Re: Timeservers in the Netherlands
At 01:30 PM 1/27/00 +0100, Remco van 't Veer wrote: [snip] -- Northold Saddam Hussein amfetamine Roel van Duin Honduras PLO Kosovo thrust supercomputer NSA Beatrix FSF XTC hacker semtex You KNOW that you're being logged... Euh, SH*T now I'm logged too!!! Sincerely, Usama Bin Laden.
Re: CPU question
Most of the time servers are I/O bound. So it would be wise to spend your money on a good NIC, SCSI adapter, SCSI hard disk, fast (IO wise) MOBO, etc. If you can re-use the P166 then do it, otherwise put a nice celeron in your server and spend your money on the I/O stuff... Regards, Onno At 05:19 PM 1/28/00 +1100, Shao Zhang wrote: Hi, I am going to set up a new server which will serve pretty much everything except www. At the moment, it has a P166, and the load avg is around about 5. The highest once reached to 22. In Australia here, the price of getting two P550 and one P700 is roughly the same. So should I get two slower CPUs, or one fast CPU. Which case will the server run faster/perform better? Thanks in advance. Shao. -- Shao Zhang - Running Debian 2.1 ___ _ _ Department of Communications/ __| |_ __ _ ___ |_ / |_ __ _ _ _ __ _ University of New South Wales \__ \ ' \/ _` / _ \ / /| ' \/ _` | ' \/ _` | Sydney, Australia |___/_||_\__,_\___/ /___|_||_\__,_|_||_\__, | Email: [EMAIL PROTECTED] |___/ _ -- Unsubscribe? mail -s unsubscribe [EMAIL PROTECTED] /dev/null
Primary and Secondary DNS one 1 server...
Because of an upgrade of our computer network I must run the primary and secondary DNS one 1 server with two NIC's for a while... Has anyone experiance with this? My guess would be to run named twice and point to two config dirs and edit the named.conf seperatly to run each named on the right interface... But thats just a guess! Regards, Onno
Re: Attention: imapd gpoing back to $HOME as mailbox root
At 10:23 PM 1/24/00 -0500, Jaldhar H. Vyas wrote: On Mon, 24 Jan 2000, Joe Block wrote: You aren't. Count me as a vote for $HOME/mail I can't keep flip-flopping on this. The default mailbox root will remain $HOME. You have three choices. 1. Learn to live with it. (Remember you can always tweak your client.) 2. Recompile the package to your taste. 3. Find a different package. I hate to be dictatorial but unfortunately their is no solution that will please everyone. Maybe there is... You could patch the program to set the variable at startup from a parameter or configfile. I haven't seen the source but reading a variable from a parameter or config file at startup shouldn't be hard... Regards, Onno
Re: ipchains diagnostics
IP forwarding and IP masq-ing are enabled in the kernel? Regards, Onno At 10:29 AM 1/24/00 +, Patrick wrote: I can't get ipchains to work and get no error messages when I run echo 1 /proc/sys/net/ipv4/ip_forward /sbin/ipchains -P forward DENY /sbin/ipchains -A forward -s 10.0.0.0/24 -j MASQ How can I tell what's wrong? I see nothing in logs nor tail -f /var/log/messages. Patrick
Re: ipchains diagnostics
At 01:56 PM 1/24/00 +, Patrick wrote: ip masq is definately enabled in the kernel but not sure about ip forwarding. My real question is how can I diagnose the error. Where can I get a message what setting I've missed as I know this is Operator Error. If 'ls /proc/sys/net/ipv4' yields a 'ip_forward' then the kernel supports ip forwarding. But futher diagnostics is done step bij step: (others: please correct me when I'm wrong because I'm doing this 'on the fly') - check the kernel for support - check kernel parameters - check ifconfig - check route - check ipchains Good luck, Onno
Re: how many users per apache proc?
At 10:11 PM 1/19/00 -0800, aphro wrote: i was wondeirng if anyone knew approx how many connection 1 apache process could handle? just 1? or is it more.. As far as I know just 1, they [apache team] are working on threaded processes for increased performance while using less memory. However they still plan to use a few processes so that if one process dies with all it's threads not the whole webserver dies... Sounds realy good, you can read it on thier web site. Regards, Onno
Re: Transparent network bridge+filter?
At 08:49 AM 1/19/00 -0600, Jeff Noxon wrote: You have an interesting idea, but it won't work in my case. I have to put this between a pair of Cisco routers running EIGRP. They won't see each other if the router discovery packets (etc.) aren't forwarded by a bridge. I also can't guarantee that the address of the router on one side won't change -- it is not under my control. As far as I know Linux doesn't understand EIGRP. I can't even find it in /etc/protocols... Regards, Onno
One step further (WAS: Re: Advanced hard disk mirroring!)
The high quality replies I received (especially from Jens B. Jorgensen) solved my problem. (see http://www.debian.org/Lists-Archives/debian-user-0001/msg02027.html) It also gave me an idea: Wouldn't it be great if there was ONE Linux boot-floppy that would mount (SMB or NFS?) a complete filesystem after the initial boot over a network (and complete the boot sequence?). I'm probable not the first who came up with this idea, so does anyone know If such a thing is already pulled off? Regards, Onno
Strange LILO question
Is it possible to set LILO on a DOS floppy disk with the default boot for the hard disk and an alternate boot for the DOS floppy itself? Regards, Onno
Re: Strange LILO question
At 01:43 PM 1/20/00 +, Ethan Benson wrote: On 20/1/2000 Onno Ebbinge wrote: Is it possible to set LILO on a DOS floppy disk with the default boot for the hard disk and an alternate boot for the DOS floppy itself? if you mean install the lilo boot sector on the DOS floppy then no, dosfs does not leave any room for a bootblock, so installing lilo on it would destroy the dos filesystem, which would not be problem except you say you want to boot from it. Yes, that is what I ment, maybe I should put LILO on the hard disk and then set the default boot to the hard disk with an option to boot from A: ??? you can however install lilo on a floppy and have it boot the hard disk. Is there then realy no way to boot the floppy also WITH a DOS filesystem? Probably not :-( Thanks, Onno
Re: Transparent network bridge+filter?
At 02:08 PM 1/18/00 -0600, Jeff Noxon wrote: Can anything that runs on Linux do reliable network bridging filtering? I need a transparent filter that I can drop into an existing network. Ipfilter will do the job with Open/NetBSD. It may work on Linux, but requires kernel 2.0.35 and isn't compatible with glibc. Another guy ask -something like that- before, I replied with an answer that worked ;-) Here is my reply and maybe you can use parts of it: (You don't want to use this route config ;-) This has been a while but here it goes: Please test if the next settings will do the trick. The debian box cannot be reached from the inet or lan, We can do something about the lan connection though... Note: Filtering firewall is WIDE open! Note: There is a route for all IP's because they are on the same subnet (netmask) but NOT on the same network device! Note: Youre gateway is 63.225.131.78 root# ifconfig lo 127.0.0.1 root# ifconfig eth0 0.0.0.0 promisc root# ifconfig eth1 0.0.0.0 promisc root# route add 63.225.131.73 eth0 root# route add 63.225.131.74 eth0 root# route add 63.225.131.75 eth0 root# route add 63.225.131.76 eth0 root# route add 63.225.131.77 eth0 root# route add 63.225.131.78 eth1 root# ipchains -P input ACCEPT root# ipchains -P forward ACCEPT root# ipchains -P ouput ACCEPT root# ipchains -F root# ipchains -X Please send me your results Good luck! Onno
Re: New drive ready to partition. Just what some recommendations and suggestions.
Hmmm, does incremental backups sound good in this situation? Anyone? Regards, Onno At 07:44 PM 1/18/00 +, John Gay wrote: I've got some good suggestions, and apparently raised a few questions as well. Let me outline my reasons for asking and what I hope to do: I've got a CD-RW. I plan to use this for back-ups as well as software publishing. I've also got a SCSI tape drive, but I'm not quite sure how to use it yet. I've got / as a 500M partition. This is perfect for putting onto a bootable CD-RW for emergency recovery. Being new, I've already screwed up my system to the point where it wouldn't boot. Last time I had to do a complete install. I would like to keep the other partitions small enough to put onto just a few CD's each. Worst-case scenario, I trash a complete partition, I can recover from just a few CD's. This also makes backing up each partition less work. I also want to have a few partitions set aside for CD images. I would feel better having a number of smaller partitions that I can back-up and recover quickly, plus fsck would run faster, that just a few really big partitions and lots of sym-links to hide the facts. I've read the multi-disk HOW-TO, as it has some good info on partitions sizes and such. I've also read the FHS info as well. I know StarOffice wants to install in /opt, and apt-get uses /var quite heavily. I expect only three users, and maybe a business account, so mail shouldn't be too much. At the moment I'm more concerned with being able to recover the various system and user programmes before I make my next mistake as root, rather than and user data. This would indicate a good scheme for recovering /usr. Probably spitting it up may help, as long as my / partition would have enough to recover the rest of the system in case of catastrophic failure. Thanks again for your input. I'll have another read of the FHS documents and a good think.
Advanced hard disk mirroring!
I'm a sysadmin and have two Debian GNU/Linux potato servers and 50 windows 95 workstations under my care. My problem is with the 50 workstations: (the 50 workstations have the same hardware) I want to install ONE workstation and then mirror the hard disk to all other workstations. The first time I did it was with ghost and I hooked up the installed hard disk to every client and mirrored it... Not a nice job to do and there must be a better way do do it! I know that ghost CAN use NETBIOS connections and such but I don't know how to boot it from a 1.44 flop and then use ghost. My question: I want to boot the workstation to be installed from a floppy with Linux or DOS. Then I want to make a connection to my server (or the installed workstation) and mirror the hard disk from a file (or hard disk). Frankly I don't care what is used or how it's being done (Linux/DOS with dd/ghost or something else!) but I don't want to hook-up all 50 workstations again... Thanks for any ideas, Onno
Re: New drive ready to partition. Just what some recommendations and suggestions.
Sometimes I don't understand the stratagies used in disk partitioning. Please correct me if I'm wrong but I always thought that you split the partitions by long term usage: 1- 2 GB / 1- 2 GB /var 1- 4 GB /var/spool rest on /home Then I link /tmp to /var/tmp WHY: The chance that / is being filled by a user is small this way. Email and printing activities are split from other partitions. Logging, tmp and such are split from other partitions. And last but not least home directories are split from other partitions. Ideas and critical remarks are welcome... Groetjes, Onno At 12:48 PM 1/18/00 +0100, Ron Rademaker wrote: I recently got a new 13G hard drive. I've installed it as hdb, and moved my CD-RW to hdc. At the moment I've got a 6G drive with 2G for WindowsNT, 100M for /, 1G for /home and 2G for /usr. I really need more room for both /home AND /usr, but I also need more space for /var and /opt and some others. I would like to make several partitions and use them to my best use, I just wanted to get some recommendations from the Expert/Experienced before I partitioned this drive. I also would like to have a few partitions set aside for CD images as I hope to start selling Software on CD as well. I don't know if you're willing to reinstall completely or just want to add the space, if you're willing to reinstall I would: 1) Your current hd 4 GB on / 2) New hd 5 GB on /home 3) New hd 8 GB on /usr That way you'll have more on /home, /usr, /var and /opt (both /var and /opt will be on /). If you don't want to reinstall you could set a 5 GB on /home and remount your current /home to /var or /opt (whatever you prefer). That way you'll have more on /home and on /var or /opt. Then you create the 8 GB, mount it somewhere, cp the /usr to that partition, following you mount your current /usr to /var or /opt (just the one you hadn't used) and mount that new partition to /usr. (It'll probably need a reboot because you can't umount your /usr (maybe you can after a init 1 (runlevel 1 single user), I've never tried). Ron -- Unsubscribe? mail -s unsubscribe [EMAIL PROTECTED] /dev/null
Re: help -- UDMA install....
Just a few things you should check: Master/slave settings, isa card interferance, broken PCI cards, BIOS init, BIOS: NO PNP OS, broken chipsets, etc I hope this helps... Regards, Onno At 01:26 PM 1/14/00 -0600, Marc D Chapman wrote: I'm currently trying to help a friend install Linux onto his system. However, we are running into some massive problems. Whenever it gets to the point where it is checking the disk for errors, or formatting the disk (basically any large amount of disk access), the system will lock up completely. One of his drives is a UDMA with on-motherboard support for this (there are a total of 3 hard drives and 1 CD-ROM on the IDE), the drive is a Maxtor 6 GB (I don't recall the model number). We tried a number of different distributions (Debian 2.1, Redhat 6.1, Mandrake), but all do the same thing. Does anyone have any ideas as to what could be up (I checked the UDMA HOWTO, which didn't help, and also tried playing with drive settings (LBA vs. Large) also to no avail). Any help is appreciated marc Any intelligent fool can make things bigger, more complex, and more violent. It takes a touch of genius -- and a lot of courage -- to move in the opposite direction -Albert Einstien If we ban whatever offends any group in our diverse society we will soon have no art, no culture, no humour, no satire. -Erica Jong [EMAIL PROTECTED] [EMAIL PROTECTED] [EMAIL PROTECTED] [EMAIL PROTECTED] -- Unsubscribe? mail -s unsubscribe [EMAIL PROTECTED] /dev/null
Re: root should be able to do anything... right???
Try lsattr and chattr... Regards, Onno At 12:57 PM 1/17/00 +0100, Ron Rademaker wrote: Last week I've send a mail about a weird file, that should be a directory but it was a character device. (/usr/X11R6/include/X11/fonts). It also had weird permissions: c---r- 1 8224 10280 49, 117 Dec 1 2031 fonts Now I'm trying to remove this file, the system won't allow me (yes ofcourse I'm root). I've also tried rm -f. I've tried to chown the 'file'/'character device' to root and a chmod, the only thing I get to see is Operation not permitted. Any suggestions??? Ron -- Unsubscribe? mail -s unsubscribe [EMAIL PROTECTED] /dev/null
Re: /etc/limits
At 06:34 PM 1/9/00 -0500, Jim B wrote: OK another issue I'm having with setting resource limits. How can I [snip] I look in my /etc/limits and see a way to restrict just about all those [snip] Where can I find more info on /etc/limits ? Regards, Onno
Ignore this test...
I'm very sorry but I have to do this test. One of my email filter failed and I have to see if they work properly now. I hope you understand... Regards, Onno
Ignore this test...
I'm very sorry but I have to do this test. One of my email filter failed and I have to see if they work properly now. I hope you understand... Regards, Onno