Linux gurus unite!

I guess our first task is to set up a suitable topic for discussion...
What about the future of Linux? What lies ahead? Will MS keep its actual campaign of FUD? Will it increase? Decrease? World domination, at last? Linux the desktop choice, for once? Drop yer 2 cents...

I have used Red Hat for a few years, but stopped after they raised the price. I have used Mandrake and was very impressed. I experienced problems with the early Debian, and thus never really pursued it. My favourite is SuSE though, because of the price, the vast number packages it includes also because it is fairly customizable and finally because of the easy installation. The SuSE update is also quite impressive, if you can get a stable connection.

I am running dns servers, dhcp servers, firewalls (iptables), samba pdcs and servers, squid and squidguard.

My recent successes includes swapping out several W2k pdcs and putting in place Linux samba servers.

Linux is the way to go, cheaper, functional and pretty good support from the community.

I've also found that Linux people are much more willing to share their knowledge than Windows people.

Just ordered Red Hat 9 - if it is an improvement on Red Hat 8.0 then it will be WOW!.

I have been using Linux (Red Hat 5.2-> and Mandrake 7->) for a number of years. I have looked at other distros like Caldera Open Linux, Knoppix (which is great for teaching introductions to Linux on most systems), Debian (stable but servers only realistically), etc and it is about time TR started supporting other than MS.

continued
We have in that same period decided to retire Novell, and replace it with M$ Win2K Office and network. We evaluated Linux on the desktop at the time, but it was just emerging with suites like Star Office, the numerous GUI's were too user unfriendly, and document interoperability was not smooth enough.

Well our parent company is based in Germany. They are heavily invested in Novell and NT4 on the desktop, and have a hundreds of seats of UNIX based CAD seats for product design, and SGI UNIX for analysis. Two months ago, I was informed, that they are evaluating Linux on the desktop. Since then I have begun to explore the desktop again. There seem to be a number of commercial distributions, that are aimed at a M$ crossover strategy to attract large concerns to migrate to some flavor of Linux.

Xandros (debian based) + Windows emulation for core office apps+ Enhanced Windows-like Desktop + functional office suite
Suse Windows emulation for core office apps+ Enhanced Desktop + functional office suite + improved admin utilities
Lindows (debian based) + functional office suite

Maybe TR would like long messages to become articles?

On a more serious tack, Linux (or, as Richard Stallman prefers, GNU/Linux) isn't the only choice for OS. There's the BSDs -- I'm on FreeBSD myself. There might be some Linux-specific software that won't run even under FreeBSD's Linux emulation -- e.g., Borland's Kylix -- but most commonly used software do.

Regards.

4.) The fragmentation frightens people. Do we go with Debian or Red Hat? Gnome or KDE? etc...

5.) I haven't thus far been able to make the networking side of it work as nicely. Novell Logon script automagicaly creates drive letters that point to the users personal, program, and Global data directories. M$ AD let's them browse for a printer, which then automagically installs the correct drivers....

6.) Even some of our web app - based systems don't play well with linux. example -We have this cool voicemail system, you can do everything from a web browser. Doesn't work under linux. Needs activeX bits, I guess...

7.) Hardware support can be a challenge. I have a wireless card in my laptop - 3com w/ XJack antenna. Can't get it to workunder Linux. This is getting better thgough. Hopefully, with HP, IBM, etc... this will continue to improve.

Making the hard drive bootable is usually not
that important/worth the trouble. Older machines'
BIOSes may be so broken that you will spend hours
trying to line up what they do for LILO with
what the kernel sees. It's an arduous
integrationstep the motherboard vendor has to
do (under NDA) with Microsoft, but not for you.

Instead, use a boot floppy until everything else
is working, including your backup/recovery
strategy. If your system boots from CDs,
burn your boot floppy into an El Torito CD.
Or make a LILO floppy with boot.b and map on
the floppy, and kernel on the hard drive.

Red Hat's oversimplified install doesn't know how to
handle your problem. Force it.

1. Let the BIOS do a "low level format" (zero fill,
actually) so nobody gets confused by anybody's
leftover boot sector. Write down the
"geometry" numbers the BIOS detects for the drive.
Notice whether BIOS thinks you are lba32 or "normal"
(that is, "CHS"), or "large".

2. When you boot Red Hat's intall disk,
stop at the boot prompt and read how to pass
parameters to their boot kernel. Force their
kernel to see your disk the same way your BIOS
does. That's the usual problem: BIOS sees one
thing, but kernel sees something bigger and better
and wants to use it all. You probably have to
type something at the boot prompt like
linux ide=255,16,12345
instead of just hitting enter.

3. Before partitioning, get a shell and go
cat /proc/ide/ide0/hda/geometry
and it should match.

4. Make the first partition 10 MB.
Mount it as /boot. Don't let Red Hat put
one giant partition on the drive. /boot is
all BIOS needs to be able to see.

Cameron

Forget Red Hat's "make it bootable" which obviously
doesn't work with old BIOSes.

Partition your drive with whatever tool you have.
A Windoze-98 rescue disk with FDISK will work.
Knoppix will probably work. Table as follows:

1. 10 MB not to be used by Linux, at the beginning
of the drive. Give it the FAT-16 or FAT-32
fs type.
2. 100 MB root.
3. 250 MB swap.
4. The rest, mount as /usr.

Install RH, get everything working, booting from
CD or floppy. (While /var and /homeand /opt are still
small, move them to /usr and point to them with
symlinks from root. Why doesn't Red Hat do that for
you? Do that in single user mode, of course.)

Get H. Peter Anvin's excellent "syslinux" program.
Put a FAT file system onpartition 1,
copy a kernel file there, and install syslinux
according to the instructions.

BIOSes know how to boot DOS. There is no shame
in letting them do that.

Cameron

If it is, go to linuxnewbie.org and look at the post there.

While I have never used syslinux to boot up, i've always used lilo or grub. The problem here is not your boot loader, it is the drive geometry.

The poster at linuxnewbie hit the nailon the head. Follow his advice, he gave you a suse link, but it will work for red hat.

Also, I know the poster here meant well. But his partition sizes are way off.

in Gui mode
100 megs /boot
1636 / (txt mode can be 612)
1024-1636 /usr (i use 700 in txt mode)
320 /swap (amt of ram * 2 min, you will need more if you plan on upgrading, so I would go with with either 512 or a gig of swap)

I personally always keep home and var on sep parts
612-1024 /var
100-1024 /home (or if in gui, fill up home)

Alot of it will depend on what you are planning on installing. For everything, I beleive you need an insane amount for / like 2 gigs.

These are just personal numbers that I use. Consult the install doc for RH's reccomendations.

BTW, the "green screen" does include bouncing-bar menus, so it's not like you're installing thru a TTY select-by-number interface. For that, look at OpenBSD 2.x -- dunno what the 3.x install looks like altho I wouldn't be surprised if it hasn't changed much.

Regards

To me, the biggest problem with FreeBSD is configuring X,
something most GNU/Linux prepackaged systems now do
for you. If you know how to confugure X, or you don't use
X, then FreeBSD is actually quite easy to install. It's also
quite optimized, though building a really complete system
can take a while, unless you have both a fast network and
a fast system, capable of quickly downloading source
code, then configuring, building, and installing it. The end
result is hard to beat, but it can be daunting to get there;
well worth it, though, especially if you want a real UNIX
system with current software.

I've been working in the IT industry for far too long as when I started there where no PC's only Mainframes with dumb workstations and these all ran Unix.

Now I work for a bank and all their gateway boxes are running Linux with MS stuff deper into the network or "Forest" as MS like to call it. Everywhere that I've ever worked theres always been at least one Linux/Unix box in the place doing the things that the MS stuff couldn't do as well and they are getting more prodomoniate nowdays as we are using more and more of the things.

Lets be real here for a moment Windows was only a development of Unix as used by Xerox in the Lidia project and honestly even to this day we have not as yet reached the level of sofiscation that Xerox had way back when they decided that there would never be a PC market and decided to dump the whole project but not before inviting all the outside developers in to have a look see {read that as Stev Jobs and Bill Gates}.

This is where the idea of a GUI interface came from and although I never saw the Xerox Lidia items I have read about their specs and they where the greatest leap forward in Unix coding that has ever occurred. We have been playing catchup ever since.

Unix has been around alot longer then windows has.
And is more stable, and has had more testing then windows can ever hope to have.

You are only looking at *nix as a desktop operating system with a gui like winbloze.

All because M$ dominates the desktop market, does not make it better. Don't get me wrong, I love 2K. In fact it is my OS of choice for building out networks.

But there is very little that Windows can do that Linux can't.

The only reason I don't put Linux on every users desktop is for the headaches that will ensue from changing what the end users know and are comfortable with.

But if the company I worked for was Pure IT, then you bet your ass that *nix would be on every desktop.

There is a reason why *nix powers the net.

I have been using DansGuardian in the office to stay familiar with it and also at a distant non-profit hospital site. I am quite tickled with the product.
Daniel has certainly done some great work with this!

Thre is, of course, a long wish list (that he no doubt has collected from users) but nothing that deters me from using it proudly.

The hospital site is used in a firewall config, using squid and squirm redirectors for the dmz servers.

The administrator uses Webmin to maintain blocklists and watch log files. A goal is to have it logged into a local mysqlserver. They report very easy usage, very effective filtering and easy list maintenance/updates.
A big jump in knowledge to start using it effectively, but it is required knowledge for an admin.

Overall, using ver 2.4.5, I'm impressed.
Well recommended.

Fundamentally cars are all the same, but you don't drive a * model car

PC's are fundamentally all the same, but you don't run a * type PC (what would DELL, IBM, ... say about that?)

So why them is *nix valid?

Windows ****, so by your argument any version of windows I choose can be used in context. This stands up to any version now or in the future.

If you really want to learn Linux, give Slackware 9.0 a try... www.slackware.org

You can check out JetBlackz Simply Linux Installation Guide here http://jetblackz.freeservers.com/

The Slackware Installation and Configuration guide there will be big help.

You can also lurk around the Distribution forum at www.linuxquestions.org and read all the great posts comparing distributions.

They have a Slackware forum that is most helpful too.

http://www.linuxquestions.org/questions/forumdisplay.php?s=&forumid=14

I read your bit about OO 1.1 beta. I love this software suite and recommend it to everyone.

I saw your disc. about pdf export and thought I
needed to give this a try. So I did. Unfortunately,
the pdf produced from OO's word processor could only
be read in the linux version of Adobe Acrobat. It could not be opened in another pdf viewer I have on my linux machine. It could not be viewed in the latest Windows version of Acrobat. I tried 2 separate files with the same results.

This is a disappointment because my campus was looking at a campus wide site license to the full version of Adobe - the one that let's you create and edit pdf documents. We couldn't afford the Adobe license. I was going to send the notice to the campus listserv about OO's ability to create pdf files but now I can't. This would have been a major "selling" point on my campus for OO if the Adobe export function worked as advertised. Bummer.

Due to the ease with which I could learn to use
Anjuta, I have been able to cut down actual
development time. Also, being new to makefiles,
Anjuta took away all the hassles of learning to
maintain makefiles and the like. With its
integration with Glade, there is no better IDE
to use on Linux.

There are at least three possible solutions:

1) VMware. This creates virtual machines. Each virtual machine can run a different virtual operating system.

2) WINE - this is a windows emulator. The version that ships with RH8.0 is an imperfect emulator in that some software won't work with it. I am told that the version in RH9.0 is better.

3) Lindows - this is a set of libraries that provide the same functionality as real windows. I am not sure I understand how WINE and Lindows are fundementally different.

If the applications are written in visual basic, there are tools that will translate the VB into Java.

So it can be done.

Great beta, fast and smooth. Gets me to where I want to be. Only issue currently is css and javascript support, but I've just added them to the "Features" wish list on their web site. It's nice to see a developer encouraging all users to offer suggestions and recomendations. You can even offer your own fixes, plugins and extensions, ideal for the home developer. A real Maclaren F1 of browsers.

You can even upset the natives with some suggestions.

I've been using Phoenix occasionally since 0.4, a bit more when 0.5 came out, but started using it with interest when 0.6 came out and it became MozillaFirebird.

This browser has most of the features of the Mozilla browser, many of the features of the Galeon browser, and seems to have the speed and quickness of the best of either of them.

When I heard that Firebird would be a standalone browser and Thunderbird would be a standalone Email client, both based on Mozilla, I took interest. This should take some of the bloat out, yet keep the great features in. So far, it's already looking very good, even in pre 1.0 form!

And I suppose that having Microsoft pay them large chunks of money has absolutely nothing to do with this huge u-turn from SCO?

http://www.businessweek.com/technology/cnet/stories/1007528.htm

(I got that link from SCOs homepage at www.sco.com)

We're still using SNF at the moment, since it works fine, but the ext2 filesystem is a lot to be desired on this one.

I've tried out MNF once, but I didn't read that much of the documentation so I only got to have DHCP working and not even NAT orSquid.

I think it's really good. I have plans to upgrade from SNF since having an updated kernel alone is reason enough to do so.

Thanks for pointing out the flaw. You may not like it, but you just proved a point.

Were it a bug in a closed source OS, I may well only find out after it got used and abused and then have to wait on a patch from MS.

As it is, I've just been warned by someone without an interest on keeping things hushed up (it appears), and I have the opportunity to do something about it.

So, I have the option of finding another vendor with a non vulnerable product (although yes, that would be extremely inefficient to do), or I can go and find more information on the flaw and the possibilities of DoS attacks with it, and maybe find a way to stop it before it happens until I get a patch from my vendor.

That is exactly why people are supporting open source. I'm not even an active sysadmin, and I'm not actively keeping track of OS issues, yet I find out about a flaw which is still only "reportedly" affecting some versions of Linux.

Thankyou, and keep up the good work.

Spuddy_M, your post shows how little you know about the workings of open source.
Most likely that flaw you are referring to will dissapear in a copule of months at most, since that's how fast the Linux community responds. When something like that happens to Windows (which is quite frequently) users have to wait for longer time for M$ to release a patch (which incidentally may introduce some other security flaws. This has happened a lot for M$).
Please get this: Linux IS stable, Windows is not.