Re: devices compatible list
On Tue, Feb 22, 2000 at 11:58:50PM -0500, Allan M. Wind wrote: On 2000-02-22 20:24:10, Bret Rice wrote: Would appreciate a direct to a list of devices which are conmpatible with Debian. My particular interest at the moment is 3Com 509 vs 905 cards. Got two 905 in my box and haven't noticed any problem. In the past, I think the Intel EtherExpress Pro were favored though. Might be worth exploring. Had heard that the kernel 3c59x driver freezes the computer under some situation when used against 3c905C. Have not seen anything strange personally, but doesn't hurt to use the 3COM driver.. it's GPL'ed also. http://support.3com.com/infodeli/tools/nic/linux.htm
OpenBSD SSH in potato.
I just noticed a strange thing In the default /etc/ssh/sshd_config there is a line ServerKeyBits 768, however, the post-installation script creates a key with 1024 bits. I thought the ServerKeyBits option should correspond to the host key as generated by the script? Is it a bug, or did I misunderstood something?
Problem with DHCP
I am running Potato on an P3.. and have a DHCP server with some Windows 95 clients. However, when those Windows 95 boot up they cannot get an IP from the DHCP server. (according to the logs the DHCP server have already offered the IPs.) When I run winipcfg and manually refresh the DHCP settings it works most of the times. Sometimes the interface refuses to update the IP. There was once a Windows NT DHCP server running with no problems (and was replaced by the Debian server). Is it possible that I have misconfigured something? Clients uses Intel EEpro* cards, and the server uses a D-Link card with the chip from Davicom (using via-rhine modules)
Re: Problem with DHCP
On Thu, Jan 27, 2000 at 10:09:20AM -0600, Phil Brutsche wrote: A long time ago, in a galaxy far, far way, someone said... I am running Potato on an P3.. and have a DHCP server with some Windows 95 clients. However, when those Windows 95 boot up they cannot get an IP from the DHCP server. (according to the logs the DHCP server have already offered the IPs.) When I run winipcfg and manually refresh the DHCP settings it works most of the times. Sometimes the interface refuses to update the IP. There was once a Windows NT DHCP server running with no problems (and was replaced by the Debian server). Is it possible that I have misconfigured something? It's likely that you've misconfigured something, however we need to know more about how your network is set up: the locations of routers hubs so on. I am not responsible for the network set-up so I cannot tell for sure, but I think everything is connected to a big Ethernet switch.. The network is connected to the external network through a machine (IP: 10.1x.x.1) that acts also as a WWW proxy server. First DNS server, as well as the DHCP server, is on 10.1x.x.2.. I can't remember the exact IPs, but I guess they shouldn't matter..? Are you using ISC's DHCP server software? If you are, it might help to see a configuration that's been working for me for quite some time: http://tux.creighton.edu/~pbrutsch/dhcpd.conf. Yes, I am using ISC's DHCP server. My configuration file is similar... equally short. The only differences that I can remember (save from choices of IPs) are: 1) I put the routers and domain-name-servers lines outside of the subnet bracket 2) I have set up more than 1 DNS server, separated by commas 3) lease times are shorter.. like a few hours. Am I supposed to use lease time that long? Clients uses Intel EEpro* cards, and the server uses a D-Link card with the chip from Davicom (using via-rhine modules) The ethernet cards shouldn't make a difference.
Re: ipchains
On Mon, Jan 17, 2000 at 02:02:30PM +0100, Ron Rademaker wrote: On my network there is one linux server and some windows things, the linux server is used as a gateway (and some other things), but it doesn't work properly. When internet exploder is used, all goes fine (most of the time), but when for example someone tries to ping from within the dos box it doesn't. I think it has something to do with ipchains. Both the input as the output chain accept anything and the forward chain: ipchains -A forward -j MASQ -s 192.168.0.1 -d 0.0.0.0/0 Shouldn't it be 192.168.0.0/24 (or something similar) instead? However I'd try to forget about it and apt-get install ipmasq :) I'd also add the -i argument. (But I find that ipmasq doesn't handle dhcp requests correctly. I have to enable those udp stuff before dhcp will work) (for every IP on the network). What can I do to fix this?? Ron
Re: Standard way to change IP?
On Sat, Jan 08, 2000 at 09:17:05PM +0100, Robert Waldner wrote: I find that sometimes I cannot use that interface after the network is changed I get something like network unreachable when I try to ping some hosts on that network. The NIC is fine after a reboot. Most of the time I was using slink with kernel 2.2.13. Are there any other things that I need to do? Or was it driver dependent? sounds like a missing route, route -n will show you the actual routing table, which should be something like Destination Gateway Genmask Flags Metric RefUse Iface 193.154.142.0 0.0.0.0 255.255.255.0 U 0 0 116 eth0 127.0.0.0 0.0.0.0 255.0.0.0 U 0 04 lo 0.0.0.0 193.154.142.1 0.0.0.0 UG1 0 46 eth0 the kernel needs to know which ip-address of its default-gateway and over which interface that can be reached. /etc/init.d/network should add the appropriate routes when invoked via /etc/rcS.d/S40network. if that doesn´t help, do a ping -v to see where the unreachables come from, could be your host or some router on the way... But the routes are there. At least the routes for the local network are there. (2.2 kernels add them automatically, don't they?) I even tried to manually bring down all ethX, remove all routes and start them up again. Still doesn't work until I reboot in frustration. Can't even ping other hosts in the same network. Perhaps I overlooked something To Lindsay: I tried manually using ifconfig already :(
DPT SmartRaid V cards
Hi, I have been playing with a DPT smartRaid V card for some days, using the driver from their homepage. (the kernel drivers don't seem to work for SmartRaid V?) However, for some reason the Raid V drive doesn't seem to perform very well. It often just stops during disk writes without doing anything (no sound from the harddisks) I have a 1554U2 with 5 IBM SCSI harddisks. If anyone has used it before... was it an expected problem? I was using the driver for kernel 2.2.5 patched to 2.2.13. Or shall I play with 2.2.5 instead?
Re: Using dselect With a Proxy Server
On Fri, Jan 07, 2000 at 09:15:12PM -0700, Jason Gunthorpe wrote: do export http_proxy=http://10.10.10.2:80/; Then edit /etc/apt/sources.list and change all your URIs to http Alternativly, you can setup APT to speak to that ftp proxy, but that is probably lots more trouble than it is worth. Jason I have these in /etc/apt/apt.conf : ---START--- Acquire { http { Proxy http://proxy.hkbu.edu.hk:8080/;; No-Cache true; Max-Age 3600; No-Store true; }; ftp { Proxy http://proxy.hkbu.edu.hk:8080/;; }; }; END And a complete listing of options is in /usr/share/doc/apt/examples/apt.conf.gz . BTW I am using potato, perhaps it's different with the apt in slink?
Re: Standard way to change IP?
On Sat, Jan 08, 2000 at 12:10:41PM +0100, Robert Waldner wrote: On Sat, 08 Jan 2000 15:04:33 +0800, Jonathan Chang writes: Hi, all Is there any standard way for debian to change ip address? Or I need to modify files in /etc manully? Thanks in advance. modify /etc/init.d/network accordingly, do an ifconfig interface down, then run /etc/rcS.d/S40network I find that sometimes I cannot use that interface after the network is changed I get something like network unreachable when I try to ping some hosts on that network. The NIC is fine after a reboot. Most of the time I was using slink with kernel 2.2.13. Are there any other things that I need to do? Or was it driver dependent?
Re: Problem with exim rewriting rule
On Tue, Jan 04, 2000 at 03:45:36PM +0100, Klaus Drews wrote: Hi Ronald, wich section you put this in ? the section after the comments REWRITE CONFIGURATION, probably the same place as you put your rewrite rules? I am doing my rewrites this way: [EMAIL PROTECTED] [EMAIL PROTECTED] Ffr [EMAIL PROTECTED] ${if eq {$sender_address}{} [EMAIL PROTECTED] Tt Regarding to the docs, b and c stand for b rewrite the Bcc: header c rewrite the Cc: header Since my entry in exim.conf is bcfrF it should only work for b rewrite the Bcc: header c rewrite the Cc: header f rewrite the From: header r rewrite the Reply-to: header F rewrite the envelope From field Since I didn't say t, it shouldn't rewrite the To: header. But I will try your alternative if you tell me about the section to put it. I think so too. But perhaps there are some hidden caveats that you and I didn't recognize? If you need it I could send you my exim.conf. :)
Re: Exim
On Mon, Jan 03, 2000 at 03:14:02PM -0600, Matthew W. Roberts wrote: I have just recently upgraded from debian 2.1 to potato and I cannot send any email at all (either internal or external) i keep getting the following error message. 1999-12-21 14:34:58 [EMAIL PROTECTED]: procmail director deferred: file existence defer in procmail director: Permission denied Does anyone know what is going on and how to get it working? I have run eximconfig again and i select option 1 (internet mail) but that still doesn't work! This looks like a procmail problem, so running eximconfig probably won't help. Are you using a procmail director in your exim.conf file, or using the .forward method? I know that using procmail with a .forward file is different with exim than with sendmail so you might want to look at the exim documentation: http://www.exim.org/ If you send me your exim.conf file I might be able to help... I find that recent eximconfig will add a procmail director to exim.conf. with bugs. # This director runs procmail for users who have a .procmailrc file procmail: driver = localuser transport = procmail_pipe require_files = +${home}:+${home}/.procmailrc:/usr/bin/procmail ^^ hopefully these + shall fix the problem. no_verify IIRC, without these exim defers the mail when it cannot check the existence of .procmailrc (when the home dir is not readable) With the + signs exim just ignore the director and proceed to the next one.
Re: Problem with exim rewriting rule
I am doing my rewrites this way: [EMAIL PROTECTED] [EMAIL PROTECTED] Ffr [EMAIL PROTECTED] ${if eq {$sender_address}{} [EMAIL PROTECTED] Tt And I don't have the problem (mail delivered to [EMAIL PROTECTED] is ok) Fetchmail works too. I guess the b and c are for CC and BCC (need to look-up the docs for that) and should not matter, but well, perhaps it does. (The second line is there so delivery failures don't go out, and shouldn't affect normal mails too.) On Mon, Jan 03, 2000 at 04:09:22PM +0100, Klaus Drews wrote: Hi there, exim works fine with me, but produced one silly bug because of low -level configuration. I wanted to replace my loginname with my original email-adress, so that my ISP will accept the sender adress as correct. Therefore I created a mail-address file and put the following into the rewriting rules section: [EMAIL PROTECTED]${lookup{$1}lsearch{/etc/email-addresses}\ {$value}fail} bcfrF This part works fine. The bug occurs, when I get mail from for example lpd for my local account. Then the To-header will be changed and the mail will be delivered over my ISP instead of directly to my local account. But all I wanted was to change the From-header part. Now I know, that I can do it with exim this way, but I do not know how. I tried a lot and got lots of error messages. Can someone fill me an example to do this and a short explanation ? I don't want it to work without knowing why ... . (And please do not only refer to the docs. I read them all - it helped little). Thanks in advance, Klaus.
Re: Linux NAT and stuff.
There were actually 2 problems. 1) I cannot connect from the desktop machines to the NATd machine using either the private IP nor the real IP. This I have solved. I guess the problem came from the wrong routing table in the Notes server.. (however I am not allowed to change settings there, so I just moved the masquerading job to FW2, and it's working now) 2) I cannot connect from any parts of the internal network to the Notes server with the real IP. (I can connect from internal network using the private IP, so I guess it is not the same problem as you stated?) Everybody is satisfied with the current configuration, but it is really ugly to get an invalid argument like that. So... are there any possible solutions for this? On Tue, Dec 28, 1999 at 08:27:51PM -0800, aphro wrote: are you trying to access the NAT'd machine from infront of the debian box doing the NAT ? from the looks of it you are doing NAT on only part of the network.. the desktop PCs section (?) You will not be able to access the NAT'd machines from infront of the debian box doing the NAT even if its on the same network. If you need this functionality you need something that can do reverse NAT. i hope i understood your problem :) nate
Linux NAT and stuff.
Hi all, I am starting to use Debian (potato) as a firewall with NAT functions. I have fast NAT compiled into the kernel, installed iproute2, read through the documentation ip-cref and did what was suggested in Appendix C. Everything looks fine. Except ... I cannot connect to the NATed machine from the internal network. My (approx) network topology: INTERNET --- FW1 [172.16.29.254] ---+--- [172.16.28.2] NT Lotus Notes | | [172.16.29.1] FW2 [172.16.28.1] | | [172.16.28.x] desktop PCs (don't ask me why 2 firewalls are needed, I don't know :( ) I have IP Masquerading and the NAT running in FW1 (172.16.28.x uses MASQ, 172.16.29.2 uses NAT and ipchains is set to just forward packets) I can connect to the Notes server from the Internet. desktop PCs can connect to the Internet and the 2 FWs. The 2 FWs, of course, can go anywhere. I can connect from FW1/2 to the Notes server through 172,16.29.2. However (here's the problem), I cannot connect from desktop PCs to the Notes server. Also, if I try to connect to the Notes server from FW1 using the NATed address I get an invalid argument error. What was the cause of these 2 error? The ip commands are something like this: /sbin/ip route add nat $EXTIP via 172.16.29.2 /sbin/ip rule add prio 1000 from 172.16.29.2 to 172.16.0.0/16 table main /sbin/ip rule add prio 1001 from 172.16.29.2 nat $EXTIP The documentation mentioned a table called inr.ruhep. Was the name arbitrary? Appendix C mentioned this table should contain route to the destination, but I don't know what that is supposed to be.. Shall I use FW2 to do masquerading, and FW1 to provide NAT for FW2 and Notes? Will it help the situation? I just noticed that it should be easier to manage this way. (I really think I should have posted it somewhere else. should I? And if yes, where should I post?) Hope it doesn't look too difficult to understand. My english isn't that good. :(
on openldapd
Hi all, I have a openldapd running on a potato. Whenever I upgrade it, I find that the ldbm files fail me (ldapsearch fails). If I revert to the old version the query is ok. At last I downgraded, dumped all the contents, then upgrade and run ldif2ldbm. (Argh, I just find it in the critical bugs list..) So, are there better ways to bypass that for now? Or should I just mark it as hold for the moment? And that I find that I can never login with the default admin password that I typed when I install openldapd. Everytime I tried (re)installing it it give me a {crypt}ed password that is much longer than the expected one of 13 characters. And then.. how to stop slapd from reporting every query? I was trying -s. for a value of 0 it is giving me more stuff...
3c905c wierd troubles
Hi all .. I have 2 3c905c on a Debian potato and have configured one to have $IP1, the other $IP2. When I ping $IP1 or $IP2 I get an error like neighbour table overflow. But ping'ing other hosts work just fine. I was wondering if anyone else have similar problems? Is it a mis-configuraton of mine, a kernel bug, or is it a debian-related bug? The machine has an ATX board with wake-on-lan features, but I think I have it disabled. (I may have forgotten to save the BIOS changes... though unlikely) The CPU is some PII or PIII, with 64/128MB RAM.. Having 2 IDE harddisks and an IDE CD-ROM. (I think they don't matter?) And then. on another machine I keep seeing System.map doesn't match kernel data. Are there anything I can do? It's one of the Compaq Proliant servers (I think 1600). I compile the kernel with make-kpkg clean ; make-kpkg --revision=custom.1.0 kernel_image It doesn't seem to hurt anything up to now, but it's ugly when you see such errors when you do ps...
Mail servers for large numbers of users
I am about to build a (group of?) e-mail server for a large number of users... more than 65536. Using Debian of course. :) It should include SMTP and IMAP. Users do not need to have login accounts. Probably I will be using Potato. What should I start with? Are there good open sourced MTA and IMAP servers that do not operate on system password accounts? (so I don't have to create 6+ accounts) For now I will be trying to distribute the accounts on more than 1 server, set-up the DNS to have more than 1 MX, and forward the mails internally through SMTP for non-existent local users. Which MTA is good for this purpose? For ease of configuration I may be choosing exim, because this is what I have been using at home (it's the default MTA for Debian.. :) I see that exim can match user names with dbm/ldap lookups as well.. which may prove useful. What about qmail or postfix? Are there other precautions that I have to take care? Are there any other precautions?
Re: Mail servers for large numbers of users
On Sat, Dec 11, 1999 at 05:07:53PM -0200, Henrique M Holschuh wrote: A long time ago, in a galaxy far, far way, someone said... It should include SMTP and IMAP. Users do not need to have login accounts. Probably I will be using Potato. What should I start with? Either exim or postfix, definitely - they're very easy to configure. I've not had any experience with anything else. I've seen setups like these being mentioned in the Postfix ML (more than one person there claimed more than 10 users under Cyrus IMAPD+Postfix). Sorry if this question looks dumb... How could I deliver mails to accounts that don't really exist? (I can't allocate 10 uids on a single machine, right?) I have only read the FAQ and anatomy for postfix Shall I play with the mailbox transport option for local delivery, or do I create local users with same uid (and disable their login) ? Do the 2nd solution really work? But still it doesn't seem very efficient to store 10 files in a single mail spool directory? Cyrus IMAPD seems to be the way to go for that many accounts, but beware their non-dfsg license. They like ISPs, though ;-) Cyrus is not available as a debian package, though there's a pending ITP. Are there good open sourced MTA and IMAP servers that do not operate on system password accounts? (so I don't have to create 6+ accounts) The MTA won't give you any problems - they typically don't care about the password database. However, here's another reason why you don't want to But the MTA have to lookup the local user from the password database.. and mails are stored under owner's uid. That's the case for sendmail and exim at least. And, thanks for all the replies. :)
Compaq Smart Array 221
Hi all, I have a Compaq Smart Array 221 on a Proliant 1600 server, and have already configured the disk array. I plugged another SCSI drive into the first SCSI channel and could detect the logical drives with a 2.2.13 kernel booting from that SCSI drive. So hopefully I can get the RAID running without trouble. At this point it's pretty late so I left the office. There are some little questions that I probably could solve myself, but help would save me some time and trouble. :) So the questions: 1) The CD allows me to make quite a number of logical drives. Should I treat them as a disk partition and run mke2fs on them or shall I treat them like harddisks and fdisk them? (in this case where are the devices for the partitions?) 2) The only Debian specific question. :) I learned from the documentation that I need to patch LILO in order to boot from these devices. A search from freshmeat shows me that it is available from: http://www.insync.net/~frantzc/cpqarray.html I think I will fetch the source package, apply patch and build a binary package from it. - Will it work? - Is it the best way of doing it? - How could I prevent apt from overwriting custom-built LILO? 3) Are there any other precautions that I should know of when playing with Smart Array cards (or RAID5) with Debian potato? Must I read through the whole user manual? :( Any help would be appreciated. And, it might not be that good to check my home mail in office. (I have other personal mail) If at all possible, could suggestions be CC'ed to my mail address at work too? ([EMAIL PROTECTED])
Re: debian-user-digest Digest V99 #1885
No. That's Chinese. Specifically that's a Hong Kong company selling e-mail addresses. (phone number starts with 852) The Subject showed correctly (as BIG5 Chinese) on the digest too. (I am using crxvt,mutt with locale set to zh_TW.Big5) Yes, I am living in Hong Kong, where stupid marketing guys will send Chinese ads into English-only mailing list. Date: Fri, 29 Oct 1999 16:32:50 +0200 From: Jean-Yves BARBIER [EMAIL PROTECTED] To: debian-user@lists.debian.org Subject: Re: 7s(%\%+3ufP,9q$l*=P2000: Message-ID: [EMAIL PROTECTED] Content-Type: text/plain; charset=us-ascii On Fri, Oct 29, 1999 at 03:16:00PM +0200, Ingo Reimann wrote: On Fri, Oct 29, 1999 at 03:16:24PM +0200, HCI wrote: kewl, what kind of language is that? Ingo Seems to be korean.
EXIM settings for dialup.
Not sure if I should post here (but I'm running exim on a debian :) So, I have a Debian unstable running exim 3.03, connecting to the Internet through PPP without fixed IP. I have configured exim to use smarthost(*) to send mail, performing address rewrite like this: [EMAIL PROTECTED] [EMAIL PROTECTED] Ffr And can send mail just fine (or else you can't see this) However there is something stupid when I send mail to myself: Address rewrite is performed even for local deliveries. How can I disable this? I guess I should disable address rewrite for everything inside the directors. But... how? Or am I plain wrong? Should I disable address rewrites for everything, and enable them only for the routers instead? *) In fact I have more than one router. The smarthost router runs only if the sender is me. I does this through condition = ${if eq [EMAIL PROTECTED] {yes}fail} It works fine but is less then clean. Are there better configuration options for this? From what I read in the spec this is the best? ( And that I am subscribed to the digest instead. Do I get every single post, including this one and follow-ups for it? )
Re: EXIM settings for dialup.
Sorry that I didn't make it clear.. :( I don't have much problems sending the mail through alumni.ust.hk. And no problems for sending local mail too. (There was quite a number of problems, but were solved before I decided to post here :) The problem is that local mails suffers from the address rewrite.. Just a bit ugly :( And yes, I find that I can see my mail and replies to it from the digest. Thanks for forwarding me a copy, Jonathan) On Sat, Oct 23, 1999 at 03:11:59PM -, [EMAIL PROTECTED] wrote: Date: Sat, 23 Oct 1999 10:43:21 + From: Jonathan Heaney [EMAIL PROTECTED] To: Ronald Tin [EMAIL PROTECTED] CC: debian-user@lists.debian.org, [EMAIL PROTECTED] Subject: Re: EXIM settings for dialup. Message-ID: [EMAIL PROTECTED] Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Ronald Tin wrote: Not sure if I should post here (but I'm running exim on a debian :) So, I have a Debian unstable running exim 3.03, connecting to the Internet through PPP without fixed IP. I have configured exim to use smarthost(*) to send mail, performing address rewrite like this: [EMAIL PROTECTED] [EMAIL PROTECTED] Ffr And can send mail just fine (or else you can't see this) However there is something stupid when I send mail to myself: Address rewrite is performed even for local deliveries. How can I disable this? I guess I should disable address rewrite for everything inside the directors. But... how? Or am I plain wrong? Should I disable address rewrites for everything, and enable them only for the routers instead? *) In fact I have more than one router. The smarthost router runs only if the sender is me. I does this through condition = ${if eq [EMAIL PROTECTED] {yes}fail} It works fine but is less then clean. Are there better configuration options for this? From what I read in the spec this is the best? ( And that I am subscribed to the digest instead. Do I get every single post, including this one and follow-ups for it? ) -- Unsubscribe? mail -s unsubscribe [EMAIL PROTECTED] /dev/null I had to do this myself, I got this info. from Oskar Liljeblad ([EMAIL PROTECTED]) (in /etc/exim.conf): # Main Configuration Settings qualify_domain = alumni.ust.hk hem.passagen.se is my email ISP, and without this they wouldn't let me relay mail through their server. qualify_recipient = localhost This is necessary so you can do mail vamp and receive the mail directly without going through the ISP. The new mail will have To: [EMAIL PROTECTED] but still From: Ronald Tin alumni.ust.hk where the latter was rewritten by the local MTA. local_domains = localhost If you use eximconfig to generate a smarthost setup, it is possible that your smarthost will be included here. But it shouldn't. Sorry for the plagiraism, Oskar, but you knew how to fix it Jonathan