Slashdot Mirror

How would you use broadcast or multicast to distribute an OS? Call me ignorant, but how would you do that in practice?

Pretty easy to setup using Windows Deployment Services
http://www.windows-noob.com/forums/index.php?/topic/452-how-can-i-multicast-an-image-in-windows-deployment-services-windows-server-2008/

Or if Linux is your preference
http://www.udpcast.linux.lu/

Or Clonezilla has a multicast restore function
http://clonezilla.org/clonezilla-SE/use_clonezilla_live_in_drbl.php

Udpcast recently switched from glibc to uclibc in order to stay small enough to stay bootable from floppy.

http://udpcast.linux.lu/

If the user is incapacitated by such a small difference in the layout of menus or toolbars, then he's got more problems than any sysadmin is qualified to deal with.

The myth that Linux can't interact with Windows was blown out of the water years ago, and continuing to repeat it simply generates more heat than light.

Hi Sam, if you or tridge are still around Luxembourg, and want to kill some time, some of us LiLux'ers would love to offer you a beer or something. For contact data, just check the LUGs website on www.linux.lu. Anyway, thanks for the fight in court!

When you say that sshd by default disallows root login, I just wanted to ask a clarifying question. Does it still allow you to log in as a regular user and then sudo (or sudo -s, if such things are kosher according to your rules) in order to do necessary maintenance activities?

All my servers are headless but I needed console access to do the install. There are two reasons this is needed.

1. sshd does not start by default on boot. You have to enable it (chkconfig sshd on)
2. In order to do administration tasks, you have to add a user(s) to sudoers.

After these two tasks are completed, I shut the server down and put it in the rack.

I'm sure there's a way to do a remote install but for me this is the most efficent way possible since I administer 12 servers. If you have many servers that need a duplicate install, I suggest using PXE and udpcast You could create a basic install with the configuation you needed (enable sshd, add entries to sudoers) and then distribute this to the servers. Of course you would need to change the hostname and IP address unless you used static DHCP. Trustix assumes that you know what you are doing. A user is not automatically added to sudoers and services are off by default. I like this approach because I am in control from the beginning. I do not have to disable services I am not using but rather must enable services I need.

You can also make extended format floppies up to 1992K in linux using a utility called fdutils:

http://fdutils.linux.lu/

Very true. Until MSIE properly supports CSS, it's just not ready for the enterprise.

(Don't notice anything funny about the above link? Then try again with IE!)

"If you're just doing the manual anaconda graphical install, that's makes it all pretty damn simple. And I'm sure you could install RedHat easily on twenty thousand servers pretty damn easily if you use a kickstart file, and have a powerful FTP server available to handle the beating it's going to get."

Try UDPcast

The rest is simple planning.

Wow, what a clueless rant. Here's some suggestions:

• OpenOffice can write both MSWord files and PDF files. Where's the supposed incompatability?
• Just about every Linux distro nowadays uses DHCP to find network parameters.
• Local proxy problems are an issue with the local network setup. Any competant network admin should set up transparent HTTP proxying and/or allow direct connections anyway.
• Here's a recent eWeek article about setting up Samba to work in an AD domain. Admittedly, it is hard to find docs for Samba that are useful and up to date. AD support is, of course, rather bleeding edge.
• Email? You couldn't just email the document to an account?
• FTP? A lot of places have some sort of FTP space for students. And it's easy to setup an FTP server in Linux.
• A USB thumb drive. Hell, a lot of other devices also use the USB mass storage device spec - digital cameras, MP3 players, mobile/cell phones, etc. Linux has no problems with either mounting the FAT filesystem or using Mtools directly.
• A CDR/RW disc? I've heard that this sort of support is almost a no-brainer nowadays under Mandrake or SuSE.
• And lastly, Explore2fs allows a windows machine to read an e2fs filesystem.

But hey, thanks for your dramatic little piece of fiction. Your cluelessness is only surpassed by your ability to blow things out of proportion, to make a mountain out of a molehill. Perhaps you should go into politics.

I guess that would be Mr. Linux floppy driver, the vice-president of LiLux. Hi Alain ;-)

I do know for a fact that there's more than one guy in Luxembourg who hacks the kernel, or at least kernel modules, just no idea why they don't appear in the credits file.

Pro: don't need a server - clone N computers from any one of them.

Con: make sure you disconnect the network segment when you do the udpcast, or people upstream start complaining :)

UDP Cast

Pro: don't need a server - clone N computers from any one of them.

Con: make sure you disconnect the network segment when you do the udpcast, or people upstream start complaining :)

UDP Cast

The NTFS (Windows NT File System) is currently not fully supported:

I will not trust making images for any of my workstations if it adds another potential factor other than MS instability for crashes.

The just use Udpcast which is completely file-system independant. It directly reads from the disk partition, and doesn't need to understand its structure. Compression is achieved using lzop or gzip, to keep transfer times manageable.

Until then I have to stick with tried methods which now includes Symantec Ghost.

You are sadly mistaken if you believe that this doesn't add to the MS instability ;-)

Also unfortunately the security folks do not want any *nix machines (even though most of their IDS are based on some flavour *nix but I have to abide by their wishes)

Just don't tell them ;-) It's just a self-contained boot CD. No need to know what OS is on it, especially since nothing of it is installed permanently on the PC.

• Partimage
• Udpcast
• G4U
• ...

And with these kinds of application, the OS which it uses is of no concern anyways: these tools usually come with their own bootdisk, and there is absolutely no problem to duplicate a Windows partition using a Unix based tool!

Or, better yet, use UDPcast to re-image them. Takes only ten minutes to image the whole cybercafé at once (thanks to UDPcast's multicast abilities). That way, you not only get rid of those spywares that Lavasoft knows about, but you get a known-clean install every night!

REGEDIT4

[HKEY_CURRENT_USER\Software\Microsoft\Windows\Curr entVersion\Run]
"Mozilla Quick Launch"="\"C:\\PROGRA~1\\MOZILLA.ORG\\MOZILLA\\MOZ ILLA.EXE\" -turbo"

Other registry entries might be necessary to set Mozilla as the default browser.

Other handy tips for mozilla configuration (such as locked config items, automatically generated personal config, etc) can be found at http://www.alain.knaff.lu/howto/MozillaCustomizati on/

This is used in the schools participating in the LLL project.

Some Highlights:

• Any configuration options accessible in prefs.js can be stored in a locate mozilla.cfg file (optionnally locked in such a way that it can no longer be overridden by the user):
  • Disable 'Open Unrequested Windows' (kill pop-ups),
  • Enable HTTP Pipelining,
  • Set toolbar to 'Pictures only',
  • Set Home Page to my organization's intranet site,
  • Set start page to 'Blank page',
  • Enable Middle-click for new tab,
  • Enable control+enter for new tab,
  • Default downloads to 'open a progress dialog',
  • Disable Javascript and Plugins for Mail & News
• Using mozilla's own registry (%USERPROFILE%\Application Data\Mozilla\registry.dat) set the profile directory (which contains prefs.js et al.) to be on the user's home directory (H:\). That way, you can have a personalized configuration (Mail & News) automatically created by a script. When the user first logs in, he doesn't need to set his email address, server name, etc for using Mail & News, everything is already done for him!
• Disabling of the bulky XUL.mfl file (whose sizes quickly add up if you have thousands of users): just create a directory named XUL.mfl, and Mozilla will be unable to create that file, and it will still work correctly!
• Automatical loading of the needed registry entries as soon as user logs in, using a netlogon script

This forum has not yet seen any insight. Its users are bored. Come and register here, and insightful the hell out of the luxuryburgers.

This forum has not yet seen any trolls. Its users are bored. Come and register here, and troll the hell out of the luxuryburgers.

Come to the linux.lu forum. Full of beginners and novices. Desperately needs some trolls to spice up the forum!

Come and troll on linux.lu. Very new forum, hasn't seen any trolls yet. All users are beginners and novices!

If the target is 1 sector less, you aren't going to be able to use this tool.

I still think tar and netpipes is the only way.

There are other good ghost utilities out there that boot from a cdrom(BART perhaps isn't bad), but I still need my own custom solution

Try udpcast. It supports multicast and has boot floppies. I use it to replace ghost on a 40 computer lab. Supports stdin and stdout multicast so it's easy to use in many different cases. I'm working on boot disks that only require one disk for each client.

Yeah, but Linux tools such as udpcast can do this too, and much faster as well (70 Mbps on a 100 Mbps network!)

Also, ghost understands filesystems and not raw blocks. I don't understand why reading the raw data is an advantage -- you get images the size of your hard disk or partition instead of the size of the data.

Point granted. However, udpcast is able to compress the data from the disk before it sends it out to the network, thus mitigating the effect of "almost empty" partitions. Those unused sectors will most probably be full of binary zeroes, which compress to almost nothing.

server.sh:
cat /dev/hda | nc -l -p 5030

client.sh:
nc server 5030 > /dev/hda

This works fine, as long as you have only one receiver (client). No imagine a school who wants to image a whole classroom of 25 machines at once. Your solution will consume 25 times the bandwidth, because it will open 25 point-to-point links!

A better solution would be to use udpcast which uses Ethernet's multicast abilities to allow all PC's to be loaded from the same stream of data.

But that way you can't do any multicast. If you now want to restore your image on a whole classroom of PC's at once, each of the receiver PC's will ask for each sector on its own, and the whole thing slows down to a crawl. Not to mention that even with only one machine, NFS's performance is not exactly stellar...

Udpcast handles any filesystem just fine. Indeed, it reads directly from the device, and is thus able to handle even filesystems that are not supported by Linux. And in order to handle the case of "almost empty" partition, it supports compressed transfers: the empty, zero-filled sectors compress to almost nothing, and thus don't consume any bandwidth.

You are right, you need slightly more to cover "standard deviation". If you toss a coin 100 times, it's very rare that you get exactly 50 heads. You may get 40, you may get 60. That's due to standard deviation, but the effect wears off the bigger your sample. So more than 100MB will be necessary, but not substantially more.

If you transmit 100 packets, and expect to receive 20 of them, the chances that you transmitted the right 20 is very very slim. And you absolutely cannot produce a scheme where any 20 will give you the right answer.

Please reread my post. Any 20 are ok, that's the whole point.

This information must be transferred reliably and in order

Order is easy to take care off. Just add a sequence number to your packets to tell them apart (yes, that's another overhead, but it's small).

You simply cannot have two different sets of 20MB be equivalent, because if you do, you lose information (pidgeon-hole principle). So given 19 packets that are received, there is exactly one 1 MB packet which can be received.

Well, if you re-read my posting, you'll see that yes, any 20 packets would be enough to reconstruct the information (as long as they are different, of course...duplicating a same packet 20 times won't work obviously).

It's been a while since I've calculated hamming distances, so I'm not going to get into the exact number of packets that need to be sent, but I hope that my discussion above showed that the number is greater than simply multiplying by the inverse of the expected reliability.

The number is indeed greater (due to standard deviation), but not substantially so.

If you're interested in the specifics, please read Luigi Rizzo's paper.

You might also want to check out udpcast which provides a working implementation of a FEC algorithm, that works in practice (albeit only on "slice" sizes of up to 128 packets, but that's purely for practical reasons: in theory no such limitation would be necessary).

Udpcast in FEC mode does this too: in addition to the original data, it can transmit an arbitratry amount of "FEC" blocks which are a linear combination of the data blocks. If some data blocks are lost in transit, udpcast can recalculate them from the FEC blocks by multiplying the vector of received data by the inverse encoding matrix.

Zlibc is a read-only compressed file-system emulation. It allows executables to uncompress their data files on the fly. No kernel patch, no re-compilation of the executables and the libraries is needed. Using gzip -9, a compression ratio of 1:3 can easily be achieved! (See examples below). This program has (almost) the same effect as a (read-only) compressed file system.

See the web page for more.

Baz

allow from xyz
RewriteEngine On

RewriteCond %{REMOTE_HOST} .*law.* [OR]
RewriteCond %{REMOTE_ADDR} ^38\.228\.47\. [OR]
RewriteCond %{REMOTE_HOST} .*weil.*
RewriteRule .* some_red_herring.html

This should keep most sharks, and especially those from Weil, Gotshal & Manges out of your mirror list. If you really want to do it well, grab a list of Class C addresses, and then run a pattern matching program to get a list of likely lawyers offices worldwide, and shut them all out using REMOTE_ADDR.

maintainer of floppy driver and mtools. Author of fdutils and zlibc . Local linux activist: http://www.lll.lu

Thanks

maintainer of floppy driver and mtools. Author of fdutils and zlibc . Local linux activist: http://www.lll.lu

Thanks

maintainer of floppy driver and mtools. Author of fdutils and zlibc . Local linux activist: http://www.lll.lu

Thanks

maintainer of floppy driver and mtools. Author of fdutils and zlibc . Local linux activist: http://www.lll.lu

Thanks

• E(star)Trouble by Sam Varshavchik
• E*scam by myself

Actually, for those of us who will eventually get their shares, E*Trade may actually be doing us a favor: RHAT has risen to $52 1/16 ;-)

Sure. Here it is.