2000 at 02:09:23PM +0800, Cam Ellison wrote Me, too .. and quite a few with empty bodies. On Fri, 25 Aug 2000 09:33:26 -0700 (PDT), Greg Strockbine. wrote: No its not just you. I've noticed the dups too. - g.s. On Fri, 25 Aug 2000, Christopher W. Aiken wrote: Is it just me or has every message posted to this Debain list been posted 3-4 times today? I've had 150+ emails and they were just duplicates of previous postings over and over !! Here, it looks like a broken usenet gateway; all the duplicate mail I've seen has the header X-Envelope-Sender: [EMAIL PROTECTED] As well as repeating messages that originate from the list, it looks like it's broken in other areas too, as many of the messages show up in Mutt as MIME messages with broken or missing multipart boundaries. Hopefully, someone in .hk knows who to contact to get it shut off... in the meantime, I've added this to my Exim .forward file: if $h_X-Envelope-Sender: matches [EMAIL PROTECTED] then seen finish endif which will hopefully mean I don't have to wade through them by hand. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: Linux Mail Client (was: Re: Web browsers for Linux (was: Re: Netscape Bus Error))
On Wed, Aug 23, 2000 at 10:39:01PM -0700, Seth Cohn wrote On Thu, 24 Aug 2000, John Pearson wrote: On Wed, Aug 23, 2000 at 07:31:07AM -0700, Steve Lamb wrote Technically, yes. However, if your boss says that work email is not to touch outside SMTP servers as a matter of policy how far do you think Well, the SMTP server will route it correctly anyway, that is what they do will fly? There are reasons other than technical to different servers. *sigh* bosses, bosses, bosses. All other arguments in this thread aside, this one is a bit weird. Does your boss realise that any non-local mail you send via your work SMTP server will be handed, unencrypted and with only the most rudimentary checks, to an outside SMTP server for forwarding or delivery? Um, reverse that. Steve was saying _work_ email touching _outside_ servers. In other words, company email shouldn't pass thru outside mail servers. This is actually a sound practice, if a bit paranoid, but I can understand the requirement. My misunderstanding. To me, work email is email either to or from work. Even so, if they don't trust an ISP to recieve and forward mail, they have little reason to trust it to receive and forward packets. I might have plonked Steve, but don't misstate what he asked. Never my intention. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: Linux Mail Client (was: Re: Web browsers for Linux (was: Re: Netscape Bus Error))
On Tue, Aug 22, 2000 at 09:36:14AM -0700, Steve Lamb wrote On Tue, Aug 22, 2000 at 07:21:38PM +0930, John Pearson wrote: .forward file allows you to filter your mail into any number of separate mailfolders at delivery time, based on a wide range of criteria including the contents of the headers. Now take it a step further, what do you do on the MUA (not mail client) side to address that? Well, that certainly indicates one reason why I'm having difficulty coming to grips with your requirement; we have a problem over terminology. I differentiate between MUAs, MDAs, and MTAs; examples are: MUA: mutt MDA: procmail MTA: exim Obviously, you mean something different to MUA to me (and, perhaps, others); what, in your view is an MUA if not a mail client? John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: Potato vs Realtek8029PCI NIC
On Tue, Aug 22, 2000 at 09:33:45PM +0200, Vitux wrote Trying to build a tiny lan here... Recompile fresh 2.2.16 with the ne2k-pci driver as module. No signs of nic when booting, insmod ne2k-pci.o gives me unresolved symbols-error, and ifconfig -yadayada gives me error to the effect that there's no hardware to configure. So the question is: to nic or not to nic? I've determined that the kmod-bit works (all the other modules are inserted automagically on request), and little lights are shining from nic and hub. Is there any other voodoo I have to do to get the card recognised/ installed/ configured/ whatever? It said in a howto somewhere (I forget which, I've pored over so many of'em lately;-), that the Realtek RTL8029 is a NE2000-clone, and so should use the driver for same... Right?! # modprobe ne2k-pci Ne2k-pci, like many other modules, depends on other bits that may also be built as modules. insmod alone won't work, unless the other bits are already present; modprobe handles module dependencies for you. Assuming that works, just adding ne2k-pci to /etc/modules should do the trick on a more permanent basis. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: Potato vs Realtek8029PCI NIC
On Wed, Aug 23, 2000 at 11:53:18AM +0200, Vitux wrote More stuff: I did a # modprobe ne2k-pci and I get this: ne2k-pci..: PCI NE2000 clone 'Realtek RTL-8029* at I/O 0x20a0, IRQ 9. eth0: RealTek RTL-8029 found at 0x20a0, IRQ 9, 00:00:B4:B8:94:CC # I suppose this means that the modular driver has been installed in the kernel and detected my NIC. So far, great. Wonder why kmod won't autoload it when I do ifconfig-yadayada up ? BTW: What does the hex-part at the end mean? It has no reason to associate this driver with eth0; you need to give it a big hint as to which module to use for the interface. Try adding alias eth0 ne2k-pci to /etc/modutils/aliases and running update-modules as root. The hex string is your card's MAC, an allegedly unique identifier that is used for packet addressing at the physical link layer for machines on your local network. The first few octets probably identify the card's manufacturer, the rest are up to them. I say allegedly unique because some early clone NICs (back when a cheap name-brand NIC might be $300) had cloned firmware that gave each card the same MAC, or used ranges that had been assigned to other manufacturers; that meant if you used bad cards, you might have to sort out which ones couldn't share a network. MACs are associated with IPs using ARP (Address Resolution Protocol). You can monitor the MAC-to-IP translation on your network with something like this (assuming your network is 192.168.1.0/24): $ ping -c 2 192.168.1.255 $ /usr/sbin/arp -a You won't see your own machine in the arp cache, because your TCP/IP stack recognizes packets addressed to itself and doesn't get as far as attempting ARP for local addresses. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: Exim and multiple domains
On Sun, Aug 20, 2000 at 07:03:17PM -0500, Will Trillich wrote /lurk mode i'm still hoping to find the king james version of the exim manual (the original greek is beyond me)... this flashed by a few days ago, and i thought i'd be able to apply it to my own situation; alas... On Thu, Aug 17, 2000 at 12:09:41AM +0930, John Pearson wrote: There's more than one way to skin a cat, but here's what I've done; it allows you to have an arbitrary number of virtual domains with each having its own alias file, which allows you to do most things including what you are looking for. Some of my virtual hosts are in .com.au, and some are in .com; I have a file for domains with three parts (e.g., mydomain.com.au) at /etc/exim/domains, and a file for domains with two parts at /etc/exim/domains2. Each file contains lines like this: *.mydomain.com.au myfile *.otherdomain.com.au otherfile for each domain. My setup allows me to match arbitrary subdomains (e.g., mail.outgoing.mydomain.com); if you don't want to match subdomains you can omit the *. from each line, and use a single file for all domains. The second field is the name of an aliasfile for the domain matched by the first field. in my /etc/exim/domains file, i've got: *dontuthink.com will To forward all mail for a domain to myuser's local mailbox, the specified file need contain only the line: * myuser but it can also use any construct that Exim recognises in alias files (pipes, :blackhole:, etc.). btw--does the second field denote an aliasfile name or a user name to redirect mail to? In exim.conf, I set local_domains like this: local_domains = localhost:my.net.au:*.my.net.au:\ partial3-lsearch;/etc/exim/clients/domains:\ partial2-lsearch;/etc/exim/clients/domains2 The partial*-lsearch; allows me to match arbitrary subdomains (e.g., mail.outgoing.mydomain.com.au). If you don't want to match subdomains you can put all virtual domains into one file and use just lsearch; my exim.conf contains: local_domains = localhost:*serensoft.com:lsearch;/etc/exim/domains and virtual: driver = aliasfile domains = lsearch;/etc/exim/domains no_more file = /etc/exim/domains search_type = lsearch* mail to the main site (serensoft.com) gets through as it should, but of course (if your eyes are trained on what to look for i'm sure it'll be obvious) email to [EMAIL PROTECTED] gets frozen: lowest MX record for dontuthink.com points to local host short version: what am i missing? long version: how would a newbie diagnose this kind of thing? If you only want to match the top-level domain dontuthink.com and not subdomains like mail.dontuthink.com (as seems to be the case, from how you've set your local_domains line) then leave out the wildcards from /etc/clients/domains, like so: dontuthink.com will The second field is the name of a file. The director that I use (virtual) retrieves this field as $domain_data. The file contains aliases for users in that domain; the simplest case is a file like * will which forwards all mail for the domain to will. Your virtual director should look more like this: virtual: driver = aliasfile except_domains = localhost:serensoft.com:*.serensoft.com domains = lsearch;/etc/exim/domains no_more file = /etc/exim/$domain_data search_type = lsearch* The except_domains is probably not really necessary, but prevents the driver matching your 'non-virtual' domains (the domains= line should be enough). The lsearch* (in place of just lsearch) allows wildcard matching in the aliasfile. In the example you gave, there would be an aliasfile for dontuthink.com at /etc/exim/will that, for example, might contain the line * will to forward all mail for dontuthink.com to the local user called Will. HTH, John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: Linux Mail Client (was: Re: Web browsers for Linux (was: Re: Netscape Bus Error))
On Wed, Aug 23, 2000 at 07:31:07AM -0700, Steve Lamb wrote On Wed, Aug 23, 2000 at 09:27:40AM -0400, David Zoll wrote: [snip-o-rama] Which can then route the mail to the appropriate mail server. This is how SMTP was designed to work. Technically, yes. However, if your boss says that work email is not to touch outside SMTP servers as a matter of policy how far do you think Well, the SMTP server will route it correctly anyway, that is what they do will fly? There are reasons other than technical to different servers. *sigh* bosses, bosses, bosses. All other arguments in this thread aside, this one is a bit weird. Does your boss realise that any non-local mail you send via your work SMTP server will be handed, unencrypted and with only the most rudimentary checks, to an outside SMTP server for forwarding or delivery? John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: What are MUA, MTA, MDA? (Was Re: Linux Mail Client)
On Wed, Aug 23, 2000 at 03:42:16PM -0400, David Teague wrote On Wed, 23 Aug 2000, John Pearson wrote: [snip] I differentiate between MUAs, MDAs, and MTAs; examples are: MUA: mutt MDA: procmail MTA: exim John, 1) What do MTA, MUA, MDA stand for? MTA - mail transport agent, responsible for machine-to-machine routing of mail messages; MDA - mail delivery agent, responsible for local delivery of mail messages to the user's mailbox/whatever. Most MTAs include at least a rudimentary MDA. MUA - mail user agent, responsible for providing a user access to his mailboces/folders. I know that mutt is a mailer, not unlike exim and smail, but has other functionality. procmail filters mail, but what else? exim seems to be a drop in for smail and sendmail, so has similar functionality. 2) What are the words for these acronyms? I have a bit of the answer: MTA is probably Mail Transport Agent (guess). MDA is Mail Delivery Agent. (from procmail man page, I can guess it delivers mail) man mutt doesn't tell much and there no exim man page on my system. Exim has extensive documentation in info format; install exim-doc and run info exim, or use dwww to browse it. What is MUA? 3) What is the function of these? 4) Where would I look this up? What is TFM I should R? NRS which FM these came from; I think I picked them up from some HOWTO or other. It's not a hard and fast separation (as witness the common blurring of MTA and MDA), but provides a convenient conceptual division along functional lines. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: exim and multiple alias files
On Mon, Aug 21, 2000 at 01:14:55PM -0500, Brent Harding wrote Can exim handle multiple alias files? I tried switching to sendmail for this, but whatever I did playing with postfix to see what it was about before removing it, made everything halt up in mqueue under var/spool, I think. Mail wouldn't even go to local users, using exim fixed it, but I like sendmail's advanced features, so I can get majorcool going, it won't append majordomo aliases to /etc/aliases. It needs a file that has only majordomo aliases in it, so when mj_build_aliases is run from the cgi, it can rebuild what lists are there, using /etc/aliases would cause all my other aliases to go, I suppose setting up a mailing list with one user subscribed, the destination address would kind of fix it, but it's really not needed. No problem. Just add a second aliasfile director to /etc/exim.conf, right after the normal aliasfile director: majordomo_aliases: driver = aliasfile file_transport = address_file pipe_transport = address_pipe file = /etc/aliases.majordomo search_type = lsearch That should do it. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: Linux Mail Client (was: Re: Web browsers for Linux (was: Re: Netscape Bus Error))
On Tue, Aug 22, 2000 at 12:54:58AM -0700, Steve Lamb wrote On Tue, Aug 22, 2000 at 05:46:00PM +1000, [EMAIL PROTECTED] wrote: .fetchmailrc can have: [] user x is mark here [] user y is julie here Requires a local account for what really isn't a separate account on the local machine. This is a piss-poor hack. Alternatively, if you don't want separate acounts for work / home, you can use an exim .forward file to filter and save your home stuff to a seperate mailbox file and mutt -f the file. Alternatively... I have already addressed this in this thread. IE, dumping all mail into a single account and then filtering out from there. This is not acceptable. Again, a hack to the extreme. Perhaps it would help if you re-stated what it is you want, and explain why this isn't a part of the solution. Using an Exim .forward file allows you to filter your mail into any number of separate mailfolders at delivery time, based on a wide range of criteria including the contents of the headers. If you don't want all mail delivered to a single mailbox, and you don't want mail delivered to several mailboxes belonging to different mail users, and you don't want mail delivered to several mailboxes all belonging to the same user, what is it you *do* want? John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: ** Emegancy Request **
On Thu, Aug 17, 2000 at 04:54:42PM -0500, Nathan E Norman wrote On Fri, Aug 18, 2000 at 12:17:28AM +0300, Shaul Karl wrote: For some reason that I do not understand the authors of the Hrad Disk Upgrade Mini How-To claim about this or very similar one that Previous versions of the Mini How-To stated that you could also use tar to copy the disk, but this method was found to have a bug. tar doesn't deal well with device files. Also, last I checked it didn't correctly handle hard links or sparse files, and ignored sockets entirely (not necessarily a big deal). I use afio for this kind of thing, and it's never failed me yet. e.g.: # find / -xdev -path '/lost+found' -prune -o -print | afio -p /newroot John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: Duplicating a file system / re: ** Emegancy Request **
On Fri, Aug 18, 2000 at 08:53:29PM +0200, Sven Burgener wrote On Fri, Aug 18, 2000 at 10:38:20AM +1200, Dan Griffiths wrote: This command will take care of duplicating everything including device files and permissions: find source dir -mount | cpio -dumpv target dir I have a (bigger) SCSI disk that I want to move my system onto. (Currently my system lives on a smaller IDE disk) I created a /boot, a swap and a / partition on /dev/sda: # fdisk -l /dev/sda Disk /dev/sda: 255 heads, 63 sectors, 131 cylinders Units = cylinders of 16065 * 512 bytes Device BootStart EndBlocks Id System /dev/sda1 * 1 3 24066 83 Linux /dev/sda2 411 64260 82 Linux swap /dev/sda312 131963900 83 Linux I next mkXX'ed them and mounted /boot and / under the two mount points /mnt/newboot and /mnt/newslash, respectively. After that, I did the command described above to copy over everything. Next, I adjusted /mnt/newslash/etc/fstab to reflect the new disk. I also adjusted lilo.conf accordingly and ran lilo -C /mnt/newslash/etc/lilo.conf. This step may cause you trouble. Unless you used paths like /mnt/newslash/vmlinuz in /mnt/newslash/etc/lilo.conf then lilo will be pointing to your old kernel/other files; this shouldn't be a problem the first time you boot, but you'll need to run lilo again to get things sorted and stable. Now, I suppose if I boot up next time, this should work just fine, correct? I can't actually test this right now as I am recompiling a kernel on a different box that I ssh'ed to from this box... :) So, does anyone have any suggestions / comments on this topic? While many BIOS/SCSI combinations play nicely together, many combinations also have real and sometimes intractable issues over booting. In your position I'd probably make a boot floppy, just in case. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: problems upgrading standard kernel to 2.2.17-ide
On Fri, Aug 18, 2000 at 02:14:06PM -0400, Dave Bresson wrote On Fri, 18 Aug 2000, David Wright wrote: a) If you're compiling a kernel module, you need kernel headers. When people write /usr/src/linux, they really mean the kernel headers for the running kernel. Okay, i *do* have the kernel headers in /usr/src, (/usr/src/kernel-headers-2.2.17/include) however, i have /usr/src/linux linked to the kernel source in /usr/src/kernel-source-2.2.17. What's the difference between the kernel-header include/ and the kernel-source include/ ? And, also, before i untar'ed the entire kernel-source package into /usr/src, the /usr/src/ directory only contained the kernel headers and i tried compiling the 3c90x driver module using the those headers. I was able to get a good compile, however, i still got the stupid version problem when trying to do a 'insmod 3c90x'. The error it gives is basically that the module was compiled for kernel 2.2.17 (the version given in /usr/src/kernel-headers-2.2.17/include/linux/version.h) and that it could not install the module since the running kernel was 2.2.17-ide. I know i could force the insmod with the -f option, however, i would like a better solution. I mean, the running kernel is *still* a 2.2.17 kernel + the dma66 stuff, so shouldn't the version for it still be 2.2.17? After all, i'm sure the stock 2.2.17 kernel and the 2.2.17-ide kernel came from the same source. How do i solve this? b) The kernel-source tarball/debian package is installed into /usr/src as it's owned by Debian. However, Debian practice is to unpack it wheresoever you like. When you compile it, the Makefile's TOPDIR looks after all the path adjustments. That way you can have multiple versions around, and can build everything as an ordinary user with fakeroot. Ah, right, thanks for the info. So, when compiling programs, alot of them need to know where to get the includes and they usually want to get them at /usr/src/linux/include, am i to assume that i *should* be sym linking /usr/src/linux to the kernel headers i have in /usr/src/kernel-headers-2.2.17? That makes sense i guess. Actually, no. Applications that link against kernel headers usually do so because they need to share datatypes and other declarations with the C library; the important thing there is that they use the same version of kernel headers as was used to build the C library. In the days when most people built their own C library this wasn't a big issue, but it is more of one now. For the views of LT himself on this, search the linux-kernel archives for /usr/include/linux; the short version is that he considers symlinking /usr/include/linux into /usr/src/linux wrong, and building the kernel as root or in /usr/src as, at best, unnecessary. The way that Debian handles this is to place linux headers at /usr/include/linux in the libc6-dev package. There are times (like building stuff that really does talk to the kernel, not the C library) when you may have to massage things a bit, but it's really not up to Debian to predict and deal with the idiosyncrasies of individual pieces of third-party software. In your particular case, you could try editing version.h to contain whatever version the running kernel thinks it is. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: tk8.0 drops /usr/bin/wish8.0 despite being in tk8.0_8.0.5-6.deb
On Tue, Aug 15, 2000 at 10:19:01PM -0400, Jameson Burt wrote Three times I entered, dpkg -i tk8.0_8.0.5-6.deb Each time, the file /usr/bin/wish8.0 was not installed. I even purged the package then reinstalled the package. All other files were installed from tk8.0; only this file /usr/bin/wish8.0 was not installed. I finally entered, dpkg -X tk8.0_8.0.5-6.deb /tmp then copied /tmp/usr/bin/wish8.0 to /usr/bin/wish8.0 This is a weird event. Double checking, I see that I have the latest dpkg, version 1.6.14, and am reinstalling this package under potato, 2.2, as successfully upgraded a month ago from slink. Is something amiss with the package tk8.0_8.0.5-6.deb or do you suppose something is amiss with my installation. Well, it works for me. Do you, perhaps, have tkstep8.0 installed? It's possible that this has set up a diversion of wish8.0, or some such... John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: Exim and multiple domains
On Wed, Aug 16, 2000 at 07:23:15AM -0400, Alec Smith wrote Using Exim, how can I configure it to process mail for multiple domains? Specifically I want mail to domaina.com processed by the .procmailrc in User A's home directory while mail to domainb.com is processed by the .procmailrc in User B's home directory. Exim is already configured to accept mail for the domains, but doesn't do any special processing based on the domain name. At the same time, I'd like to take some mail to [EMAIL PROTECTED] and deliver it directly to someotheruser on the same machine. Is there a way to do this without having to use .procmailrc and process the e-mails twice? (ie [EMAIL PROTECTED] - someotheruser and someuser2 - domain.com are mapped in an Exim database -- Something along the lines of Sendmail's virtusertable file) There's more than one way to skin a cat, but here's what I've done; it allows you to have an arbitrary number of virtual domains with each having its own alias file, which allows you to do most things including what you are looking for. Some of my virtual hosts are in .com.au, and some are in .com; I have a file for domains with three parts (e.g., mydomain.com.au) at /etc/exim/domains, and a file for domains with two parts at /etc/exim/domains2. Each file contains lines like this: *.mydomain.com.au myfile *.otherdomain.com.au otherfile for each domain. My setup allows me to match arbitrary subdomains (e.g., mail.outgoing.mydomain.com); if you don't want to match subdomains you can omit the *. from each line, and use a single file for all domains. The second field is the name of an aliasfile for the domain matched by the first field. To forward all mail for a domain to myuser's local mailbox, the specified file need contain only the line: * myuser but it can also use any construct that Exim recognises in alias files (pipes, :blackhole:, etc.). In exim.conf, I set local_domains like this: local_domains = localhost:my.net.au:*.my.net.au:\ partial3-lsearch;/etc/exim/clients/domains:\ partial2-lsearch;/etc/exim/clients/domains2 The linebreaks and whitespace are there only for legibility in this post, and my.net.au is not the real domain; my apologies to whomever owns that name. The partial*-lsearch; allows me to match arbitrary subdomains (e.g., mail.outgoing.mydomain.com.au). If you don't want to match subdomains you can put all virtual domains into one file and use just lsearch; I have two directors that handle virtual domains: virtual: driver = aliasfile except_domains = localhost:my.net.au:*.my.net.au domains = partial3-lsearch;/etc/exim/clients/domains no_more file = /etc/exim/clients/$domain_data search_type = lsearch* virtual2: driver = aliasfile except_domains = localhost:my.net.au:*.my.net.au domains = partial2-lsearch;/etc/exim/clients/domains2 no_more file = /etc/exim/clients/$domain_data search_type = lsearch* Again, if you don't want to match arbitrary subdomains (or all of your virtual domains are the same length) you need only one of these. You can put these directors at the head of your directors section, but I'm not clear on how expensive this is as it appears to involve a file lookup for normal local messages (yes, in spite of the except_domains line). This isn't a big deal if the bulk of your mail is for the virtual domains; if not you can avoid it by putting these directors at the end of the directors section, but you must then add the line domains = localhost:my.net.au:*.my.net.au to the other directors, so that they are not matched for your virtual domains. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: still problems w/ horde/imp
On Wed, Aug 16, 2000 at 08:45:27AM -0300, Mario Olimpio de Menezes wrote Hi Ashley, On Tue, 15 Aug 2000, Ashley Clark wrote: It should have at least: extension=pgsql.so extension=imap.so yes! I have all these lines and more. and possibly: extension=ldap.so Did you restart Apache? /etc/init.d/apache restart yes! I even restarted the system! Sounds right, I've got php3, php3-{imap,ldap,pgsql}, postgresql-7.0.2, horde, imp. however, I still can't use postgresql-7.0.2 with horde/imp :-( phpinfo() doesn't show any support for postgresql even with the pgsql.so extension in the php3.ini file. I have no idea what's happening here! anybody? Two possibilities: - php3 can be run either as an interpreter or as an Apache module, and each has their own config file: /etc/php3/{cgi,apache}/php3.ini Try checking the *other* config file, in case you are mistaken as to which you are using. - Potato's php3-pgsql is linked against libpq from Postgresql 6.5.3, which may not be compatible with Postgresql 7.0.2; if you are using the potato php3-pgsql, have you checked for success reports with this combination from other users? John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: debugfs and superblock
On Wed, Aug 16, 2000 at 05:23:48PM +0200, Ron Rademaker wrote I want to change the filesystem size (according to the superblock), because it's different from the physical size. I guess I should do this with debugfs (or perhaps fsck), if anybody has any ideas on how to do this I'd like to hear them. I'd normally look to use ext2resize for this sort of thing, but I'm not sure whether it will work (e.g., whether it would require updating any areas beyond the end of your partition); but then, I'm not convinced that just updating the superblock would work, either. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: Exim configuration
On Wed, Aug 16, 2000 at 05:26:53PM -0300, Alberto Pereira wrote Hi All, Someone knows how i can set exim to when a e-mail arrives for a unknow user its transfer to a mailbox of a know user. I try to put lsearch* in exim.conf and *: userknow in the aliases file. But all e-mails comes its tranfer to this userknow!! I want to transfer only the unknow users to this user to the all. You previously posted this question about a week ago, and I posted a solution on 11 Aug that used the smart_user driver. Did that not work, or not get to you? John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: Routing Problem
On Mon, Aug 14, 2000 at 11:04:27AM +0100, Sian Leitch wrote On Sun, Aug 13, 2000 at 02:27:47AM +0930, John Pearson wrote: I'd check your ipchains/ipfwadm rules. If you're running kernel 2.0.x, what does the output of # ipfwadm -I -l -e # ipfwadm -O -l -e # ipfwadm -F -l -e look like? If you're running kernel 2.2.x, what does the output of # ipchains -L -v look like? I'm running a Dell Inspiron 3800 with a newly-compiled 2.2.17 kernel. I tried running `ipchains -L -v' and got the following output: ipchains: Incompatible with this kernel According to `dpkg', ipchains is one of the TCP/IP programs supplied with the `netbase' package of which I have version 3.18-4 (potato). Does anybody know which version of `ipchains' is suitable for a 2.2.17 kernel? IPChains is it for kernel 2.2.x. This suggests that your kernel doesn't include IPChains support, in which case this isn't the source of your problem. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: using outgoing smtp server with mh (where's the fm? :)
On Mon, Aug 14, 2000 at 06:41:02PM -0400, Richard E. Hawkins wrote There's probably an easy, stock answer for this, but I'm not finding it. To keep the network folks happy, this machine needs to be incapable of sending mail on its own. I've reconfigured exim to deliver locally only, but I can't find what I need to do to have mail from mh sent to my outgoing smtp server with the appropriate name and password. Could someone point me to the fm I should rt? :) Rather than configuring exim to only deliver locally, you should configure it to send all non-local mail via a smarthost, and specify your outgoing smtp server as the smarthost. You can do this part easily enough using eximconfig. It may be possible to do this using mh alone but I'd be surprised if it went so far as to support smtp authentication on outgoing mail, which is an unusual requirement in the Unix world. Connecting to an outgoing SMTP server amounts to Sending mail on its own; I suspect that they mean sending mail directly to outside hosts, or via hosts other than their outgoing SMTP server, so using exim with them as a smarthost shouldn't be a problem (just make sure you aren't an open relay). Because your outgoing server requires authentication, you should read the section of the exim docs that deals with this (SMTP Authentication). You will need to find out what authorization type your smtp server uses, and configure exim to use appropriate credentials for that server (AFAICT the standard Potato exim supports the MD5 and PlainText authentication methods). I haven't done this myself (I've never seen the point of SMTP authentication, except perhaps to work around braindamaged systems that have no other way of controlling connections), but it seems to be reasonably well documented in the exim-doc package. Good luck, John Pearson. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: ran out of input data
On Mon, Aug 14, 2000 at 11:11:39PM -0500, Kent West wrote Andrew Martin wrote: I am trying, to no avail, to install Debian 2.2 thru Windows. I haven't been able to get my CD drive to work since it is not a conventional IDE drive. It attaches to its own sound card. Anyhow 1: I have defrag'ed the drive 2: I turned off the virtual memory in win 95 3: I fips to split the partition 4: I rebooted the computer and re-enabled the virtual memory in win95 5: I ran scan disk and everything looked ok 6: I copyed the following files to the desktop boot.bat linux loadlin.exe root.bin These were all copyed from the install folder on the CD 7: I right clicked on the boot.bat file 8: Went to properties 9: Clicked on the advanced button 10: Checked the ms-dos mode box 11: Exited the properties menu 12: I then double clicked on the MS-DOS shortcut to boot.bat 13: It gave me a warning that this was going to run in dos 14: I clicked ok and below is what I got uncompressing linux ran out of input data system halted I am installing this on a 66mhz 486DX2 with 16MB of memory. From what I have read that should be enough to do it. Should I quit trying to use my current CD drive and just spring for a new IDE drive? This is so frustrating. If anyone wants to they can e-mail me direct at [EMAIL PROTECTED] Thanks, Andrew Can you get to your CD-ROM from DOS only? Rather than trying to run boot.bat from a Windows icon, exit Windows with the MS-DOS Mode option. If the DOS drivers for your CD are in CONFIG.DOS (SYS) and AUTOEXEC.DOS (BAT), you should be able to access your CD-ROM drive directly and install from there. Alternatively, do the following: Create a directory on C: such as C:\DEBIAN. Place the files you mentioned in this directory instead of on the desktop. Also, you might want to include the resc1440.bin and drv1440.bin files. I'm not positive they're needed, but I THINK one or both of them is needed. Then, as above, instead of trying to run boot.bat from Windows, exit Windows into the MS-DOS Mode. Then cd \debian to get into the C:\DEBIAN directory, and then enter boot. If your CDROM isn't an IDE or SCSI CDROM, or it's connected to an ISA/PNP card, then you'll definitely need rescue.bin and the file drivers.tgz, so that you can load the driver required for your CDROM and continue from there via CDROM. You may also want to consider putting base2_2.tgz there, so that you don't have to rely on your CDROM drive to install the base system. That'll probably get you farther along. I think that when you run boot.bat from the desktop, Windows uses C:\WINDOWS as the working directory instead of C:\WINDOWS\[Profiles\UserName\]DESKTOP, which is why the installation can't find the files that it needs. Putting everything in C:\DEBIAN should solve this problem. You also should try to identify your CDROM type before proceeding, so that you don't end up sweating over it trying to figure out which of the half-dozen or so likely drivers is appropriate while your machine is no longer running Windows, but not yet running Debian. What make model is it? John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: can't ftp through IP Masq
On Tue, Aug 15, 2000 at 03:37:30AM -0500, John Reinke wrote I did some research, and the ip_masq_ftp.o module is automatically compiled when CONFIG_IP_MASQUERADE_MOD is selected during kernel config. I already have it selected, and the file is in my modules directory. And like I mentioned previously, I've tried changing the passive settings on the ftp clients. I re-read the IP Masq howto at http://ipmasq.cjb.net and I had included everything I needed to have in the kernel. I had compiled everything into the kernel, with nothing compiled as modules - that shouldn't hurt, should it? There were a few items that I don't have which were shown at that web site. They put a lot of settings in the /etc/rc.d/rc.firewall file on a RedHat system. Where would I put that in my potato system, in case some of those settings help? Here's what my problem is (for those just joining): I have IP Masqing set up on a potato system, and everything works through it except ftp. The ftp clients on machines on the private network connect to external sites, but never are able to get a listing of the files or even retrieve files from those systems. So, just to check... if you go # lsmod does it list ip_masq_ftp? John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: ran out of input data
On Thu, Aug 14, 2008 at 04:18:14PM -0500, Andrew Martin wrote [snip] If your CDROM isn't an IDE or SCSI CDROM, or it's connected to an ISA/PNP card, then you'll definitely need rescue.bin and the file drivers.tgz, so that you can load the driver required for your CDROM and continue from there via CDROM. You may also want to consider putting base2_2.tgz there, so that you don't have to rely on your CDROM drive to install the base system. That'll probably get you farther along. I think that when you run boot.bat from the desktop, Windows uses C:\WINDOWS as the working directory instead of C:\WINDOWS\[Profiles\UserName\]DESKTOP, which is why the installation can't find the files that it needs. Putting everything in C:\DEBIAN should solve this problem. You also should try to identify your CDROM type before proceeding, so that you don't end up sweating over it trying to figure out which of the half-dozen or so likely drivers is appropriate while your machine is no longer running Windows, but not yet running Debian. What make model is it? [snip] OkI tried to but all of the files in my new folder on c drive. I also changed my drive letter for the cd in config.sys I was able to boot from either one but still got the ran out of input data message. My cd drive is a Sony CDU31A-02. It isn't a regular IDE drive. It has it's own card which is also the sound card. I didn't try putting base2_2.tgz in the folder. I was pretty tired when I did this. Is that the file that contains the kernal? If it is then that should get me to where I can install drivers for the cd right? Andrew You will need to load the cdu31a driver durinmg installation before you can access files from the CDs. Because of the order that the installer does things, you must load both the kernel and drivers from somewhere other than your CD; you should then be able to use the CD for installing the base system and packages. The kernel is normally loaded from a rescue floppy, from the CD or from your hard disk; you can't use the CD, so you must use either a rescue floppy or copy 'rescue.bin' to your hard disk, as you have done. The drivers get loaded either from a set of floppy disks (the number of disks depends on the kernel flavour you use), or the 'drivers.tgz' file that you have copied to your hard disk. 'base2_2.tgz' is an archive containing the base system, and you *should* be able to access that from you CDROM once you've loaded the CDU31a driver module. If your system is very low on memory (under 6MB installed) then you may have trouble running the Debian installer, or booting from a standard boot disk; however you have 16MB, so this shouldn't be a problem. It may be that whatever default settings your system uses for MS-DOS programs interfere with loadlin's operation; one thing you may like to try is exiting to MS-DOS and running the program from there rather than running it as a full-screen program under Windows: - From the Start menu, select Shutdown... - Choose Exit to MS-DOS - At the DOS prompt, assuming you copied all of the files to c:\debian: C c: C cd \debian C boot.bat Alternatively, try making boot root floppies and seeing how that works; it is possible that Loadlin has issues with your BIOS. HTH, John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: ran out of input data
On Tue, Aug 15, 2000 at 09:11:42PM -0500, Andrew Martin wrote Well I broke down last night and went to wal-mart at 3:00am and bought a IDE cd rom drive. So now everything will install but..The instructions in the book that came with the dist. said to make a boot floppy and then reboot the system. I did that. Now it wants me to type in my root password and the key board is dead. No action what so ever. In the setup I picked the default keyboard settings. How can I go back and change them? Andrew First up it won't echo the characters you type at the password prompt, so just because you don't see anything happening doesn't mean it isn't seeing what you type. Type your password, then hit Return, and repeat as requested. If even Return doesn't do anything, then it probably isn't a keyboard configuration problem; the keyboard configuration option affects what symbols go with which keys (so chooisng the wrong keyboard will mean you may not type what you think ypu're typing), but it shouldn't stop it dead; that's more likely to be caused by a loose keyboard cable, or a faulty keyboard. If you *do* need to change your keyboard configuration the official command to use is kbdconfig, but you need to get to a root prompt to do that. HTH, John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: still unable to get kernel source
On Mon, Aug 14, 2000 at 02:27:26PM +0200, Florian Friesdorf wrote On Sun, Aug 13, 2000 at 10:21:43PM -0500, John Reinke wrote: Dumb question: should all the documentation within /usr/share/doc/kernel-package be .gz files? Is there an easier way to view all the documenation here without having to gunzip them first? I think zless displays either gzipped and plan text files. So if you put an alias less=zless into somewhere it gets executed, your less will work for both. Using zless works, but you lose functionality by going through a wrapper like that. Even better, put eval `lessfile` in /etc/profile or .bashrc and just plain $ less myfile.{tar,gz,deb,...} becomes almost magical. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: Problem with potato
On Sun, Aug 13, 2000 at 10:23:41AM -0400, Brian Schramm wrote I just got through installing Potato on my HP Pavilion 133 computer with 80 meg of ram and a 8gig HD. When I come back to it from letting it sit idle for about 8 hours this is on the tty1 screen: DEBUG: Pages 0: Changed 0, Reapped 0, Empty 0, New 0; Tup 0: Vac 0, Keep/VTL 0/0, Crash 0, UnUsed 0, MinLen 0, MaxLen 0; Re-using: Free/Avail. Space 0/0; EndEmpty/Avail. Pages 0/0. Elapsed 0/0 sec. DEBUG: --Relation pg_rewrite-- DEBUG: Pages 3: Changed 0, Reapped 0, Empty 0, New 0; Tup 5: Vac 0, Keep/VTL 0/0, Crash 0, UnUsed 0, MinLen 2568, MaxLen 4609; Re-using: Free/Avail. Space 0/0; EndEmpty/Avail. Pages 0/0. Elapsed 0/0 sec. DEBUG: [snip] I'm now running the 'woody' version of Postgresql, and on my setup this is fixed by setting PGDEBUG=0 (or, for that matter, PGDEBUG=anything). If PGDEBUG or PGECHO are set the Postmaster's output is redirected to a log file, whereas if they aren't then the output is not so redirected, presumably in the mistaken belief that if you don't set those options then no output is produced. The offending code is in /usr/lib/postgresql/bin/postgresql-startup, which on my system includes this chunk starting at line 232: # Ready to go: stand clear... echo Starting PostgreSQL postmaster cd ${POSTGRES_HOME} if [ -n ${DEBUGLEVEL} -o ${PGECHO} = yes ] then touch ${POSTGRES_LOG:=/var/log/postgres.log} chown postgres.postgres ${POSTGRES_LOG} chmod 660 ${POSTGRES_LOG} su postgres -c ${POSTMASTER} -b ${POSTGRES} ${BUFFERS} ${BACKENDOPT} \ -D ${PGDATA} ${DEBUGLEVEL} ${TCP} ${PORT} ${OPTIONS} \ ${POSTGRES_LOG} 21 else su postgres -c ${POSTMASTER} -b ${POSTGRES} ${BUFFERS} ${BACKENDOPT} \ -D ${PGDATA} ${TCP} ${PORT} ${OPTIONS} fi HTH, John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: Protecting a single file?
On Sat, Aug 12, 2000 at 10:50:07AM +0200, Oliver Schoenknecht wrote Hey everyone, just a short question : I have a certain file where php-passwords are stored in - the problem is that it could usually read by anyone. Does anyone of you know how to protect this single file without doing a .htaccess on the whole folder and keeping the chown-rights of 777 ? Is it even possible to set a password or so on this file ? Any help is appreciated very well :-) ! The short answer is, don't store documents like this in directories that your web server can serve files from. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: Routing Problem
On Sat, Aug 12, 2000 at 04:37:50AM -0700, Peter Welte wrote hey there... I have a linux computer that is supped to act as a gateway to a school network and the internet for some linux clients, but im having this problem right now where the gateway itself can't even ping another computer on the school network. The gateway's ip address is 192.168.1.12 (er, that is eth0, which is connected to the school network) and the computer i am trying to ping is 192.168.1.5. i cannot figure out what is wrong with my network setup, so if anyone out there could help me, i would greatly appreciate it! [snip fairly normal looking ifconfig route details] ##and here is me pinging an internal network comp: PING 192.168.1.12 (192.168.1.12): 56 data bytes 64 bytes from 192.168.1.12: icmp_seq=0 ttl=255 time=2.6 ms 64 bytes from 192.168.1.12: icmp_seq=1 ttl=255 time=1.5 ms 64 bytes from 192.168.1.12: icmp_seq=2 ttl=255 time=1.5 ms 64 bytes from 192.168.1.12: icmp_seq=3 ttl=255 time=1.5 ms --- 192.168.1.12 ping statistics --- 4 packets transmitted, 4 packets received, 0% packet loss round-trip min/avg/max = 1.5/1.7/2.6 ms This is actually you pinging your own NIC according to the ifconfig output you posted, so it proves little. ##And finally, here is me trying to ping the computer on the school network: PING 192.168.1.5 (192.168.1.5): 56 data bytes ping: sendto: Operation not permitted ping: wrote 192.168.1.5 64 chars, ret=-1 ping: sendto: Operation not permitted ping: wrote 192.168.1.5 64 chars, ret=-1 --- 192.168.1.5 ping statistics --- 2 packets transmitted, 0 packets received, 100% packet loss I'd check your ipchains/ipfwadm rules. If you're running kernel 2.0.x, what does the output of # ipfwadm -I -l -e # ipfwadm -O -l -e # ipfwadm -F -l -e look like? If you're running kernel 2.2.x, what does the output of # ipchains -L -v look like? John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: Exim
On Thu, Aug 10, 2000 at 03:51:10PM -0300, Alberto Pereira wrote I see the question Q0432 but, my problem is in Q0401. But my versionof Exim is .2.05-2 I put this on the exim.conf: system_aliases: driver = aliasfile domains = z10.com.br file = /etc/aliases search_type = lsearch* and put this on de final line of aliasfile: *: [EMAIL PROTECTED] But ALL mails are delivery to [EMAIL PROTECTED] I want to delivery only the unknowuser. Can someone help me? OK. First, take out the wildcard line from /etc/aliases (if you haven't already), and delete the domains = z10.com.br line from your system_aliases director, and reload the exim configuration with the command # /etc/init.d/exim reload so that normal delivery to real users works again. Check that it's basically doing the right thing for both local and remote addresses with a command like # exim -bt root [EMAIL PROTECTED] Then, add something like this to the end of the DIRECTORS section of /etv/exim.conf, just before the end statement: unknown_user: driver=smartuser new_address=admin Reload the exim configuration with # /etc/init.d/exim reload and then try it out with something like # exim -bt nonesuch John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: restricting user in HOME using ProFTP
On Wed, Aug 09, 2000 at 06:40:51AM -0700, Paulo Henrique Baptista de Oliveira wrote Hi all! I'm trying to restrict my FTP users in own directories using ProFTP. I used the configuration examples found in ProFTP's page but something is wrong. AFAICT, all you need to do is add DefaultRoot ~ somehwere near the top of /etc/proftpd.conf and run # /etc/init.d/proftpd reload They should then be locked into their home directories; it seems to work for me. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: Has Corel been violating the GPL for approx 6 months?
On Fri, Aug 11, 2000 at 06:12:35AM +, Mr Smith wrote Has former copies of the Corel LinuxOS Open Circulation CD-ROM violated the General Public License? This seems like a straight forward question but one that Corel has beat around the bush in answering. This is going to be a rather lengthy email so I will break it into six major parts: 1) Questioning Corel's distribution practices in January 2000 2) Questioning Corel's distribution practices in April 2000 3) Corel declairs the situation corrected in July 2000 4) Corel's excuses for distribution practices 5) What Corel doesn't seem to be willing to dispute 6) Plan of action to bring about a complette correction [snip] One thing you didn't discuss in your post was what remedy, if any, you proposed to Corel before you decided legal action was appropriate, and how they responded. As to their failing to apologize it may be graceless and unflattering, but provided that they rectified their original error it's hardly actionable. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services Do I *look* like a lawyer?
Re: Problems with XF86Config
On Fri, Aug 11, 2000 at 04:07:09AM -0700, [EMAIL PROTECTED] wrote I want to install Linux on two HDs, one is an old 850 MB Conner, from which all traces of Gates Co.have been banished and this will be pure Linux, and a new 6.5 GB HD which is dual-boot Win95 and Linux. I downloaded Star Office in both Win95 and Linux forms, and have been using it under Win 95 for several months and I never want to have MS Office on my box again! Gates does not like Star Office, and the PC goes bananas regularly, but Star is crash-proof and always saves correctly, and nothing gets lost. I have tried SuSe and Red Hat ( Mandrake), but I much prefer the Debian basic installation. Dselect on the Conner is OK. The problem: Debian can't find an XF86 config file, the result is that X-Windows cannot start, although Dselect has installed everything else. Unlike Suse YAST Debian never asked me anything about my monitor, and never mentioned setting up an XF86 config file. Please say how I can create the missing file, how I get it into the correct directory and let startx work? Debian doesn't configure X for you, but it does provide two tools that walk you through it: XF86Setup (in the xf86setup package) and xf86config (in the xserver-common package). XF86Setup uses 16 color VGA mode, so that's probably not an option for your system. Run the command # apt-get install xserver-mono xfonts-scalable xfonts-75dpi to install an appropriate Xserver and some fonts, and run # xf86config and tell it about your hardware. Be sure to choose the mono Xserver when it asks you which one to use (it'll probably recommend XServer-SVGA, but that doesn't do mono). You'll probably also want to run a command like # apt-get install icewm wmaker xterm to install some other window managers and xterm, if you haven't already done so. Monitor: Triumph-Adler monchrome 14 Video Bandwith more than 30 Mhz analog (80x 25) line frequency 31467 Hz frame frequency 60-70 Hz Video card - Venus VGA Tests Actual mode (3h Text) 80 x 25 V H 70.00KHz31.43Khz mode 12h (graphic 640 x 480) 64.08Khz31.54 Khz I hope you can help! As I have no Internet connection here at home, I have to prepare this message at home under Star Office and then cart this on diskette to the public library and load it under *** MS-Office to send it under Netscape This all supposes that you have a Debian CD or some other medium that you can use as a source of Debian packages, and that you have configured apt to use it. If you haven't, then this will involve a *lot* of to-ing and fro-ing with *many* floppy disks. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: Installation: error messages
On Fri, Aug 11, 2000 at 04:43:41PM +0200, Helgi Ćrn wrote Hello Debians! I just installed 2.1 slink (one more time) and got these messages at the end of the dselect package installation: Errors were encountered while processing gs gv pstotext pstoedit ghostview Installation script returned error exit status 1. This happens because the order that dselect unpacks and configures packages in is sometimes not quite right. If you see messages like this after selecting Install, try selecting Configure, then Remove, and then Install again. That will usually fix things. mv: /var/lib/pdmenu//pdmenurc_auto.new : No such file or directory Update-menus: Cannot remove lockfile /var/run/update-menus.pid Not necessarily anything to worry about, providing the menus work :-. I would be very grateful if someone could inform me about the importance of these messages and in what way they effect my system. I have installed 2.1 several times but never managed to configure it to my satisfaction, no internet connection and no sound for example, and i have REALLY tried. I really would like to use Debian/GNU instead of SuSE which i otherwise have as my default OS. Good luck, John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: More on compiling Galeon
On Wed, Aug 09, 2000 at 07:54:56AM -0400, Jonathan Markevich wrote I don't know if I missed it in the thread, but I can't figure out how to compile Galeon either. I have the Potato version of Mozilla installed (M16, I understand) but when I ./configure, I get configure: error: Could not find the gnomeConf.sh file that is generated by gnome-libs install gnome-libs install? Is that a command; or a process? I installed the libglade and gnome glade -dev libraries and it doesn't make a difference. Where is this file? # zgrep gnomeConf.sh debian/dists/potato/Contents-i386.gz usr/lib/gnomeConf.sh devel/libgnome-dev # Looks like it's in libgnome-dev. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: setting up x terminal workstation
On Sat, Aug 05, 2000 at 06:57:43PM -0500, Chris Hoover wrote Can someone point me to some sites that explain how to get an older computer working as a remote x-term/workstation. Here is what I currently have: 1 486DX/2 50 laptop w/ 20 megs - This has a complete potato install with a working X server. I would like to convert this laptop into a remote X workstation with everything possible (X wise) running off of my local server. How can I do this? Easy. Get X configured locally (so you can use startx to go into graphics mode, and have a working mouse cursor). Remove XDM from your 486, if you have it installed; install it on your server (which doesn't need a working X server) if it isn't already. On the server, edit /etc/X11/xdm/Xaccess so it contains lines like *.localnet *.localnet CHOOSER BROADCAST substituting the name of your network for 'localnet' (or, the name of your client for '*.localnet'). Don't just use *, and remember that X traffic is unencrypted: don't do this if you don't trust your local network, don't use a wildcard that matches untrusted machines, and don't use a wildcard that matches machines the path to which includes untrusted networks or machines. The first line says that XDM should allow the specified machines to login, and the second allows the specified machines to run a chooser (which is irrelevant, if you only have one workstation to login to). If you've had to make any changes, run # /etc/init.d/xdm reload so that XDM sees them. You can use GDM or WDM instead, but then you need to find and appropriately edit their equivalent of XDM's Xaccess file. Run this command as root on your 486: # /usr/bin/X11/X :0 -query server.localnet /var/log/x0.server.log substituting the name of your server for 'server.localnet'; you should get an XDM prompt from your server on your local display. If you have more than one server that you might want to login to, you can use this command instead: # /usr/bin/X11/X :0 -indirect server.localnet /var/log/x0.server.log This should get your server to present you with a menu of hosts to choose from. Note that currently this doesn't work if the server is running gdm or wdm. To make it more permanent, add a line to /etc/inittab like this: x0:3:respawn:/usr/bin/X11/X :0 -query server.localnet This sets the local Xserver to run in runlevel 3; test it with # telinit 3 and, if it works, you can change :3: to :2: in inittab to make it start in Debian's default runlevel. If you're already running a local Xserver on your 486, change the X :0 to X :1 (or whatever) so that they each use a different display. Hope this helps, John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: Buggered up my router somehow.
On Fri, Aug 04, 2000 at 09:55:25AM -0600, Adam Scriven - Lore wrote Ok, by some great and wonderful streak of stupidity, I seem to have somehow completely fscked up my router. It's hooked up to an ADSL modem, running PPPoE (Roaring Penguin), and that part looks like it's working great. I've got 2 other network cards, both 3Com 905B. I have the 3c59x module loading with modprobe, and I've checked ifconfig, and both cards look to be setup correctly. eth0 is 192.168.0.1, and eth1 is 192.168.1.1 (Incedentally, eth2 is an NE2k-pci card, for the PPPoE client). I can see the world just fine from the router (I'm telnetting to an ISP where my maail is hosted to send out this message, and I'm on the router now), but I can't ping anything on my 192.168.0.0 network (192.168.1.0 isn't used yet). The route command returns: Kernel IP routing table Destination Gateway Genmask Flags Metric Ref Use Iface ADSL-NAME * 255.255.255.255 UH0 0 0 ppp0 192.168.1.0 * 255.255.255.0 U 0 0 0 eth1 192.168.0.0 * 255.255.255.0 U 0 0 0 eth0 default ADSL-NAME 0.0.0.0 UG0 0 0 ppp0 (I had to tyype this in by hand, so any formatting problems are mine.) This all looks OK to me, but pinging just gives 100% packet loss, and traceroute to 192.168.0.2 from the router gives: traceroute: sendto: Operation not permitted 1 traceroute: wrote 192.168.0.2 38 chars, ret=-1 Assuming your router is a Linux box: This means that your kernel is dropping the packets, probably due to an IPChains rule or because forwarding is disabled. Check # ipchains -L input # ipchains -L output # ipchains -L forward and # cat /proc/sys/net/ipv4/ip_forward (should be 1) and # cat /proc/sys/net/ipv4/conf/eth0/forwarding # cat /proc/sys/net/ipv4/conf/eth1/forwarding (should both be 1). As a last resort, you can also try # cat /proc/sys/net/ipv4/conf/eth0/rp_filter # cat /proc/sys/net/ipv4/conf/eth1/rp_filter These are both normally 1. If your router receives packets for forwarding from an address that doesn't match the network address of the interface they are received on and this is set to 1, the kernel drops the packets; this is intended to prevent spoofing. You shouldn't normally have to play with this, but I found that I had to when (e.g.) I had multiple networks on the same cable. I've never used PPPoE and I doubt that's it, but as I say, it's worth checking as a last resort. You can change any of these parameters with echo, e.g.: # echo 1 /proc/sys/net/ipv4/ip_forward OH, I also meant to ask. I used to use linuxconf on my RH systems. Is there an equiv. for Debian? Something better, perhaps? Well, there's always Linuxconf. Never used it myself, but it wouldn't be there if it didn't work at all. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: more on syslogd remote logging
On Wed, Aug 02, 2000 at 10:45:59PM -0700, [EMAIL PROTECTED] wrote ok, i got syslogd working it is recieving log entries from my router, now im curious how i would redirect those to a dedicated file? i tried various things in /etc/syslog.conf and the log file is empty still. I'd like to redirect everything from 10.10.10.1 to /var/log/dsl.log sample log entries: Aug 2 15:26:25 10.10.10.1 000:23:31:30 ATMInfo Wan0 Up, 640 Kbps Down, 544 Kbps Up, 340 Baud^M Aug 2 15:26:25 10.10.10.1 000:23:31:30 ATMInfo Wan0 Up*, +11.3 dB TX Power, +18.7 dB Rem TX Power, 42 dB RX Gain, No Change Margin^M Aug 2 15:26:25 10.10.10.1 000:23:31:30 PPPInfo PPP Up Event on wan0-0^M Aug 2 15:26:25 10.10.10.1 000:23:31:30 ATMInfo Wan0 Up*, 23 dB Line Quality^M Aug 2 15:26:40 10.10.10.1 000:23:31:45 SERIAL Info Serial Connection Timeout^M Aug 2 15:28:05 10.10.10.1 000:23:33:10 PPPInfo PPP Down Event on wan0-0^M any ideas ?? Use /etc/syslog.conf to control where logging goes. This allows you to specify things by facility and priority. Your router should allow you specify the syslog facility used for messages, probably with a config statement like logging facility local3 if it's a Cisco (it's on the documentation CD which you should have). Edit /etc/syslog.conf to add a line like this: local3.* -/var/log/dsl.log If you want quick console access to the messages and aren't too fussed about other peopel seeing them, you can also use a line like local3.* /dev/tty12 to direct them to an unused vt as well. You may also want to add local3.none to some of the other lines, if they use a wildcard for the facility and you don't want those lines to catch messages from your router. Then run # /etc/init.d/sysklogd reload and tell your router to use syslog facility local3 (or whatever you chose). John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: noise from monitor, HELP!
On Wed, Aug 02, 2000 at 09:57:12AM -0700, kmself@ix.netcom.com wrote On Tue, Aug 01, 2000 at 11:24:59PM -0700, Jaye Inabnit ke6sls wrote: On Tue, 01 Aug 2000, Ron Farrer wrote: I've been running Debian for years and my monitor (20 IBM P200) has been working fine with it's current setting for months. Now all of a sudden I get a VERY high pitched noise from it when in X. It does not do this on the console! It has become more and more frequent and is starting to drive me nuts (well more then usual :) I know it's the monitor because I can turn it off while it's doing it and the noise stops. Does anyone know what the problem might be? FYI I am running potato using a Matrox Mill. (original) with 8MB memory @ 1024 x 768. Ron, I have a 17 'emachines' monitor. It was making some pretty heavy noise too when I first fired up x after running XF86Setup. I don't know why but in windoz it was quiet but x was giving it grief.. I finally tweaked the settings and cured the noise (Think I increased the res).. I'm sure it's probably unrelated, but I recall seeing upgrades to x the last two times I ran the 'apt-get update/upgrade' so perhaps you need to tweak things on ur end... Just a thought. Given this, it's possible that modifications to video settings via xvidtune might dispose of harmonics leading to monitor hum, if the cause is resonance rather than failing bonds or fasteners. Haven't tried it myself, but the logic seems to work mentally. Actually, the cause is usually a combination of the two. The yoke should be designed to provide good damping across the usable range of the monitor, but as adhesive perishes and things get freed up modes start appearing which may be close enough to the scan frequencies you actually use (or a subharmonic thereof) to cause hooting. While hitting a monitor may provide a short-term fix, I'd suspect it would compound the problem of a degrading joint. Not generally recommended. I don't know what he paid for his 20 monitor but I could take a guess; for myself, I try to avoid thumping things that cost that much. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: noise from monitor, HELP!
On Tue, Aug 01, 2000 at 07:07:18PM -0700, Ron Farrer wrote I've been running Debian for years and my monitor (20 IBM P200) has been working fine with it's current setting for months. Now all of a sudden I get a VERY high pitched noise from it when in X. It does not do this on the console! It has become more and more frequent and is starting to drive me nuts (well more then usual :) I know it's the monitor because I can turn it off while it's doing it and the noise stops. Does anyone know what the problem might be? If it is a regular CRT-style monitor, there are wire coils wrapped around the CRT to deflect the electron beam to provide horizontal vertical deflection; these are probably glued in place (or at least, in bundles) using epoxy resin or something similar. Probably, the epoxy has cracked or come loose from whatever it's anchored to at some point and what you can hear is some or all of the windings on the horizontal deflection coil rattling back and forth in time to the horizontal scan, in accordance with Newton's laws (For every action there is an equal and opposite reaction). Just maybe, it's some other part of your monitor's yoke doing the same thing. It's irritating as hell if you can hear it, but it shouldn't affect the monitor's performance or reliability. Possible solutions: - Pay someone to fix it. If you take it in for a service, be *very*clear* about the problem or it probably won't get fixed (chances are, most of their techs won't be able to hear it). - Swap monitors with someone who can't hear it; - Listen to a few hours of something load every day until the problem goes away (what most people, wittingly or not, do); - Get a new monitor (big bucks for something that size). John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: Help(!) with Slink install -- disk repartitioning problem
On Wed, Jul 26, 2000 at 11:41:55PM -0400, Christopher Lee wrote *cc me on any replys, since I am not subscribed to debian-user* Please read this if you know something about hard-disk partitioning, and think you can tell us where the mystery 2 Gigs went. My wife is a little stressed-out that I may be messing-up her computer I'm installing Slink on my wife's laptop (I have the CDs lying around, and will upgrade it to Potato later). She has a HP Omnibook 4150 Laptop with a 10Gig HD running Windows 95, which I am modifying to work as a dual-boot machine. [snip] The long and the short of it is that cfdisk doesn't deal properly with disks over 8Gb; you should use the comand line fdisk to create the partition, just as you used it to delete the fips'd partition. Don't forget to create a swap partition, as well as a Linux partition. If you do so, you may then want to select Partition a Hard Disk but quit cfdisk without making any changes, to con the installer into taking you to the next step (Initialize a swap partition). Because you are installing beyond the area supporeted by conventional BIOSes, booting Linux directly from the hard disk may take a little work (booting from Windows using loadlin should be fine); be sure to make a boot floppy when prompted rather than/instead of installing Lilo on your hard disk, so that you can reboot your new system. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: Unidentified subject!
On Thu, Jul 27, 2000 at 01:32:56PM -0400, David Teague wrote Folks, I have a question from a buddy who asks the following question, which I could not help him with. I have had wonderful support, I hope one of you can help him. Send flames to me for any lack of information, I'll ask him for what you need. It ain't his fault ;) Please send answers to both of us. [EMAIL PROTECTED], [EMAIL PROTECTED] Victor asks: . I am trying to compile a very minimal Linux kernel for a 486 with . limited memory and disk space. It will just be a simple networking . gateway. . I have compiled a monolithic kernel that seems to run . ok but it has a small problem... . The rc.sysinit script invokes the command: . mount -a -t nonfs, smbfs, ncpfs, proc . This produces an error message: . mount: fs type devpts not supported by kernel . . This error message results from attempting . to mount the nonfs filesystem mentioned in this . mount command. . . The mount man page says that something is mounted on /dev/pts but . I don't know what that is. . . Can you tell me what filesystem, or other support . needs to be included in the kernel configuration to . fix this, or is it even needed? Devpts is the pseudofs used by Unix98 PTY's. You don't need it, and not using them may save you a little in memory/kernel size. Read /usr/src/linux/Documentation/Changes for a little more info. To avoid the message (which isn't really doing any harm) comment out the line in /etc/fstab that reads (approximately) none /dev/pts devpts gid=5,mode=620 0 0 John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: terminal goes funky
On Tue, Jul 25, 2000 at 10:19:23PM +0200, Sven Burgener wrote On Tue, Jul 25, 2000 at 10:05:00PM +0200, Morten Liebach wrote: On Tue, Jul 25, 2000 at 09:38:07PM +0200, Sven Burgener wrote: [snip] It seems to display things fine, but just with weird characters instead of proper ones. Some characters are fine though, like the GNU part of the login prompt. :) [snip] He, I think this is interesting. I haven't tried it myself in Linux, but it has been a problem in some snapshots of OpenBSD that I have used, and it surfaced on a mailinglist that it's an age-old artifact in UNIX. It seems so, particularly as I saw someone undoing this problem with a key-combination. This person comes from UNIX and not Linux. Porbably, at the command prompt, they typed $ echo ^V^O (that's CTRL+v, CTRL+o) That plus $ reset fixes things in every Linux I've seen; most recently, $ reset alone has been enough. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: init.d
On Tue, Jul 25, 2000 at 04:38:58PM +0900, keke abe wrote John Pearson wrote: Scripts in the Debian init.d directories are run using run-parts. I believe run-parts is used to run scripts in rc.boot, not those in init.d. Oops, right you are. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: init.d
On Mon, Jul 24, 2000 at 12:03:17AM +, R K wrote I've been having some problems getting startup scripts to work. Particularly with MySQL (latest binary release). From what I know, you're supposed to put the script in /etc/init.d and make a sym-link to /etc/rcX.d right? In any case, I put the script in init.d and made a symlink (S99mysql.server - ../init.d/mysql.server) in rc3.d, but nothing happens at startup. I remeber having this problem before and fixing it, but now I can't remember what I did. Running potato (not like it makes a difference, I was running slink the other time). Anyway, if I've missed some critical step, please let me know. =) Oh yeah.. the script and the symlink DO work directly from the command line. i.e. /etc/init.d/mysql.server start works. I've even tried echo testing the script, and have determined that it's not even being run at startup. Scripts in the Debian init.d directories are run using run-parts. Run-parts ignores scripts that don't conform to certain naming conventions: DESCRIPTION run-parts runs a number of scripts or programs found in a single directory directory. Filenames should consist entirely of upper and lower case letters, digits, under- scores, and hyphens. Subdirectories of directory and files with other names will be silently ignored. So, you shouldn't use . in script names. This means that if you leave any backup files lying around (e.g., mysql-server.bak, mysql-server~, #mysql-server#, etc.) they won't be inadvertently executed. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: mailbox postprocessing
On Mon, Jul 24, 2000 at 03:05:08PM -0400, Nakul Hoelz wrote Hello, I have a 2 mail servers for our domain, both running debian gnu linux. The first one has a DNS mailexchange value of 0 the other has a DNS mailchange value of 5... i.e. all email should be pouring into the main mail machine for our domain somehow though email ended up on the secondary mail server and I would like to send the email in the mailboxes of the secondary mail server to the primary mail server... How can I do that? I have so far not been able to come up with anything... shouldn't there be an easy way of postprocessing the mailbox files under /var/spool/mail/USERNAME split the messages into the original messages and to send them on to the main mailserver? Any help is appreciated. Nakul You should do two things: - set up your secondary MX so that it forwards all mail to your primary, and does no local deliveries; - look at man 1 formail, which can feed /usr/sbin/sendmail to do what you want on a mailbox-by-mailbox basis. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: cannot talk with AT modem
On Sun, Jul 23, 2000 at 11:53:02AM -0700, Tomasz Barszczak wrote Having the modem recognize the Hayes AT command language is not a reliable indicator. Many winmodem drivers have an AT command interpreter to satisfy older programs. When I bought the modem I was assured it is not a winmodem (software modem) but a hardware modem. I also searched the web and it seems it is not a winmodem. According to the USRobitics web site, this is a PCI modem device; are you able to open the case on your computer to verify that? Yes it is a PCI modem, I am sure of this. The contents of the file /proc/pci will also tell us if this is a PCI modem. I don't really understand output, but typing by hand what seems to be relevant: PCI devices found: Bus 0, device 11, function 0: Serial controller: Unknown vendor Unknown device (rev 1). Vendor id=12b9. Device id=1008. Medium devsel. IRQ 5. I/O at 0xd400 There are 5 more devices: Multimedia audio controller, vendor 1274(Ensoniq) device 5800 IDE interface, vendor 1022(AMD) device 7409 ISA bridge, vendor 1022 device 7408 PCI bridge, vendor 1022 device 7007 Host bridge, vendor 1022 device 7006 Because it's at a non-standard location, Linux will probably need you to locate it configure the driver by hand. Does running the command # setserial -b /dev/ttyS5 io 0xd400 irq 5 help at all? If it does, you can add this to /etc/init.d/setserial to get it run at each boot. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: PNP hardware and dual boot machine.
On Sat, Jul 22, 2000 at 04:06:34PM -0400, adam b. wrote isapnptools works okay, probably. I have gotten cards to recognize and load drivers, but I have never gotten them actually working before giving up. Be prepared to edit long config files from pnpdump and also you must know free IRQs, IO hexes, and Memory ranges for all your devices. They will probe, but he'll come up with like 5 or 6 viable configurations, only 1 or 2 of which might actually work (especially if you have a commercial machine-in-a-box). Also, you _may_ have to put your devices in Memory-Mapped mode (I've heard rumors to that effect). In this case, you have to find a free memory range that's in the on-limits range for Debian. Not too hard, but get the range from Debian.org before you go killing things. Well, if you already have Win95 installed it's not that bad; you can use Window's autoprobing to do the work for you. Open Control Panel - System, click on the Device Manager (? working from memory here) tab, find your PnP devices in the list, write down what works for Windows; reboot, use that configuration in /etc/isapnp.conf. I've yet to see a device that requires a memory-mapped area (other than for ROM or video memory, anyway). John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: erratic ps/2 mouse behavior
On Sat, Jul 22, 2000 at 06:06:51PM -0800, Ethan Benson wrote I have installed debian potato on a 600Mhz Athlon system and have been having problems with the mouse: the problem is most noticable in X but also occurs in the console with gpm, when i move the mouse it will at times go out of control, the pointer starts jumping around the screen extremely fast with the buttons going off randomly (i am not clicking the mouse buttons, but the windowmaker menu, and window menus appear and my workspaces are switched around when the mouse `clicks' my windowmaker clip several times, it has even dragged icons off the dock and thrown them away) the pointer usually ends up in the upper right corner of the screen but it jumps around so fast i can only see where it has been by where menus pop up and windows get closed etc. it is exceedingly annoying. i have tried disabling gpm, with no effect, furthermore this problem DOES occur in gpm without X and X without gpm, and with both running. i have also tried 3 seperate mice, 2 three button, one 2 button. all exhibit this problem. the mouse only starts going into these `fits' when i use it, it never does this when the mouse is sitting idle. it appears to go into these fits more often when i move the mouse faster, but moving the mouse at all (even slowly) still triggers it (just not quite as often) X and gpm are configured for ps/2 mouse at /dev/mouse (symlink to /dev/psaux) i am using the same configuration on two other intel systems with no such troubles. here is my gpm.conf, but note that i have disabled GPM and still had these problems in X: device=/dev/mouse responsiveness=15 repeat_type= type=ps2 append=-3 -a 3 -d 5 -l \a-zA-Z0-9_.:~/\300-\326\330-\366\370-\377\ and here is my X config: Section Pointer ProtocolPS/2 Device /dev/mouse Resolution 100 Buttons 3 EndSection the 2 seperate mice were all different brands, the one i am using now is a 3 button Logitech PS/2. i am at a loss, i have NEVER seen anything like this before, any ideas? -- Ethan Benson http://www.alaska.net/~erbenson/ It sounds like either: - Wrong mouse type (maybe, try type mman for gpm / Mouseman or MouseManPlusPS/2 in XF86Config); or - IRQ or I/O conflict: perhaps you have two devices on IRQ 12 (or wherever your PS/2 post is located), and you're waking the wrong one up? John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: help with file timestamp
On Thu, Jul 20, 2000 at 04:04:37PM +1000, Shao Zhang wrote Hi, I am having some problems here with timestamp. The system is hamm. # date Thu Jul 20 15:59:32 EST 2000 # date -u Thu Jul 20 05:59:34 UTC 2000 # touch /tmp/hello # ls -al /tmp/hello -rw-rw-r-- 1 root root 0 Jul 20 05:59 /tmp/hello As you can see /tmp/hello's timestamp is using UTC rather than EST. How do I make it using EST? This sounds more like a bug in ls or touch than anything else, as times in the filesystem will always be recorded as UTC, but displayed in localtime. Do you get similar results when you do $ date -u $ cat /dev/null /tmp/hello $ ls -al /tmp/hello If not, it looks like touch has problems. Alternatively, check the actual time and filestamp with perl: $ perl -e 'print time, \n' $ touch /tmp/hello $ perl -e 'print ((stat(/tmp/hello))[9], \n)' If they produce numbers that correlate well, then it looks like your ls is displaying dates in UTC, ignoring the timezone that date uses; could be time to upgrade, if you want to fix it... John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: UMAX 1220S, SCSI card (436P?)
On Thu, Jul 20, 2000 at 10:40:47AM -0400, Juan Alejandro Diaz MuƱoz wrote drivers please Buckleys, sorry. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: network/ethernet card configuration problem?
On Mon, Jul 17, 2000 at 02:55:54PM -0300, James Polson wrote Hi, I'm completely new to Linux, so the solution to my problem may be trivial to all of you experienced Debian Linux users (so I hope!). I'm trying to install Debian Linux (mostly with success), but I think that I'm having problems connecting with the ethernet card in my computer. The symptoms are: (1) A message during booting that is something like network is unreachable (I don't know what log file to look for to get the exact wording). (2) When I type ifconfig -a, I should see listings for two interfaces, lo and eth0, but the second one (the ethernet one) is not there. (3) Not surprisingly, telnet, ftp, etc. can't be used. (4) In the /var/log/syslog file, there is a message: cardmgr[189]: starting, version is 3.0.5 cardmgr[189]: no sockets found! cardmgr[189]: exiting I think that this problem reported is related. These lines relate to the PCMCIA support software; if you have no PCMCIA slots (which is what the card manager seems to believe) then it isn't a problem; you can ignore these messages, or be rid of them by removing the pcmcia-* packages. Information about my system: (1) I have a D-Link DE 220 card. (2) I have the NC2000 driver in /lib/modules/2.0.36/net This is the appropriate driver for the card, according to the D-Link website. The correct way to proceed depends on which version of Debian you are using (slink (2.1), potato (2.2), etc.); which is it? You can check if things are going to work out for you by trying the following commands as root: # lsmod This lists the driver modules currently loaded. If the ne or ne2k-pcidriver is loaded then you should see it listed in the output to this command, like so: ne2k-pci4136 1 Assuming it isn't loaded, try loading it by hand. If your NIC is a PCI card you should use the ne2k-pci driver, like so: # modprobe ne2k-pci PCI cards shouldn't need any extra parameters. If it's an ISA card you will need to use the ne module, and will need to pass at least the IO port as a parameter, with a command like this: # modprobe ne io=0x300 irq=10 You can skip the IRQ parameter, but if you know what it is then it makes things a little more bullet-proof. If you don't know what IO address the card is using, the Windows Device Manager (under System in Control Panels) will probably tell you. If the card is a bona-fide ISA/PNP card (as opposed to a traditional ISA card) then this will fail after a cold boot, and you will need to set the card up using isapnp before you can use it under Linux. The way modules are handled changed a couple of times in older (pre-2.1) Debian releases; you should check if the following files exist: /etc/modules.conf /etc/conf.modules /etc/modutils/ (a directory) /etc/modules This won't get you on the network, but it will tell you what you need to go further. Get back to us when you've tried this and let us know what Debian release you're using, and you should get some better-targeted help. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: telnet vs ssl
On Mon, Jul 17, 2000 at 02:09:03PM -0400, Ethan Pierce wrote Hi, im wondering about the security of telnet - people say its really unsecure...does this mean that someone would need to be running a network sniffer on my ip to catch the data? I telnet to my ISP from work, (only telnet available) then run SSH from ISP shell to home machineis this just as insecure as a straight connection via telnet right to my home machine? Maybe, it depends how your data gets from work to your ISP. If you're dialling up your ISP from work it's not so bad, because only machines on your ISP's net can see the traffic from your telnet session; if you're telnetting from your corporate LAN across the Internet to your ISP then it's no better than using Telnet alone. You don't need an SSH server to use SSH from your workplace: there are at least a couple of free Windows terminal programs (teraterm pro and putty) that can use SSH, and you could probably run them off a floppy. Of course, your workplace may be paranoid about you using or installing unauthorised software, in which case you should try to persuade them to authorise one of these (if they are across the security issues it shouldn't be a problem, provided they don't mind you using your home machine on work time). HTH, John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: network/ethernet card configuration problem?
On Tue, Jul 18, 2000 at 12:00:46PM -0300, James Polson wrote Hi, Thanks to all who answered my call for help! This message here is in response to the one from John Pearson. Date: Tue, 18 Jul 2000 13:15:56 +0930 From: John Pearson [EMAIL PROTECTED] Subject: Re: network/ethernet card configuration problem? To:debian-user@lists.debian.org The correct way to proceed depends on which version of Debian you are using (slink (2.1), potato (2.2), etc.); which is it? It is slink (2.1) You can check if things are going to work out for you by trying the following commands as root: # lsmod This lists the driver modules currently loaded. If the ne or ne2k-pcidriver is loaded then you should see it listed in the output to this command, like so: ne2k-pci4136 1 It was not loaded! Assuming it isn't loaded, try loading it by hand. If your NIC is a PCI card you should use the ne2k-pci driver, like so: # modprobe ne2k-pci PCI cards shouldn't need any extra parameters. If it's an ISA card you will need to use the ne module, and will need to pass at least the IO port as a parameter, with a command like this: # modprobe ne io=0x300 irq=10 You can skip the IRQ parameter, but if you know what it is then it makes things a little more bullet-proof. If you don't know what IO address the card is using, the Windows Device Manager (under System in Control Panels) will probably tell you. If the card is a bona-fide ISA/PNP card (as opposed to a traditional ISA card) then this will fail after a cold boot, and you will need to set the card up using isapnp before you can use it under Linux. The full name of the card is D-Link DE220 ISA PnP -- is this a bona-fide ISA/PNP adaptor? I tried using the command # modprobe ne io=0x300 irq=03(these are the proper settings) and it didn't complain. However, I infer from your remarks that I will have to use 'isapnp'. I guess that this means I will have to edit (properly!) the /etc/isapnp.conf file. I looked at the website http://www.roestock.demon.co.uk/isapnptools and it looks like I will need a line in the file like: (CONFIGURE EDI0119/236861364 ( ... etc. The code 'EDI0119' identifies the ethernet card -- but how can I find what code to use for my card? Please let me know if there is anything else I have to know about 'isapnp'. OK. Firstly, some ISA/PnP cards have a Software Configuration mode that allows you to specify a configuration using their setup disk, rather than doing PnP configuration at run time. Ignore all of the ISA/PNP stuff below if your card has that option and you're using it; skip to the line (If it isn't a PNP card, start here) in that case. isapnp is the utility you use to set up ISA/PNP cards; there is a companion utility, pnpdump, that will interrogate your ISA/PNP cards for you. First, make asure that the card isn't in use by running # lsmod to list the currently loaded modules, and use # rmmod ne # rmmod 8390 to unload the ne and 8390 modules if necessary (8390 is a low-level module that is used by the ne module to talk to ne-compatible NICs). Changing the ocnfiguration while the card is in use is officially Naughty, and may hang your machine. Run # pnpdump /etc/isapnp.conf.try This should produce a template isapnp configuration file listing your card, but with all of the lines that would actually specify a configuration commented out. Pnpdump is in the isapnptools package; install that package if it isn't already on your system. Here's an excerpt, showing a single device: START of pnpdump output fragment # # Logical device id @[EMAIL PROTECTED] # Device supports vendor reserved register @ 0x39 # Device supports vendor reserved register @ 0x3a # Device supports vendor reserved register @ 0x3c # Device supports vendor reserved register @ 0x3d # # Edit the entries below to uncomment out the configuration required. # Note that only the first value of any range is given, this may be changed if required # Don't forget to uncomment the activate (ACT Y) when happy (CONFIGURE CMI0001/16777472 (LD 3 # Multiple choice time, choose one only ! # Start dependent functions: priority preferred # Logical device decodes 16 bit IO address lines # Minimum IO base address 0x0220 # Maximum IO base address 0x0220 # IO base alignment 1 bytes # Number of IO addresses required: 16 # (IO 0 (SIZE 16) (BASE 0x0220)) # IRQ 5. # High true, edge sensitive interrupt (by default) # (INT 0 (IRQ 5 (MODE +E))) # First DMA channel 1. # 8 bit DMA only # Logical device is not a bus master # DMA may execute in count by byte mode # DMA may not execute in count by word mode # DMA channel speed in compatible mode # (DMA 0 (CHANNEL 1
Re: Autologin on Serial Console
On Fri, Jul 14, 2000 at 03:34:08AM +0200, Ron Rademaker wrote On Thu, 13 Jul 2000, Simon Tennant wrote: I currently have a Dec VT320 hooked up to my Debian box via a serial line on /dev/ttyS1. This box sits in the kitchen spewing the output of a tcpdump and provides and interesting way to monitor network activity in the house. Cool ;) I currently have to login and run tcpdump -i eth1. I'd like tcpdump to be run automatically when the console is powered up and for no input to be accepted from the keyboard. To that end I've investigated getty-ps and mgetty. Both want a login or at lease a username in the case of getty-ps. Has anyone successfully managed to get a prelogged in console working? Never tried, but maybe you can do something with a script in /etc/init.d, or instead of running a getty on the terminal a script?? For the stated application, why not just add this to /etc/inittab: td:23:respawn:/usr/sbin/tcpdump -i eth1 /dev/ttyS1 then # telinit Q No login, no security problem aside from people shouldering data off your local ethernet while they make coffee. For this to work, you will have to setup the serial port to the right speed for the terminal; you can do this in your local startup scripts, or make a little script that does it and then runs tcpdump, and use that in place of tcpdump in your inittab entry. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: Getting apache, PHP3, and PostgreSQL working together.
On Fri, Jul 14, 2000 at 09:15:01AM +0200, Remco Rijnders wrote Hello all, I am trying to get apache, PHP3, and PostgreSQL to work together on my Potato machine. I want to use the phpPgAdmin to give me a web accessible interface to the PostgreSQL database. However trying to use it gives me the following error: Fatal error: Call to unsupported or undefined function pg_connect() in lib.inc.php on line 90 This suggests to me that PHP is unawareof the PostgreSQL database functions. I have read the readme files that come with php3-pgsql and searched the web for solutions to this problem. They all seem to suggest to edit /etc/php3/apache/php3.ini . In that file I have the following snippets: extension_dir = /usr/lib/php3/apache ... extension=pgsql.so After having edited this file I issued an apachectl restart but it still refuses to use the database functions I need. What do I do wrong, and, more importantly, how do I fix this? All help is greatly appreciated. You may need to stop Apache and then restart it; you haven't changed Apache's configuration you've changed PHP's, so you have to ensure that the PHP module is unloaded and then re-loaded to give effect to the change. The simplest way I've found of doing this is to stop and restart Apache. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: RADIUS benchmark program
On Fri, Jul 14, 2000 at 01:01:40AM +0200, Russell Coker wrote Currently in my Postal package I have the following: Postal - the mad postman - a SMTP benchmark. Rabid - the mad Biff. POP benchmark that eats your mail as fast as possible. Now I plan to add a RADIUS benchmark to the suite. Does anyone have any ideas what I can name it? How about Caliper? It's the tool I'd use to measure RADIUS. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: What to use as an MTA
On Sun, Jul 09, 2000 at 07:52:19PM +0200, Viktor Rosenfeld wrote Dave Sherohman wrote: 2) Unless you've registered your own domain and it can accept mail, the configuration generated by eximconfig won't quite work out of the box. Specifically, if you tell it you're @isp.net it will assume that all @isp.net addresses are local, preventing you from sending mail to other users who have the same ISP as you. If you go through the config by hand, though, the comments in the generated file make it fairly clear how to fix this. This was exactly my problem, when doing a quick configuration with eximconfig. Mail sent to my fillow students would not be delivered to the university (my ISP), because the system thought, these mails were local. Right now I am using Netscape, so this is not a big issue, but doing mail under Netscape is not the way to go. A quick solution is to change your machine's name; instead of calling it machine.isp.net call it machine.localnet, and *don't* tell eximconfig that you accept mail for people at mail.isp.net (or wherever it got the idea that those are loacl addresses); tell eximconfig to use a smarthost, your ISP's SMTP server. Then add a re-writing rule to the end of your /etc/exim.conf like this: [EMAIL PROTECTED][EMAIL PROTECTED] Fh to automatically rewrite (in this case) the envelope from field and all headers, substituting your true email address for the locally-visible machine name. That gives you a solution which works; the only drawback is that mail that you send to [EMAIL PROTECTED] is delivered via your ISP, rather than directly. The other problem I have with eximconfig is that it will only deliver 10 mails at a time and then pause for a (unknown) delay. This occurs when fetchmail gets mail from my ISP and passes it on to exim. Again, not a big issue, and I haven't even tried to fix it. To quote from exim's info page (under SMTP Processing): Exim normally starts a delivery process for each message received, though this can be varied by means of the -odq' command line option and the `queue_only', `queue_only_file', and `queue_only_load' options. The number of simultaneously running delivery processes started in this way from SMTP input can be limited by the `smtp_accept_queue' and `smtp_accept_queue_per_connection' options. When either limit is reached, subsequently received messages are just put on the input queue. The controls that involve counts of incoming SMTP calls (`smtp_accept_max' `smtp_accept_queue', `smtp_accept_reserve') are not available when Exim is started up from the `inetd' daemon, since each connection is handled by an entirely independent Exim process. Control by load average is, however, available with `inetd'. HTH, John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: su question
On Fri, Jul 07, 2000 at 11:22:38PM +0200, [EMAIL PROTECTED] wrote On Fri, Jul 07, 2000 at 03:14:55PM +0200, tom wrote: howdy guys, This is one of those things that has been pestering me. while logged in as a user in x, how can I edit files that require su privilages? (like /etc/fstab). I usually su from eterm and jed filename. can I su from within emacs? nedit? I use this method [Gianluca][~] = xhost + localhost localhost being added to access control list [Gianluca][~] = su Password: [ROOT][gianluca] # Not so good if there are other users logged into the workstation. I prefer $ ssh -l root localhost if I want to run X applications as root while logged in as an ordinary user. Easy to remember, and it just works. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: What drive is the dir on ?
On Fri, Jul 07, 2000 at 10:16:20PM -0400, Mike Werner wrote C. Falconer wrote: At 09:55 PM 7/7/00 -0400, you wrote: For example, on my box here I get: HAL9000:~$ mount /dev/hda1 on / type ext2 (rw,errors=remount-ro,errors=remount-ro) proc on /proc type proc (rw) devpts on /dev/pts type devpts (rw,gid=5,mode=620) /dev/hdb1 on /space/part1 type ext2 (rw) /dev/hdb2 on /space/part2 type ext2 (rw) /dev/hdb3 on /space/part3 type ext2 (rw) /dev/hdb5 on /space/part4 type ext2 (rw) /dev/hdb6 on /space/part5 type ext2 (rw) /dev/hdb7 on /space/part6 type ext2 (rw) Might I ask why you have seven partitions on hdb ? You sure can. I carved it up that way planning on installing a different distro of Linux on each partition. It's actually only 6 - one of those is the extended partition containing three logical partitions. However, my need for miscellaneous data storage came before I had the chance to do the Great Installfest so those partitions are now being used for random storage. Oh yeah - the drive is a 12.2 gig drive, with each partition being 2 gig except the last partition which was sligtly larger. /dev/hda is a 2.6 gig drive - that's my main drive. I.e. that's where Debian itself is installed, where the swap partition lives (/dev/hda2), and is what I boot from. However, I am now regretting carving the drive up like that as some partitions are rapidly running out of room while others are close to being empty. As soon as I can afford a CD burner I'm going to archive everything off and either turn it back into one large partition or *maybe* 2 or 3 larger partitions. I don't know yet. If you're running a recent 2.2.x kernel and you still have one of those partitions empty, consider using LVM. Together with ext2resize, you can flexibly manage the space without having to repartition stuff. It works quite well. You can get LVM (a kernel patch plus utilities) at ftp://linux.msede.com/lvm/ John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: exim question
On Wed, Jul 05, 2000 at 09:11:37PM +0200, Ron Rademaker wrote How can I use multiple domains with exim, eg both the domains domain.com and otherdomain.com run on a machine and I want this: [EMAIL PROTECTED] to user foo [EMAIL PROTECTED] to user bar [EMAIL PROTECTED] to user hello [EMAIL PROTECTED] to user bar Here's how I do it. The virtual hosts (www.mydomain.com, mail.mydomain.com) all have 'A' records pointing to my mail/web server; CNAME records don't work as well, because stupid mail programs/systems put the CNAME host in the envelope. My exim.conf contains (all single lines, however your mail client displays them): Main section: local_domains = localhost:my.net.au:*.my.net.au:partial3-lsearch;/etc/exim/clients/domains Directors: A new director, to handle the virtual hosts: virtual: driver = aliasfile except_domains = localhost:my.met.au:*.my.net.au domains = partial3-lsearch;/etc/exim/clients/domains no_more file = /etc/exim/clients/$domain_data search_type = lsearch* All other directors have a line like domains = localhost:my.net.au:*.my.net.au so that they aren't applied to the virtual hosts. /etc/exim/clients/domains contains lines like *.mydomain.com.au: mydomain The second field is just an identifier, and for each identifier there is another file, e.g., /etc/exim/clients/mydomain, which works like a regular alias file, e.g.: *: [EMAIL PROTECTED] This setup is for virtual domains with three parts in their domain name (e.g., *.mydomain.com.au); minor adjustment is required if you use American style domains (e.g., *.somewhere.com), and you can mix them easily enough (with a longer local_domains line and a further director). HTH, John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: scp alternative
On Wed, Jul 05, 2000 at 09:09:06PM -0800, Ethan Benson wrote On Wed, Jul 05, 2000 at 09:50:40AM -0400, Chris Mason wrote: I've been using scp to copy the files from my production web server to my hosting companies web server. scp works well except that there is no way to prevent it coping files that already are up to date, and also, it requires a password every time. Is there a better alternative? My remote server does not allow rsync, although they say they will in the future. you can use rsync over ssh, it requires nothing from the server except a working sshd (which it obviously has) rsync --rsh=/usr/bin/ssh Not meaning to be a killjoy, but it also requires that the rsync binary be available on the remote machine. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: Tekram DC310U? Re: Advansys 3940UA Ultra-SCSI controller recommended?
On Thu, Jul 06, 2000 at 02:15:10PM -0400, Peter S Galbraith wrote Hi again, I can get a Tekram DC310U PCI Ultra-SCSI controller (based on the Symbios Logic SYM53C860 SCSI-3 Chip) for the same price as the Advansys 3940UA. It uses the sym53c8xx.o driver. Should I go for that instead? I don't know, but I've used a couple and they worked fine for me. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: Postgresql 7.02 and Debian
On Fri, Jun 30, 2000 at 07:06:58AM -0400, Bill Barnes wrote = Original Message From jpb [EMAIL PROTECTED] = Try apt-get update again. It looks like you had network congestion when you tried before. Or you can just use ftp and go to ftp.us.debian.org and manually grab the postgresql source files (the .diff.gz, .dsc .orig.tar.gz) and use dbuild to extract them. jpb Finally got a clean run. Had to do it in the wee hours. Fetched postgresql okay. Missing files in the compile. The log follows. Snipped out all the good makes. This was a second run. Script started on Fri Jun 30 04:54:05 2000 kgb10:/home/billb/postgresql-7.0.2# ./debian/rules binary snip make -C libpgtcl all make[3]: Entering directory `/home/billb/postgresql-7.0.2/src/interfaces/libpgtcl' gcc -I../../include -I../../backend -O2 -g -g3 -Wall -Wmissing-prototypes -Wmissing-declarations -I../../backend -I../../include -I../../interfaces/libpq -fpic -c -o pgtcl.o pgtcl.c In file included from pgtcl.c:19: libpgtcl.h:19: tcl.h: No such file or directory In file included from pgtcl.c:20: pgtclCmds.h:17: tcl.h: No such file or directory make[3]: *** [pgtcl.o] Error 1 make[3]: Leaving directory `/home/billb/postgresql-7.0.2/src/interfaces/libpgtcl' make[2]: *** [all] Error 2 make[2]: Leaving directory `/home/billb/postgresql-7.0.2/src/interfaces' make[1]: *** [all] Error 2 make[1]: Leaving directory `/home/billb/postgresql-7.0.2/src' make: *** [build-stamp] Error 1 kgb10:/home/billb/postgresql-7.0.2# exit $ zgrep '/tcl\.h' /cognac/linux/debian/dists/potato/Contents-i386.gz usr/include/tcl.h devel/tcl8.0-dev,devel/tcl8.2-dev,devel/tcl8.0-ja-dev usr/include/tcl8.0-ja-int/generic/tcl.h devel/tcl8.0-ja-dev usr/include/tcl8.0/generic/tcl.hdevel/tcl8.0-dev usr/include/tcl8.2/generic/tcl.hdevel/tcl8.2-dev usr/share/doc/selfhtml/html/tcl.htm doc/selfhtml Looks like you need to install tcl8.2-dev (or tcl8.0-dev). John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: can't install kernel/modules from cd-rom
On Tue, Jun 27, 2000 at 12:06:28PM -0400, Curt Salada wrote Thanks for your help, John. Taking your advice, here's what I found: My 2 hard drives appear in the dmesg list as hda and hdb, floppies as fd0 and fd1. The CD-ROM doesn't appear specifically, but I do get the following SCSI-related errors in dmesg: NCR 53c406a: no available ports found Failed initialization of WD-7000 SCSI card! That's OK; the default kernel has drivers for all sorts of stuff, this is just a driver for a device you don't have. (Transcript of complete dmesg at http://americasisp.net/hp/debbie/curt/dmesg.txt) I saw nothing inside the PC that indicated this was an SCSI CD-ROM. Motherboard has 3 cards connected to it: 1. video board (outputs to monitor) 2. interface board(?) has IDE port (connects to HDDs), floppy port, serial port, game port, printer port 3. CD-ROM DRIVE 16 BIT I/F CARD -- goes to CD-ROM (Mitsumi 2x Model CRMC-FX0010, dated May '94) -- card also has stereo RCA output jacks Bingo! You have what's called a proprietary CDROM, which is supported by the installation system using a loadable kernel module. This means that you must make a drivers floppy (much as you presumably made a boot/rescue floppy) and load the module for your CDROM from that floppy before you install the base system, during the Install kernel and modules dialog. The driver you need is probably the 'mcd' (Mitsumi CDROM) or 'mcdx' (Mitsumi CDROM with extended features) driver, and your CDROM device will be /dev/mcd or /dev/mcdx, respectively. I don't recall now what category these drivers appear under when you load modules - it may be CDROM, Misc, or something else. HTH, John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: DPKG ERROR
On Sun, Jun 25, 2000 at 06:32:27PM -0700, Jay Kelly wrote Hello Group, I was trying to install ssh and would get an error E: Sub-Process /usr/bin/dpkg returned an error code (1). So I thought to upgrade DPKG (oh yeah I was running slink and upgraded to potato when this started to happen). Well after trying tp upgrade dpkg I got the same error message. No matter what I try to install with apt-get install I get the same error message. How can I fix this? I have posted some of the error message befare but I will resend some of them. thanks group This is because ssh depends on debconf. The following should get you one step closer: # apt-get install debconf # dpkg --pending --configure Then, you will encounter another problem with this ssh package. Look in /var/lib/dpkg/info/ssh.templates, for a line (about line 90 of 100) that reads Template: ssh/run_sshd There should be a blank line in front of this line, to separate it from the previous Template: section. Insert that new line, and then run # dpkg --pending --configure That should fix it. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: squid errors
On Mon, Jun 26, 2000 at 12:55:43PM +1200, [EMAIL PROTECTED] wrote Hi In my syslog I am getting the following message which seems to appear every minute. Hence i am worried that it may be slowing down my squid server : Jun 26 12:34:53 server squid[13325]: urlParse: URI has whitespace: {http://chann els.real.com/getlatest.glh?PV=6.0.6.99OS=WINL=en-US, en, *LID=1033ch=120+54+ 0+0+0ch=72+39+0+0+0ch=44+287+0+0+0ch=52+505+0+0+0ch=33+393+0+0+0ch=73+ 19+0+ 0+0ch=98+27+0+0+0ch=94+244+0+0+0ch=16+478+0+0+0ch=113+46+0+0+0ch=51+11 6+0+0 +0ch=97+108+0+0+0} if anyone has any clue as to what this is telling me, it would be much appreciated. Squid is complaining that the URL contains whitespace, which isn't allowed (but often happens). If you're worried about the load from logging these messages (at one per minute, I wouldn't be) you can turn them off by using debug_options in /etc/squid.conf; for instance, I have debug_options ALL,1 debug_options url_parse,0 which gives me a little debug infoprmation about some things, but none at all about this kind of thing. Also have a look at the section for the uri_whitespace option, which allows you to specify what action squid should take when such a URI is requested. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: Ctrl-C and normal kill doesn't work.
On Sun, Jun 25, 2000 at 10:22:21PM +0200, Harald Thingelstad wrote This frustrating little problem... When having a running process, ctrl-c or kill doesn't work on my system. You have to use kill -9. A simple example: ping 127.0.0.1 this process is meant to run as long as you want, then you ctrl-c it to get your statistics. However, ctrl-c doesn't work. No reaction whatsoever, the program just keeps on running. So I have to ctlr-z it to reclaim my command line, then kill -9. Normal kill is likewise. No reaction, no error messages, the process just keeps on running. It affects all programs as far as I can see. I'm using debian 2.1 (slink. Long story and a slow connection.) with the latest updates, gnome-icewm and bash. Sure gnome is marked experimental here, but.. Where could the problem be? Kernel? It's a customized debian-source-2.0.38 kernel, but the most exceptional thing about *that* is ip-masquerading support. I'm numbskulled here, so any support appreciated. It may be an stty problem; if you type stty -a at the command prompt you will see the settings for your tty; one of them is intr, which denotes the key used to send a SIGINT to applications. It should be: intr = ^C If yours is not then that would explian it, but then you'ld have to figure out how it got that way. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: ssh deb not installing correctly
On Mon, Jun 26, 2000 at 11:42:03AM +1200, Daniel Free wrote I Just reinstalled my machine at home with debian 2.2 (potato) and started apt-get installing all the things i likeon my system, but when i got to ssh it returned the error below. Begin Console Dump 1 packages not fully installed or removed. Need to get 0B of archives. After unpacking 0B will be used. Setting up ssh (1.2.3-5) ... dpkg: error processing ssh (--configure): subprocess post-installation script returned error exit status 10 Errors were encountered while processing: ssh E: Sub-process /usr/bin/dpkg returned an error code (1) End Console Dump the program itself has installed correctly and works however whenever i use apt-get it finishes with the above error. any suggestions ?? There is a (?) bug in this version of ssh. Look in /var/lib/dpkg/info/ssh.templates, for a line (about line 90 of 100) that reads Template: ssh/run_sshd There should be a blank line in front of this line, to separate it from the previous Template: section. Insert that new line, and then run # dpkg --pending --configure That should fix it. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: Bash script question (was: Re: Netscape 4.73 wrapper broken)
On Fri, Jun 23, 2000 at 08:51:44PM -0400, Peter Kovacs wrote On Sat, 24 Jun 2000, Mark Phillips wrote: What does YMMV stand for? YMMV = Your Mileage May Vary An official fix has come through, and what it does is replace for f in (cd. by for f in $(cd From man 1 bash: Command Substitution Command substitution allows the output of a command to replace the command name. There are two forms: $(command) or `command` That's why I had the backticks in there. I've never seen the $() construct either. And what is the . $d/$f supposed to do?? Again, from man 1 bash: . filename [arguments] source filename [arguments] Read and execute commands from filename in the curĀ rent shell environment and return the exit status of the last command executed from filename... It simply executes those files as a shell script. To be more accurate, it executes the contents of those files as if they had been included in the current script. No new shell, all the current parameters accessible, any parameters or functions defined in those files available in the current context. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: kernel problem
On Wed, Jun 21, 2000 at 11:04:52AM +0300, Petteri Heinonen wrote I compiled kernel succesfully, but when I try to boot with it, it says: Kernel panic: No init found. Try passing init= option to kernel. Have I left something important out of my kernel or what is going on? Probably, you've set the wrong boot partition. Presumably, it can mount the partition that it *thinks* is root, otherwise you'ld have a Panic: Can't mount root fs. If you're using LILO, you can pass the root partition at the boot: prompt; e.g., boot: linux root=/dev/hda3 John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: SGML beginners question
On Thu, Jun 22, 2000 at 12:03:44PM +1000, Shao Zhang wrote Hi, Recently, I have read a lot about SGML. It gives me the impression that it is very hard to learn and very very powerfull. However, I still don't have a clue that in what circumstances I should use SGML instead of others. Should I write a thesis report in SGML or LaTeX? Would it be idea for a general document that one would normally write in Word? Thanks for any help in advance. I use SGML for two reasons: - For documents where I want to maintain a single source but want to produce a range of output formats (e.g. .html, .ps, .rtf, .dvi) - For the sheer joy of using psgml so I don't have to memorise a document grammar. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: no lpr
On Wed, Jun 21, 2000 at 08:04:50PM -0700, cls--colo spgs wrote debs, i finally upgraded my 2.0.36 kernel by compiling 2.2.16 on my potato box. in the process, i lost lpr. ...error message: lpr: connect: connection refused jobs queued, but cannot startd daemon did my compiling miss the mark? (how may print again?--i re-ran /etc/sbin/magicfilterconfig, but no success.) Device naming for parallel ports changed in kernel 2.2.x; under 2.0.x the first port is lp1, under 2.2.x it is lp0. Also, the parallel port driver was broken into several chunks; if you have an Intel-based machine you will need the parport parport_pc (Parallel Port Support and PC-style Hardware under General Setup) modules, as well as the lp module with maybe printer_readback selected (Parallel printer support and Support IEEE1284 status readback under Character Devices). HTH, John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: Re-send contents of a mail file
On Tue, Jun 20, 2000 at 04:07:18PM +0200, Marcus Johansson wrote Hi! I have a pretty huge mail file for a user who want that transfered to his Lotus Notes mail box. Is there any way of re-sending the contents of this mail file to his new mail address? All new mails are of course forwarded to this new account, so that's no problem... Is this something sendmail can do, or is there a tool in Debian for this? Or can I make a script or something... You may want to look at formail; something like $ formail -s sendmail [EMAIL PROTECTED] oldmailbox I haven't used this myself, but the man page makes it look like it would work. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: can't install kernel/modules from cd-rom
On Tue, Jun 20, 2000 at 04:04:37PM -0400, Curt Salada wrote Installing 2.1 on a 486, eliminating Win3.11. Going through the steps in the Installation Main Menu, I get as far as Install Operating System Kernel and Modules. Program asks me to select CD interface type. Each option (/dev/hda, /dev/hdb, etc.), returns the same message: Mount failed. The CD-ROM was not mounted successfully. Selecting /dev/hdb takes a few seconds longer, as if it might work, but still returns the Mount failed message. Now what? Hardware: 2 IDE hard drives /dev/hda -- Maxtor 1.5G (1 bootable Linux partition, 1 swap) /dev/hdb -- Samsung 406M (1 Linux partition) CD-ROM Mitsumi IDE ^^ You need to identify the device that your CDROM is using. It isn't /dev/hda or /dev/hdb, as they are your hard disks. If it's on a cable of its own that connects to the same card (or the same area of the motherboard) as your hard disks, it's probably /dev/hdc or perhaps /dev/hdd; if it has its very own expansion card, or it plugs into a sound card, things get more complicated. If your CDROM is connected to a standard IDE port then you can use the Linux boot floppy/CDROM to identify it. Boot up using the floppy/CDROM and then, after you see the initial screen asking if you have a colour monitor, press ALT+F2 and then, like the screen says, press Return. You should now see a pound sign (#), which is the root prompt. Type 'dmesg' (no quotes) and hit Return; you should see a couple of screens of text scrolling past. These are the messages produced by the Linux kernel and drivers during startup, which will include details of all the IDE devices that the kernel found. You can scroll up and down through this list by holding down the left Shift key while tapping PageUp and PageDown, respectively. Look for lines a bit like this: hda: FUJITSU MPC3043AT, ATA DISK drive hdc: FX320S, ATAPI CDROM drive hdd: LS-120 COSM 02 UHD Floppy, ATAPI FLOPPY drive In this case, hdc is the CDROM drive and you would specify /dev/hdc as the device name. After you finish with this you can type exit at the prompt and then press Alt+F1 to return to the installation program. If it doesn't appear in the output of dmesg then you have more work to do. If it's on a cable of its own that connects to the same card (or the same area of the motherboard) as your hard disks, it's probably /dev/hdc or perhaps /dev/hdd, and you should check your BIOS to make sure that you haven't (e.g.) disabled the second IDE controller, and then make sure the cables are inserted correctly and the right way round. If it has its very own expansion card, or it plugs into a sound card things get more complicated, as you may have to configure the card, use special boot parameters or load a module before you can access it, depending on what kind of card it's plugged into; perhaps you should check it out, and then post again. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: setting up ssh so that ssh'ing to a system doesn't require password
On Tue, Jun 20, 2000 at 04:16:42PM -0800, Britton wrote I need to figure out how to set up ssh in such a way that ssh can be used without the user needing to enter a password. I apparently don't understand the required format of the .shosts command to achieve this. I have also used ssh-keygen, placed my public key in the auhtorized_keys file, started an agent (ssh-agent $SHELL) and used ssh-add to add my identity to the authentication agent, but doing ssh localhost still requires me to enter a password. If anyone has a valid .shosts file I would greatly appreciate a snipping showing how they have it set up, or a pointer to an example file (which I havn't been able to find in the docs). Make sure that sshd is configured to allow RSAA authentication on the target machine: look for a line like RSAAuthentication yes in /etc/ssh/sshd_config. If that's set then all you need is to generate a key pair using ssh-keygen on the host you're connecting from, and place that key into ~/.ssh/authorized_keys on the host you're connecting to. No .shosts or ssh-agent required. If you want your users to login silently, don't specify a passphrase when running ssh-keygen. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: email program for newbie
On Tue, Jun 20, 2000 at 05:20:10PM -0700, marshall paul wrote Hi, I have been plodding through the Debian woods, taking one thing at a time and eventually getting small successes. However, I am tired and would like to set up email for a dialup single user machine. Would someone recommend a mail program to start out with? (There is just a bewildering amount of variety when one is new to Linux and it has taken _weeks_ to configure X. OK, I can surf, now, but I want email, too.) If you just want the same kind of direct access to your ISP POP account as you'd get with Windows, you can use either Netscape Messenger (part of the various Communicator releases), or XFMail. There are many other (arguably better) mailers out there (my favourite is Mutt), but these two are most like what you'ld be used to, with a minimum of extra bells and whistles to obscure the issue. In the longer term, something like exim + fetchmail + mutt + some configuration would give you greater flexibility and better integration of local mail with internet mail. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: Loading X w/only netscape
On Mon, Jun 19, 2000 at 07:16:27PM -0700, [EMAIL PROTECTED] wrote I would like to setup an account on my system.. but when a user logs in (via KDM) if possible i don't want it to load a windowmanager and all i want it to load is netscape, also i want it to log back out when netscape exits.. is this possible? right now i have a user that logs in using afterstep (i can't seem to get .xinitrc to override kdm's settings) and modified the autoexec to load netscape on the special page it should be on but i can't get it to logout afterwards i just put big text in the background that says hit CTRL+ALT+BACKSPACE to logout. i also chsh'd the user to /dev/null so they can't get a shell(not that security is really an issue right now). this is basically so someone can enter a generic login and have the machine show this website when im not around to do it for them. This is basically a bit of a security hole, unless you can and do nail down Netscape's configuration pretty tight. Users who want to escape Netscape can probably (e.g.) fiddle with mime types/helper applications to give themselves an xterm or window manager, so it's probably not a good idea unless you trust your users. If you want to stop people mucking about in that way, you may want to consider using Mosaic (after ensuring that it renders your pages satisfactorily), as it has a 'kiosk' mode that is intended for this kind of application. As for *how*, if you can get KDM to use ~/.xsession it should just be a question of using an xsession like #!/bin/sh /path/to/netscape http://my.host/url John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: librep problem
On Wed, Jun 14, 2000 at 04:36:58PM -0400, cam_random wrote I'm currently trying to install Sawfish, however in order to do so I need librep. When I've tried to configure librep i keep getting the same error message: checking for mpz_init in -lgmp... no configure: error: cannot find GMP Library I've installed the gmp library...several times. any ideas? Most libraries, libgmp2 included, come in two packages: the runtime package, required for compiled software that links against the library, and a development package, required for building software that links against the library. From your mail, it appears that you have installed the libgmp2 package; have you also installed libgmp2-dev? John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: Apache and perl cgi problem
On Thu, Jun 15, 2000 at 09:27:16AM +0200, [EMAIL PROTECTED] wrote Hello I have a little problem with Apache and the execution of the cgi-script. I have already set the ScritpAlias and it work fine. Now I'd like to have under the public_html dir of one user (my user) a directory tree wicht contain one entry for every project I am working on. The problem is that I cannot execupe some perl cgi, Netscape give me this error --- You don't have permission to access /~gianluca/effeesse/p.pl on this server. --- The script have the correct attributes and the directory is accessible (I can get the html file and also run the php3 script) I already added the line AddHandler cgi-script .pl in the src.conf file, but it not work as well. Any suggestion ? Make sure you have Options ExecCGI set for the directory (probably in access.conf) if you want Apache to run your script. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: exim, sendmail, smail, qmail .... ???
On Fri, Jun 16, 2000 at 09:08:07AM +1000, Richard Lindner wrote Would someone explain Debian's logic behind choosing exim over sendmail or smail for slink and potato (and I presume woody). The exim web page is excruciatingly coy regarding comparisons between it an the other mainstream MTAs. What's exims advantage (other than a simplified config) that warrants it's use over the other industry standards? I can only speak for myself, but I like it best. - It's better-maintained and more flexible than smail; - It's easier to build, modify and debug configurations than with sendmail; - It doesn't have weird licensing issues like qmail; - Exim's bundled documentation is far better than the others. Additionally, at one stage Eric Allman definitely had his nose out of joint over Linux; while that may be long past, people take these things personally and that may have contributed. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: Exim changes
On Tue, Jun 13, 2000 at 02:24:58PM +0930, Mark Phillips wrote John Pearson [EMAIL PROTECTED] wrote: It could be that mail.infoeng.flinders.edu.au and adam.ist.flinders.edu.au are the same machine, and it's simply using the canonical name in your logs. This would be fairly normal if they are arranging a transition from one to the other: both would work until they decide to drop the old name. Try the following three things: $ host adam.ist.flinders.edu.au adam.ist.flinders.edu.auA 129.96.1.21 and $ host mail.infoeng.flinders.edu.au mail.infoeng.flinders.edu.auCNAME adam.ist.flinders.edu.au adam.ist.flinders.edu.auA 129.96.1.21 to see if they are, in fact, the same machine, and It seems they are. $ /usr/sbin/exim -bt [EMAIL PROTECTED] to confirm how exim thinks it should be routed. [EMAIL PROTECTED] deliver to [EMAIL PROTECTED] router = smarthost, transport = remote_smtp host adam.ist.flinders.edu.au [129.96.1.21] So does this mean that it is routing to adam directly, or via mail.infoeng? How can you tell? No via about it. They are the same machine; the only difference is in the name that you use to refer to it. It's likely that exim is simply using the canonical name in the logs. The other thing I still don't understand, is how mail can be delivered even though the exim daemon is not running (ie /etc/init.d/exim stop has been run). Because I tried this, and mail still seems to be delivered outside. Probably, your mail program is sending mail directly by passing messages to /usr/sbin/sendmail on STDIN. This should be a symlink to exim (or whatever MTA you're using). The exim daemon is responsible for accepting incoming SMTP connections, and for running the mail queue. If your message is sent by invoking /usr/sbin/sendmail then it may be delivered directly and never end up in the queue, and so get where it's going whether the exim daemon is runing or not; it depends on how exim is configured. Could be, everything's working OK :-) Bye for now, John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: Exim changes
On Tue, Jun 13, 2000 at 09:23:47AM +0930, Mark Phillips wrote My ISP is soon to make changes to its email system, and as a result I need to change my exim setup. Basically, whereas the host I used to send out email used to be adam.ist.flinders.edu.au, it now is mail.infoeng.flinders.edu.au. A quick grep adam exim.conf gave me: route_list = * adam.ist.flinders.edu.au bydns_a so I changed this to: route_list = * mail.infoeng.flinders.edu.au bydns_a and then did /etc/init.d/exim restart. Surely this is all I need to do I thought. But when I sent off a test message and looked at /var/log/exim/mainlog it had: 2000-06-13 08:56:03 131dal-0001s8-00 = [EMAIL PROTECTED] U=mark P=local S=1016 [EMAIL PROTECTED] 2000-06-13 08:56:06 131dal-0001s8-00 = [EMAIL PROTECTED] R=smarthost T=r emote_smtp H=adam.ist.flinders.edu.au [129.96.1.21] 2000-06-13 08:56:06 131dal-0001s8-00 Completed So it is still using adam.ist.flinders.edu.au!! What is wrong? I thought perhaps this invocation of exim is via the inetd??? But if I look at inetd.conf I see: It could be that mail.infoeng.flinders.edu.au and adam.ist.flinders.edu.au are the same machine, and it's simply using the canonical name in your logs. This would be fairly normal if they are arranging a transition from one to the other: both would work until they decide to drop the old name. Try the following three things: $ host adam.ist.flinders.edu.au and $ host mail.infoeng.flinders.edu.au to see if they are, in fact, the same machine, and $ /usr/sbin/exim -bt [EMAIL PROTECTED] to confirm how exim thinks it should be routed. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: purging X from debian installation
On Fri, Jun 09, 2000 at 11:25:19PM -0500, w trillich wrote i'll wind up reformatting and reinstalling, i can tell. i'm trying to get xwindows stuff off the hard drive, and use only ncurses console/telnet/ssh interaction, and server software -- but when i try to zap the xlib6 packages (xlib6 and xlib6g) it wants to remove elvis, perlmagick/ libmagick and a few other non-X-dependent items: [snip] X is a thing of many parts (one reason why it's spread over so many packages), and while you may want to be rid of it, you may have to learn to live with parts of it if you want to keep these other packages. I wouldn't worry too much about xlib6g itself; programs that are linked against it will require it, but it only provides the ability to talk to Xservers and so on; it doesn't represent any particular weakness unless you're actually running them under X (or they think you are). If it's purely the space you're concerned about you could consider using nvi in place of elvis, but then you miss out on some of the fancy stuff (like X support) that you probably like about elvis. Of course, this won't help you much with perlmagick. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: gv
On Wed, Jun 07, 2000 at 08:32:14AM +0700, Oki DZ wrote Hi, When I run gv, I have the following: Error: /invalidfontGNU Ghostscript: Unrecoverable error, exit code 1 in findfont Operand stack: 1 --dict:4/4-- --dict:4/4-- Times-Roman Font Times-Roman 28765 Times-Roman --nostringval-- Times-Roman Courier Execution stack: %interp_exit .runexec2 --nostringval-- --nostringval-- --nostringval-- 2 %stopped_push --nostringval-- --nostringval-- --nostringval-- false 1 %stopped_push 1 3 %oparray_pop .runexec2 --nostringval-- --nostringval-- --nostringval-- 2 %stopped_push --nostringval-- --nostringval-- -- I believe that Times-Roman is included in my installed font packages: [EMAIL PROTECTED]:~$ dpkg -l | grep xfonts ii xfonts-100dpi 3.3.5-2100 dpi fonts for X ii xfonts-75dpi 3.3.5-275 dpi fonts for X ii xfonts-base3.3.5-2standard fonts for X ii xfonts-scalabl 3.3.5-2scalable fonts for X Is there anything that I should install? Ghostscript uses its own fonts when interpreting postscript, provided in gsfonts. Try installing that. Xfonts may be used by gv to draw your document, but that comes later. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: problems with server dying.
On Tue, Jun 06, 2000 at 10:41:09AM +1000, Marc-Adrian Napoli wrote Hi all, as you can guess from the subject i have a server (debian 2.0, pentium 200, 500mb ram, 30 gig or so) that is dying on me at random times early in the morning!! (quite annoying). I've gathered the following from the logs: Jun 5 06:43:42 godzilla kernel: EXT2-fs error (device 16:00): ext2_readdir: bad entry in directory #126 28: rec_len % 4 != 0 - offset=0, inode=3326860705, rec_len=34410, name_len=31708 Jun 5 06:43:42 godzilla kernel: Remounting filesystem read-only Jun 5 06:43:42 godzilla kernel: EXT2-fs error (device 16:00): ext2_find_entry: bad entry in directory # 12628: rec_len % 4 != 0 - offset=0, inode=3326860705, rec_len=34410, name_len=31708 Jun 5 06:43:42 godzilla kernel: Remounting filesystem read-only Jun 5 07:02:50 godzilla kernel: hdc: read_intr: status=0x59 { DriveReady SeekComplete DataRequest Error } Jun 5 07:02:50 godzilla kernel: hdc: read_intr: error=0x40 UncorrectableError }, LBAsect=4512574, sec tor=4512574 Jun 5 07:02:50 godzilla kernel: end_request: I/O error, dev 16:00, sector 4512574 Jun 5 07:03:00 godzilla kernel: hdc: irq timeout: status=0xd0 { Busy } Jun 5 07:03:01 godzilla kernel: ide1: reset: success Jun 5 07:03:08 godzilla kernel: hdc: read_intr: status=0x59 { DriveReady SeekComplete DataRequest Error } Jun 5 07:03:08 godzilla kernel: hdc: read_intr: error=0x40 UncorrectableError }, LBAsect=4512720, sec tor=4512720 Jun 5 07:03:08 godzilla kernel: end_request: I/O error, dev 16:00, sector 4512720 Jun 5 07:03:19 godzilla kernel: hdc: irq timeout: status=0xd0 { Busy } Jun 5 07:03:21 godzilla kernel: ide1: reset: success Jun 5 07:03:31 godzilla kernel: hdc: irq timeout: status=0xd0 { Busy } Jun 5 07:03:35 godzilla kernel: ide1: reset: success Jun 5 07:03:43 godzilla kernel: hdc: read_intr: status=0x59 { DriveReady SeekComplete DataRequest Error } Jun 5 07:03:43 godzilla kernel: hdc: read_intr: error=0x01 AddrMarkNotFound }, LBAsect=4512574, secto r=4512574 When the techie on call at that time put a monitor on the box he saw Couldn't get free page... all the way down the screen and couldn't get a prompt. (Forcing us to hard reboot the system). Looks like /dev/hdc is in trouble. While this could be caused by other stuff on the same IDE cable or on your PCI bus, if all of the messages are pointing at the same device (they seem to be) it's the most likely source. Most likely causes: - /dev/hdc is dying. - Overheating. Especially if you have several drives, with maybe not as much air space between them as they might prefer, and have only a CPU and PSU fan. Especially if there's an extended period of disk activity (say, some 7am cron jobs) before things come unstuck. - Bad or poorly fitted IDE or power cable. I'd have a close look at the cooling for /dev/hdc and maybe give it more room or install a drive fan, but I'd probably also source a replacement drive in case. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: gv
On Thu, Jun 08, 2000 at 07:14:10AM +0700, Oki DZ wrote On Wed, 7 Jun 2000, John Pearson wrote: Ghostscript uses its own fonts when interpreting postscript, provided in gsfonts. Try installing that. Xfonts may be used by gv to draw your document, but that comes later. Thanks, it's working now. One might wonder why gsfonts pkg is not included in gv's dependency list? To someone in your position it may seem thin, but it's probably because you only need gsfonts if you want to view PS files that use fonts (or, you may have your own PS fonts squirreled away somewhere). BTW, what kind of tool do you use to convert .ps to .pdf? gv can do .pdf to .ps, acroread does the same thing too. I use /usr/bin/ps2pdf, which is included in gs and gs-aladdin. It's just a wrapper around gs (which supports PDF as one of its output formats), but it saves remebering all those options... John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: LAN/NAS problem.
On Thu, Jun 08, 2000 at 12:10:01PM +1000, Marc-Adrian Napoli wrote hi all, Hello, Marc! we have an ascend server that serves a dial up customer sitting on 203.28.51.128. routing table on the ascend server: 203.28.51.128/27 203.28.51.128 wan12rGT120 265755169 203.28.51.128/32 203.28.51.128 wan12rT 60 141505169 they are given the 203.28.51.128/27 subnet from radius. on the customers end is a linux box acting as a router for their /27 subnet. ifconfig on the debian box shows: loLink encap:Local Loopback inet addr:127.0.0.1 Bcast:127.255.255.255 Mask:255.0.0.0 UP BROADCAST LOOPBACK RUNNING MTU:3584 Metric:1 RX packets:7607 errors:0 dropped:0 overruns:0 TX packets:7607 errors:0 dropped:0 overruns:0 eth0 Link encap:Ethernet HWaddr 00:00:E8:47:F3:9C inet addr:203.28.51.128 Bcast:203.28.51.159 Mask:255.255.255.224 ^ UP BROADCAST RUNNING PROMISC MULTICAST MTU:1500 Metric:1 RX packets:85592 errors:0 dropped:23 overruns:0 TX packets:6940 errors:0 dropped:0 overruns:0 Interrupt:9 Base address:0x240 ppp0 Link encap:Point-to-Point Protocol inet addr:203.28.51.128 P-t-P:203.63.219.10 Mask:255.255.255.224 ^ UP POINTOPOINT RUNNING MTU:1524 Metric:1 RX packets:10757 errors:0 dropped:0 overruns:0 TX packets:10410 errors:0 dropped:0 overruns:0 [snip] Using the network address ([203.28.51.128]) as the interface address isn't a good idea. It may be treated as special for all sorts of reasons and you can't rely on it to behave like a regular host IP, or even for different TCP/IP stacks to deal with it consistently. For example on my local LAN, with a network address of 192.168.113.0, all of my Debian boxes (running kernel 2.2.15) respond if I ping 192.168.113.0, but it would be equally understandable if none of them did (as is the case if I run # echo 1 /proc/sys/net/ipv4/icmp_echo_ignore_broadcasts ). Change the interface address to (say) 203.28.51.129; it may not fix your problem, but it removes the most obvious reason why it may not work. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: PHP with MySQL
On Thu, Jun 01, 2000 at 03:17:53AM +, Thiago wrote I'm tring ti use PHP with MySQL, but every time I receive this message! Fatal error: Call to unsupported or undefined function mysql_connect() in /www/http/mysql_test.php3 on line 12 How do I recompile the PHP to enable MySQL support? Thaks! You shouldn't need to. If you have php3 (not php3-cgi) installed: - Install php3-mysql; - Ensure that the line extension=mysql.so appears in /etc/php3/apache/php3.ini - Stop re-start apache. If you're using php3-cgi: - Install php3-cgi-mysql; - Ensure that the line extension=mysql.so appears in /etc/php3/cgi/php3.ini HTH, John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: wvdial dials; kppp doesn't. Why???
On Mon, May 29, 2000 at 08:45:19AM -0400, Chris Joyner wrote Thanks, Jens, I have tried that, but I'm pretty sure my ISP uses PAP. Here is the ouptput to xconsole from KDE. (I finally figured out how to get it.) Any clues here? May 29 08:28:39 probe pppd[270]: sent [LCP ConfReq id=0x1 asyncmap 0x0 auth chap 05 magic 0xbf09398b pcomp accomp] May 29 08:28:39 probe pppd[270]: rcvd [LCP ConfAck id=0x1 asyncmap 0x0 auth chap 05 magic 0xbf09398b pcomp accomp] May 29 08:28:39 probe pppd[270]: rcvd [LCP ConfReq id=0x2 mru 1514 asyncmap 0x0 auth pap magic 0x3ad32ce3 pcomp accomp 11 04 05 ea 13 03 00] May 29 08:28:39 probe pppd[270]: sent [LCP ConfRej id=0x2 11 04 05 ea 13 03 00] May 29 08:28:39 probe pppd[270]: rcvd [LCP ConfReq id=0x3 mru 1514 asyncmap 0x0 auth pap This says your ISP is asking your PC for PAP authentication. magic 0x3ad32ce3 pcomp accomp] May 29 08:28:39 probe pppd[270]: sent [LCP ConfAck id=0x3 mru 1514 asyncmap 0x0 auth pap magic 0x3ad32ce3 pcomp accomp] May 29 08:28:39 probe pppd[270]: sent [LCP EchoReq id=0x0 magic=0xbf09398b] May 29 08:28:39 probe pppd[270]: sent [CHAP Challenge id=0x1 8cb2c35edaf96a8d387acdf21d83b356e528da429e081e6e3cd18fa5cd7f8e19a95a6780, name = probe] May 29 08:28:39 probe pppd[270]: sent [PAP AuthReq id=0x1 user=jhzb password=**] May 29 08:28:39 probe pppd[270]: rcvd [LCP EchoRep id=0x0 magic=0x3ad32ce3] May 29 08:28:39 probe pppd[270]: rcvd [CHAP Response id=0x1 d2b0109626c64619d32d61a12f5e9c3f, name = HiPer] May 29 08:28:39 probe pppd[270]: No CHAP secret found for authenticating HiPer This is your PC talking. HiPer is offering it an authentication token, which your PC doesn't recognise... May 29 08:28:39 probe pppd[270]: sent [CHAP Failure id=0x1 I don't like you. Go 'way.] May 29 08:28:39 probe pppd[270]: sent [LCP TermReq id=0x2 Authentication failed] So it tells your ISP to go away. May 29 08:28:39 probe pppd[270]: rcvd [PAP AuthAck id=0x1 ] May 29 08:28:40 probe pppd[270]: Hangup (SIGHUP) May 29 08:28:40 probe pppd[270]: Modem hangup May 29 08:28:40 probe pppd[270]: Connection terminated. May 29 08:28:41 probe pppd[270]: Exit. Assuming you don't require your provider to give a password, try adding noauth to /etc/ppp/options. That tells PPP that the remote peer (your ISP) doesn't need to authenticate itself. If you *do* have a password that your ISP has to provide before you will connect, add it to /etc/ppp/chap-secrets or /etc/ppp/pap-secrets, as appropriate. Jens Luedicke wrote: try in kppp the CHAP or an script-based authentication method. ---Reply to mail from Chris Joyner about wvdial dials; kppp doesn't. Why??? Dear friends, I have tinkered and tinkered but still can't make kppp work from KDE to dial my modem. I can use wvdial from a console (but only as root! BTW, how do I get to use wvdial as another user?) and it connects immediately. Any suggestionsas to getting kppp to do its thing? I'm using slink. Thanks, Chris Joyner -- Unsubscribe? mail -s unsubscribe [EMAIL PROTECTED] /dev/null ---End reply with friendly regards Jens Luedicke [EMAIL PROTECTED] -- Unsubscribe? mail -s unsubscribe [EMAIL PROTECTED] /dev/null HTH, John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services
Re: Tulip ethernet performance issues
On Sun, May 28, 2000 at 12:53:32PM +1000, Damon Muller wrote Hi gang, I know that this has come up on the list recently, but I haven't really seen anything that has helped me solve this little problem. I have a couple of tulip-based ethernet cards (I think they are made by Acton, and may certainly be re-badged), one in my Debian box and one in my win98 box. They are connected together by an 8-port 10/100 switch (specifically a LanTech MINI Switch 800 (8 port 10/100 Base-TX Switch) according to the front panel). The performance that I'm getting through this network is significantly less than I'd be expecting. In fact, it seems to be slower than the 10M hub that I had previously. I've just transfered a fairly large file from my Linux machine to my Win98 machine using the Win98 FTP client (the console based one), and here is what it said when it finished: 370238462 bytes recieved in 1872.52 secs 197.72 Kbytes/sec. With only 2 pcs, both with 10/100 tulip cards, over a 100M switch, I would have expected a much better transfer rate. (there is nothing else connected to the switch). Here are some diagnostics: Linux rei 2.2.15 #1 Fri May 5 18:30:12 EST 2000 i586 unknown tulip-diag.c:v1.19 10/2/99 Donald Becker ([EMAIL PROTECTED]) Index #1: Found a Digital DS21143 Tulip adapter at 0x6c00. Port selection is 100mbps-SYM/PCS 100baseTx scrambler, full-duplex. Transmit started, Receive started, full-duplex. The Rx process state is 'Waiting for packets'. The Tx process state is 'Idle'. The transmit unit is set to store-and-forward. The NWay status register is 41e1d2cd. Internal autonegotiation state is 'Negotiation complete'. Use '-a' or '-aa' to show device registers, '-e' to show EEPROM contents, -ee for parsed contents, or '-m' or '-mm' to show MII management registers. CPU0 0: 460597 XT-PIC timer 1: 10326 XT-PIC keyboard 2: 0 XT-PIC cascade 4: 5129 XT-PIC serial 5: 6112 XT-PIC soundblaster 8: 1 XT-PIC rtc 10: 286851 XT-PIC eth0 12: 8685 XT-PIC aic7xxx 13: 1 XT-PIC fpu 14: 151229 XT-PIC ide0 15: 738283 XT-PIC ide1 NMI: 0 eth0 Link encap:Ethernet HWaddr 00:40:C7:9A:01:5F inet addr:192.168.13.1 Bcast:192.168.13.255 Mask:255.255.255.0 UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 RX packets:181175 errors:80 dropped:0 overruns:0 frame:81 TX packets:259454 errors:12590 dropped:0 overruns:4 carrier:12586 collisions:820 txqueuelen:100 Interrupt:10 Base address:0x6c00 Particularly worrying, I guess, is the error, collisions, etc on the ethernet card (I had rebooted just before this large transfer). On my machine at work, with has a 3Com 509 which is also attached to a switch, with 3.5million packets transfered, there has not been a single error or collision (There should never be a collision with a switch, should there? I though that was the idea!). Can anyone suggest any possible solutions? Is it likely the card is dodgy (it worked fine, and fast with a 100M hub that I had, but I couldn't attach my laptop to that, as it only had a 10M pcmcia card), or might it be an interaction between the card and the hub? Is it work shelling out for a new card? You should take a long, hard look at your cables. At 100M cables that work fine at 10M can turn your data to mush. Verify that they are genuine CAT5, and if you have the opportunity verify that they give satisfactory performance at 100M with different gear; even if the cable is CAT5, poor terminations or connectors can kill them. John P. -- [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.mdt.net.au/~john Debian Linux admin support:technical services