Re: [techtalk] Download throttling?
On Sun, Sep 03, 2000 at 05:01:47AM +, [EMAIL PROTECTED] wrote: > Dancer and I have a similar situation (with a 28.8K modem!). Well we have a 56K. I actually consider this fairly good since my previous experience with modems and the net was in regional Austrlia sharing a 28.8 with the rest of the school... :) > Apparently in Linux 2.2 or later, 'quality of service and fair > scheduling', under network options. All the facilities are > implemented by that. The IProute package is used to configure it. > And, to quote him, 'Good luck figuring out how'. Oops. My flatmates regarded this as a challenge... > Dancer says he heartily recommends the 'D4x' package which is a transfer > manager for the graphic environment, but has built in rate limiting at > any levels you like. IE: if you all use somehting like this to transfer > the > files, you can restrict your own downloads. :/ That's OK, except in the situation where I'm doing some Very Important Surfing and soeone else is running something commandliney like 'apt-get update'. > Good luck. Sorry to be less than helpful - if anyone has a better > solution, PLEASE speak up! We'll use it too! Thanks, Mary. ___ techtalk mailing list [EMAIL PROTECTED] http://www.linux.org.uk/mailman/listinfo/techtalk
Re: [techtalk] installing new hd
On Sun, Sep 03, 2000 at 12:41:42PM +1100 or so it is rumoured hereabouts, Malcolm Tredinnick thought: > On Sat, Sep 02, 2000 at 06:22:54PM -0700, Allen Heinecke wrote: > > I am considering installing a new HD in my computer as the old one is > > running out of space... however the "partition" / location that is running > > out of space is root... > > > > what is the best way to migrate all of the root files, including the > > bootup files to this new HD? > > There are two situations you may be in here. I'll give some hints to > both and then others can flame me. > > (A) You are wanting to completely replace the old HD with new one (the > more difficult solution). > > Hook up the new HD to your machine and use something like fdisk or > disk druid to partition it as you like. Then mount each partition > under an appropriate directory. For example, suppose the new disk > will have partitions called '/', '/tmp' and '/usr' (as partitions 1, > 2 and 3), I would make directories under /mnt called 'root', 'tmp', > 'usr' and then use the commands.. > > mount -t ext2 /dev/hdd1 /mnt/root > mount -t ext2 /dev/hdd2 /mnt/tmp > mount -t ext2 /dev/hdd3 /mnt/user > > Then copy everything from your current HD to the new home on the new > HD. Note that you will not need to copy the contents of /proc (they > are not real files). > > Adjust /mnt/root/etc/fstab to mount the new partitions in the right > spots (bearing in mind, that after you swap the new for the old, it > will be /dev/hdc, so you would mount /dev/hdc1 under /tmp in the > above example). > I'm presuming when you refer to /dev/hdd and /dev/hdc you really mean /dev/hdb and /dev/hda which would be the more likely arrangement in a 2 disk system and /dev/hda is definitely need if LILO is to be the boot manager. (Yes, I know one could have the two disks on seperate controllers etc and am simplifying here) > I _think_ you probably need to rerun LILO so that the kernel copy > that is on the new drive understands where it lives. There may well > be a way to do this at this point in the plan, but I can't work out > how to get LILO to write the result to /dev/hdd's MBR in a way that > will work when you reboot it as /dev/hdc. Somebody else may be able > to solve this (please?). Instead, I propose a different plan in the > next paragraph. > Yeah, no bother... A few things to do though... LILO, by default uses /etc/lilo.conf for its configuration. Let's have a look at a typical /etc/lilo.conf ## lilo.conf # boot=/dev/hda #This is where lilo will write to map=/boot/map #I don't know what the map is for :-) install=/boot/boot.b#Similarly prompt timeout=50 default=linux image=/boot/vmlinuz-2.2.14-5.0 #This is the kernel to boot label=linux read-only root=/dev/hda7 #This is the root partition end Now, since you have the new disk installed as /dev/hdd at present and have its various partitions mounted under /mnt/ You need to modify lilo.conf to reflect these changes. This is a once off modification so it would be worth copying lilo.conf to lilo.conf.hdd and changing that instead. Here's the modified lilo.conf.hdd: ## lilo.conf.hdd # boot=/dev/hdd #This will be the primary MBR when the #disk is installed as the primary map=/mnt/root/boot/map #I don't know what the map is for :-) install=/mnt/root/boot/boot.b #But you need to add the /mnt/root bit #anyhow prompt timeout=50 default=linux image=/mnt/root/boot/vmlinuz-2.2.14-5.0 #This is the kernel to boot label=linux read-only root=/dev/hda7 #This is the root partition to be mounted #on boot, don't change this end Now, do a: /mnt/root/sbin/lilo -C /mnt/root/etc/lilo.conf.hdd to install. This info is derived from the Hard-Disk-Upgrade-(mini-?)HOWTO now you can skip this bit ### > Shutdown the machine, put the new HD in place of the old one and > reboot using a boot floppy (if you don't have one of these, or can't > create one, ask again on this list and I, or somebody else, will try > to help). Run /sbin/lilo on the new HD and it should correctly alter > the bootup information. ## > > ** Caveat Emptor: What I have described is completely untested!! > However, it looks like it will work. Whatever you do, don't do anything > destructive to the old HD until you are sure the swapover worked. > Alternatively, go with plan B (below)... > > (B) You want to add in the second HD to the system, but wi
Re: [techtalk] installing new hd
On Sat, Sep 02, 2000 at 08:39:45PM -0700 or so it is rumoured hereabouts, Allen Heinecke thought: > Ok... B) is the method that I was looking at using... but it just seems > extremely involved and quite bulky... I mean... look at all the extra crap > that you're leaving on the original drive... Sorry, What extra crap? Are you talking about the contents of /usr that remain on the old drive while you check out that everything transferred properly? But, you delete that later. [snip] > > On Sun, 3 Sep 2000, Malcolm Tredinnick wrote: > > > On Sat, Sep 02, 2000 at 06:22:54PM -0700, Allen Heinecke wrote: > > Alternatively, go with plan B (below)... > > > > (B) You want to add in the second HD to the system, but will keep the > > first on installed as well (much easier!). > > > > Partition the new HD as you like and make each partition a "major" > > part of the directory structure, like /usr/local or /tmp or /usr or > > whatever. For the time being, mount these new partitions under > > appropriate directories under /mnt (e.g. /mnt/usr/local) using the > > mount commands I gave above. > > > > Copy everything from the current /usr directory over to what will be > > the new /usr directory (just an example). So in this example, you do > > something like > > > > cp -r /usr/ /mnt/usr This should read cp -dpR /usr /mnt/usr To preserve permissions etc. > > > > Do this for all the directories you want to move across. > > > > **NOTE: You should leave your kernel image (probably in /boot, so > > leave all of /boot), /etc, /bin and /sbin on the first harddrive, > > otherwise you will have trouble booting! > > > > Now alter /etc/fstab to mount the new partitions under their new > > directory names. So, for example, if the partition mounted as > > /mnt/usr in the above case is /dev/hdd3, you would add a line to > > /etc/fstab like > > > > /dev/hdd3 /usrext2defaults1 2 > > > > You can now reboot and verify that the new partition was mounted > > under the right spot. > > ## See, This bit... # > > To reclaim your disk space, remove the above line from fstab, > > reboot again (this puts the old /usr directory back) and then > > remove all the contents of /usr (which frees up space on the > > first HD). Finally, add the fstab line back and reboot again. > > What could be easier? ## End of See, This bit... ## -- Conor Daly <[EMAIL PROTECTED]> Domestic Sysadmin :-) ___ techtalk mailing list [EMAIL PROTECTED] http://www.linux.org.uk/mailman/listinfo/techtalk
Re: [techtalk] Help - glibc-update -> broken tcsh!
Magni Onsoien: > We are using RedHat 5.2 on my company's server, and AutoRPM for keeping > the packages updated. Last night glibc was updated to glibc-2.0.7-29.2, > and then tcsh (tcsh-6.07.09-1) stopped working! Whenever started, it > core-dumps. RedHat did the mistake, and I have got and tested a new RPM now. Guess it will be public by Monday/Tuesday or so. Upgrading tcsh wouldn't work, there was an error in glibc. The error was only in the 5.x-RPMs, so those of you running redHat 6.x should upgrade glibc ASAP as there are severe security problems with older versions. Magni :) -- ulimit is good for you. ___ techtalk mailing list [EMAIL PROTECTED] http://www.linux.org.uk/mailman/listinfo/techtalk
Re: [techtalk] Download throttling?
Mary Gardiner wrote: > Well we have a 56K. I actually consider this fairly good since my previous > experience with modems and the net was in regional Austrlia sharing a 28.8 > with the rest of the school... :) I'm in urban Australia. :) Melbourne, actually. > > Apparently in Linux 2.2 or later, 'quality of service and fair > > scheduling', under network options. All the facilities are > > implemented by that. The IProute package is used to configure it. > > And, to quote him, 'Good luck figuring out how'. > > Oops. My flatmates regarded this as a challenge... Good! Get them to document their results? Please? :) Jenn V. -- "Do you ever wonder if there's a whole section of geek culture you miss out on by being a geek?" - Dancer. [EMAIL PROTECTED]Jenn Vesperman http://www.simegen.com/~jenn/ ___ techtalk mailing list [EMAIL PROTECTED] http://www.linux.org.uk/mailman/listinfo/techtalk
Re: [techtalk] Download throttling?
On Sun, Sep 03, 2000 at 01:31:45PM + or so it is rumoured hereabouts, [EMAIL PROTECTED] thought: > Mary Gardiner wrote: > > > Well we have a 56K. I actually consider this fairly good since my previous > > experience with modems and the net was in regional Austrlia sharing a 28.8 > > with the rest of the school... :) > > I'm in urban Australia. :) Melbourne, actually. > > > > > Apparently in Linux 2.2 or later, 'quality of service and fair > > > scheduling', under network options. All the facilities are > > > implemented by that. The IProute package is used to configure it. > > > And, to quote him, 'Good luck figuring out how'. > > > > Oops. My flatmates regarded this as a challenge... > > Good! Get them to document their results? Please? :) > > Just had a look at some of the documentation... Informative? Hah! Just half a page on traffic control but it raises some questions. I'm using a similar setup: 486/66 firewall/router/dial-on-demand with ipmasq for the net with clients and server behind it. Now, one can set up a tunnel from a source IP to a destination IP through a device (real or virtual) and limit the bandwith, burst size and queue size. Queueing cannot be implemented on a virtual device so the virtual must be mapped to a real device. So, the questions arise... 1. Can multiple virtual devices map to a single real device (eg ppp0)? 2. If ipmasq changes all packets to an outbound ip address on the way to ppp0, do you end up with all traffic going down a single tunnel? 3. Do you end up having to do forwarding to different tunnels for each client / protocol first and then ip masq the tunnels on their way to ppp0 (maybe bind the tunnels to eth0 rather than to ppp0)? 4. Can you adjust the bandwidth parameters on the fly or only at setup of the tunnel? Can a single client on-line get the benefit of the full bandwidth and get throttled back when another user goes on-line? 5. Does all this mean an 'orrible mess of a routing table? -- Conor Daly <[EMAIL PROTECTED]> Domestic Sysadmin :-) ___ techtalk mailing list [EMAIL PROTECTED] http://www.linux.org.uk/mailman/listinfo/techtalk
Re: [techtalk] Download throttling?
Conor Daly wrote: > > > > > Apparently in Linux 2.2 or later, 'quality of service and fair > > > > scheduling', under network options. All the facilities are > > > > implemented by that. The IProute package is used to configure it. > > > > And, to quote him, 'Good luck figuring out how'. > > > > > > Oops. My flatmates regarded this as a challenge... > Just had a look at some of the documentation... Informative? Hah! Dancer found a HOWTO! 'Linux 2.4 Advanced Routing HOWTO' apparently has info. It's current to five days ago, it seems, and can be found at: $ export CVSROOT=:pserver:[EMAIL PROTECTED]:/var/cvsroot $ cvs login CVS password: [enter 'cvs' (without 's)] $ cvs co 2.4routing cvs server: Updating 2.4routing U 2.4routing/2.4routing.sgml This may be the SGML source - Dancer says it seems to be LinuxDoc sources. But it's got the /data/ Jenn V. -- "Do you ever wonder if there's a whole section of geek culture you miss out on by being a geek?" - Dancer. [EMAIL PROTECTED]Jenn Vesperman http://www.simegen.com/~jenn/ ___ techtalk mailing list [EMAIL PROTECTED] http://www.linux.org.uk/mailman/listinfo/techtalk
