Kernel 2.2 - It Lives!

Build6 writes "For those of us still using 2.2 (how's that for "conservatism" eh?) -- 2.2.24 is out (and has been since last week) - see kernel.org for downloads. I see networking code tweaks, but no changelog. Time to give our old RH 6.2 machines one last kernel-recompile before Red Hat's end-of-life date arrives for 6.2? :-) What I'd like to know is - who else (besides me) out there still has machines running 2.2 and intends to keep it that way?"

Comments

Kernel Series 2.2

[Scoria]

Although Linux 2.2 may eventually become relatively obscure, I wouldn't anticipate its disappearance. It will almost certainly remain a viable contender for certain embedded and esoteric applications.

Re:Kernel Series 2.2

[Anonymous Coward]

Is it stable enough for prime-time yet? I'm not "upgrading" my 2.0 boxes until I'm sure it is.

Did anyone ever fix the uptime rollover thing? That's what I'm really waiting for. I'm sick of rebooting every year and a half.

Re:Kernel Series 2.2

[Micah]

Huh? 2.0 is ready for prime time? Why didn't someone tell me? I'm still using 1.2!

Re:Kernel Series 2.2

[Scoria]

What have I been eating, then?

Re:Kernel Series 2.2

[0x0d0a]

This newfangled Linux thing seems a bit risky. Minix is definitely the way to go.

Re:Kernel Series 2.2

[rmadmin]

I'm still running a 2.0 kernel on a production Slackware 2.3 machine. The reason for that is that I'm afriad of breaking the antique software package on the system by upgrading, but it seems I will have to do something soon since OpenSSH will no longer compile on the machine, and I don't feel like leaving it open to the world.

Re:Kernel Series 2.2

[dfries]

I'm running 2.4.18 on my Telephone so I'm not so sure how long 2.2 will last in the embedded market either.

Re:Kernel Series 2.2

[Anonymous Coward]

I ran a 2.4 kernel on a box that was largely unchanged from its original Slackware 3.2 install from November 1997. There were a few minor changes along the way to keep up with the kernel, but nothing massive.

Last year, I upgraded it in place to Slackware 8.1 by methodically installing new packages. Those old versions didn't have upgradepkg, so you have to install a few things (like pkgtools) before you do anything else.

After finishing with that and running the new version of LILO to set up the boot environment, I rebooted. It came back up, and life goes on with a modern box that isn't limited by libc5 weirdness.

You can do it. Just be slow and methodical and remember to install the new base before removing anything from the old one. That means installing all of the new libs (glibc, etc) before even thinking about touching the old packages.

By the way, I did all of this in multi-user mode since I was coming in via ssh. It worked fine.

Re:Kernel Series 2.2

[gujo-odori]

2.2 is still quite viable for use on production servers. At my former employer, the mail system (4 MXes, 2 outbound SMTP boxes, 2 POP proxies, and 2 backend mail spools) still runs 2.2 because it has been utterly stable (2.4 is pretty stable now too, but for a long time 2.4 was held to be too volatile for production use by many people, and more than a few still think so).

The default kernel remains 2.2 in Debian Stable, as well. On my personal machine I run 2.4, but if I were installing to a server that had no need for USB, etc., I would think about using 2.2 even now.

Re:Kernel Series 2.2

[edbarrett]

LEAF Bering. It rocks. seriously. Shoreline firewall config, Free S/WAN support, and more!

I'm not trying to knock you, I'm just plugging a cool product (although I'm just a user, myself).

Re:Kernel Series 2.2

[Jeremiah Cornelius]

SLACKWARE distros have NO package management tools!!!

Slackware has two. tar and gzip.

Why 2.2?

[York the Mysterious]

This is a real question not flame bait. Why would you keep 2.2? What is there in 2.4 that makes it so bad? It seems like it's pretty mature now so what's wrong with it?

-Tim

Re:Why 2.2?

[Osty]

This is a real question not flame bait. Why would you keep 2.2? What is there in 2.4 that makes it so bad? It seems like it's pretty mature now so what's wrong with it?

Better yet, why would you upgrade if 2.2 does everything you need? Any security patches will be back-ported, and that's the only time you really need to upgrade your kernel so long as it does everything you need already. For example, my bridge/firewall machine (P200MMX) is running a 2.2 kernel, and with the 2.4 bridging code backport, it works perfectly fine. I have absolutely no desire to spend a day with my firewall machine down while I upgrade all of the kernel dependencies, configure and build a 2.4 kernel, rewrite my firewall scripts for iptables (yeah, I know you can use the old ipchains interface with 2.4, but if you're going to do the upgrade, do the upgrade), and then work out all the gremlins from running "new" code.

Desktops are different, because nobody cares if you have downtime with them. Servers on the other hand can cause pain when they're down, and even for a personal server the downtime is not worth the upgrade. You gain nothing, and lose quite a bit of time.

Eventually, I'll decomission the P200 and bring the celeryonion 433 up as my bridge/firewall machine, but that's going to be timed with a move (when the machines will have to be down anyway, and it'll be a week or two before the new place has internet access). Doing it before then is pointless.

Re:Why 2.2?

[Anonymous Coward]

If I had to bet my life on it I'd still use 2.2 over 2.4. To say the 2.4 series has had some rough spots is an understatment. 2.4.0 through something like 2.4.5 was less then impressive and at 2.4.10 Linux replaced the VM. It hasn't exactly been smooth sailing. I suppose by 2.4.24 I'll have that same warm feeling I do for 2.2 but so far that hasn't happened.

Re:Why 2.2?

[the eric conspiracy]

Why would you keep 2.2?

'cause upgrading a server running a bunch of stuff would take a long time to test.

Re:Why 2.2?

[oconnorcjo]

This is a real question not flame bait. Why would you keep 2.2? What is there in 2.4 that makes it so bad? It seems like it's pretty mature now so what's wrong with it?

Another good question is why upgrade?

2.2 is a good system and has served many of us well. Linux 2.4 really only became stable and available in 2002. For people who don't live on the bleeding edge (and there is a reason it is called "bleeding"), 2.4 is just starting to be deployed. When a server runs well, upgrading it is often just a pain. I am guessing that 2.6 will come out in 2004, be stable and available in 2005 and many servers will only start migrating to it in 2006 and on slashdot some simmiliar story will be posted and somebody (maybe you) will post a simmiliar comment/question.

Re:Why 2.2?

[Tony-A]

Box du jour syndrome. Due to peculiarities in the particular hardware and bios and in the particular detection of that hardware, things work or not with no easily discernable rhyme or reason. At one point I had two versions of RedHat and two computers. One version would install on one computer and not the other. The other version would install on the other computer but not the first.
One advantage of old versions is that you can get something usable running without much effort or knowledge.

Re:Why 2.2?

[oconnorcjo]

Theory has it that 2.6 will come out around the end of June 2003. Well, this is theory, of course, but there aren't really many show-stoppers left in 2.5 nowadays. I think "by the end of the year" might be a reasonable release expectation for 2.6.0 barring major problems cropping up in the next couple of months. If this is the case, I bet it would be stable enough to use in production by mid-2004.

Developers of software always think they are close but usually software gets delayed much longer than expected. "Linus and gang" wanted the 2.5 series to be done in something like 9 months. When I was reading that, I thought the idea was rediculous. 2.5 was started 11/2001 and it is now 3/2003. A two year development cycle is really as fast as can be expected in such a sophisticated system (if fundamental changes are made) and a three year cycle (like the time given to 2.4) is probably even better. Nobody really needs a new kernel sooner (or they are using the wrong tool for the job) and the time could be used to make the new system better. Distibutions, developers and users should spend more time USING an OS than UPGRADING it. I would be shocked if Linus was actually able to release 2.6 this year and servers should not start deploying it for another year after that.

Several production servers

[benevold]

Although we are slowly phasing them out we are running quite a few machines, mostly redhat. Because of they are critical systems they have yet to be upgraded. There's one debian box that's been up for over 500 days and the kernel wasn't upgraded for a while before that! For some reason it seems to be the most stable box we have.

Re:Several production servers

[gilesjuk]

Debian use tried and tested software, their kernel sources contain quite a few bugfixes too.

Running the vanilla sources from www.kernel.org means you sometimes miss out on some bugfixes, unless you follow kernel development.

Why?

[este]

I get asked all the time.

I've still got 2.2 on my laptop, and really, I'm happy. I don't use it for much more than mobile internet access, and as tightly compiled as I have it, I don't feel a need to go through and upgrade. Just that much more work for an overall unimportant change, a least in this situation.

Of course, my desktop has 2.4. :-)

PS2 Linux users

[Vardamir]

All of us are still using 2.2 kernels, whether we like it or not.

Re:PS2 Linux users

[Mattsson]

Take a look at http://blackrhino.xrhino.com/
A debian-like 2.4 distribution for ps2. Requires the ps2linux kit though...

Fancy new software for my old POS?

[Quasar1999]

2.2? I'm still running 2.1... Works great (as a firewall)... then again, it's only a 486 DX2 66... so I don't dare fiddle with it... It might turn to dust from old age...

Re:Fancy new software for my old POS?

[FyRE666]

Bah! A 486 can easily handle a moderate (small office) amount of machines on a cable connection. We had one at a previous job (486dx33, 16MB RAM, no hard disk) that MASQ'ed/firewalled for 15 machines - it didn't break a sweat. I have an old P90/32MB on my DSL line (internal ADSL modem) with extensive firewall rules, acts as the smart mail relay and it also runs apache when the main server is down (minimal config - no CGI/PHP). It's never anything close to busy. Mind you, I think it's close to death now - it normally takes 3 attempts to boot, and I can hear the fan if I listen closely even though its up in my attic! I'll really have to back up the drive one day - took me about a week to get it working just right - that ASDL card was a real bitch...

I know people who still have routers

[leviramsey]

...that run 2.0... And of course, Debian stable is still 2.2.

Simple

[Anonymous Coward]

It works.

I ran a firewall off of my 2.2.23 box all set up to be secure to the outside and provide a fileserver/print server to the inside as well as being a DSL and dial-up router.

Why would I upgrade and possibly break something?

It does not need X, it is a PII-400, and it does not do anything that is so intensive it needs 2.4

Long live 2.2

Re:Simple

[error0x100]

I have a Pentium 166 (64 MB RAM, 2GB HD) running RedHat 6.1 w kernel 2.2 set up as a dial-up router, running squid as a web proxy (with sleezeball to block ads), plus a bit of file sharing and a remote X desktop with icewm over VNC. I am pondering loading RH 8.0 onto the machine, but quite frankly, the current system works, and it works well and fast.

Re:Simple

[AchilleTalon]

I still have my 486 with 16MB memory running a firewall/DSL router. It is now near 10 years old and still running. This machine is never powered-off, unless there is a black-out. The only problem I had was with the CPU fan I replaced many years ago and the power-supply fan I replaced about two years ago.

Linux Kernel 2.2.X has been continuously updated on this machine without a glitch.

This machine has seen pre-1.0 kernels and was my first PC. I just don't remember the very early Linux distributions I tested on this machine. For sure, Slackware was installed on it at some point in its life.

Re:Simple

[Eil]

It does not need X, it is a PII-400, and it does not do anything that is so intensive it needs 2.4

Err, I hope you aren't implying that 2.4 is either bloaty or slow or both on older hardware.

I used 2.4.18 just fine for over a year on a Pentium 166 (no MMX) and had absolutely zero problems. This box was my broadband firewall and also served 60 GB of NFS, as well as SMB, ssh, mail, and apache 2.x web pages, both static and generated. (I know you're not supposed to combine your firewall and other stuff, but I had no choice at the time.) Anyway, this box did its job(s) flawlessly without a single complaint and though building a kernel took on the order of 50 minutes, most things happened instantaneously.

I decided to upgrade it to a Celeron 366 only after I started using a python-based wiki on a daily basis for note-taking. If I really wanted to, I could have hacked up my own program in C that would have been 10x faster but I had the spare hardware and figured I might as well retire the 166. Given all of the improvements of the 2.4 kernel series, I highly doubt that 2.2 is significantly faster than 2.4 (for the same tasks) on all but the very oldest hardware.

The only places that I think would want 2.2 over 2.4 are organizations that have mission-critical stuff running on 2.2 and aren't keen to fix that which isn't broken (if you'll pardon the cliche). Other than that, using 2.4 for most tasks is simply NOT going to cause armageddon. And also remember too that just because some piece of software is OLD doesn't automatically mean it's more STABLE.

Re:Simple

[pheared]

If you run a firewall, all the more reason to upgrade to 2.4. Netfilter is far superior to ipchains, in my opinion.

Of course, you may not currently need stateful inspection, but you don't even have the option with 2.2. If you come to a point where you do, you're out of luck. (unless there is a current reliable backport out there, which is possible)

Re:Simple

[rastachops]

That also begs the question of, why would you spend $0 on an operating system for a $3000 server?

Re:Simple

[DrXym]

It certainly is superior, however from my own experience of moving up, you have to bite the bullet and rewrite your firewall script. While you could leave everything using ipchains, (2.4 has an ipchains module after all), distributions such as Red Hat don't like it very much and some of the network tools won't work when ipchains is running. IP Masquerading (NAT) and VPN also just works which is a compelling reason in itself. If you want VPN in 2.2 you have to arse around downloading kernel patches to enable it.

It's better to rewrite everything to use iptables, though this does require some effort since the syntax is not quite the same. The biggest hurdle is figuring out how to log and drop a packet. In ipchains it is one command, in iptables you must create a new chain that does both actions and redirect packets to that.

Gateway

[yoshi_mon]

My current gateway is a AST 486SX/33 with 16 megs of RAM.

I was able to install RH 6.2 on it and wittle the RPMs I didn't need to get it down to under 200 megs.

While on many of my other servers I run 2.4.x, on this type of box I think 2.2.x suits my needs perfectly.

I have a 2.2 machine...

[powerlinekid]

I have no idea what number it is but it runs my Mac Performa 6360 which is acting as a router. The 2.4 kernel panics anytime it tries to access the cd drive or I look at it funny. Don't get me wrong here, I use 2.4 on everything except in this one case.

Linux's new target market

[batobin]

I'd say with everything I'm reading about Linux's new target market, a lot of people will leave their kernels. Why? Because they either don't know how to upgrade, or more likely, are not educated on the benefits of upgrading.

Honestly, with the advent of Linux being sold at K-Mart, used in schools, and wielded by mouse-clicking Grandma's, there are bound to be lots of people who don't know they should upgrade their kernels. I personally think marketing Linux to these markets is important, but an equal amount of importance should be recognized in educating these new users in the basics of maintaining these systems.

Because what good is the open source movement if the end user doesn't know how to benefit from our work?

Re:Linux's new target market

[gilesjuk]

Probably why Linux isn't in the top 10 of uptimes on Netcraft, people bringing the system down to change a kernel or recompile a component due to a security hole.

It's quite true what you say Linux for the inexperienced home user, only it's not just kernels either. What about installing new applications? users are used to downloading a setup.exe and running it. With some distros you use a package management tool, others you compile the source code. This will be intollerable for many users, anyway this is deviating from the topic.

Re:Linux's new target market

[syrinx]

FreeBSD:

> cd /ports/whatever
> make all install clean

Just as easy as downloading and running setup.exe, wouldn't you say?

I hear Debian has something similar, though I haven't used it myself.

Not that I'm arguing with you, some people will find things like that "too hard", just because it's different. *shrug*

Re:Linux's new target market

[Alex]

Go on.... Why "should" people upgrade their kernel? What are "the benefits of upgrading"? (to the average K-Mart buying linux user),

Alex

Re:Linux's new target market

[Osty]

more likely, are not educated on the benefits of upgrading.

And those benefits would be ...? I was on the kernel upgrade treadmill for years, getting new versions as they were released, upgrading to 2.2 with the very first version, etc. It gained me very little. I reached a point where staying on top of kernel upgrades was more pain than it was worth, and stopped. Now I only upgrade if I need some new feature, better support for an existing feature, or for security reasons. Aside from security patches (which I would hope all of the "Windows Update"-ish tools would handle for these users already), 99.95% of all Linux users have no reason to upgrade their kernel, so long as they're using a sufficiently modern one to begin with. Why, then, is it such a bad thing that these users don't know how to upgrade their kernel? If anything, I'd say it's a testament to Linux that users running it can get by without ever having to touch the kernel (aside from maybe loading a module or two when they get new hardware, though even that could/should be automated).

Re:Linux's new target market

[Otter]

Not that there are vast numbers of Linux-using grandmothers in reality, anyway, but I'd propose that the overwhelming majority of desktop users, and probably a lot of server admins should be cautious in upgrading.

Distributions provide well-tested, patched kernels. Compulsively updating kernels is a fine hobby if it makes you happy, but unless there's a new feature you need, the potential for breaking something exceeds any practical benefit. The experience of the 2.4 series, where half the kernels substantially degraded performance because of some new half-assed VM only underscores that lesson.

No, if you don't know why you need to upgrade your kernel, you almost certainly don't.

Re:Linux's new target market

[stor]

> Just as easy as downloading and running setup.exe, wouldn't you say?

Non-geeks most certainly wouldn't.

When yer mum calls you regarding her shpanky new FreeBSD box and the fact that "some book... or library.. that's it.." needs upgrading are you going to tell her to fire up an xterm, cd into /ports/ and make all install clean. Then claim that's just as easy as double-clicking on a "setup" icon?

You've never done tech support right?

Cheers
Stor

Re:Linux's new target market

[coyote-san]

So do you know what's involved in upgrading your kernel?

This isn't a minor change, major kernel updates usually require updated tools. Hopefully the tools will update cleanly, but it's a non-trivial risk.

The same thing applies to upgrading distro versions. I'm hardly a naive user, yet I have been unsuccessful in every attempt to upgrade my Debian potato systems to woody. I've now given up - I just do a clean install of woody.

Re:Linux's new target market

[lewp]

I've felt the way you do for a long time. I've usually got the latest dev kernel on my laptop because of its requirement of a recent ACPI patch (it's one of those legacy-free Toshibas) to get sound and its NVIDIA video card (these two don't go together well), but for every other system I have I don't touch the kernel if I can get away with it.

In this day and age, when most common PC hardware, and certainly any PC hardware worth anything (ie. not the aforementioned laptop), is well-supported under Linux, why do people feel compelled to have the latest kernel? I compile all my kernels on my Athlon XP build machine and even when the compiles are pretty fast (compared to my old 486 box that I first used Linux on) it's not exactly *fun*.

I consider myself a pretty big geek, being thoroughly integrated into the Slashdot hive mind and all. I'm on top of CVS gaim, xine/totem, Mozilla, and GNOME (hoping it won't suck eventually) because they seem to become noticeably better by the day. But the kernel? Could someone please explain why?

Re:Linux's new target market

[cpeterso]

Maybe that new USB digital camera they bought is not supported by Linux 2.2? Improved hardware support is the only tangible reason (to a Joe Kmart user).

Re:Linux's new target market

[Blkdeath]

Go on.... Why "should" people upgrade their kernel? What are "the benefits of upgrading"? (to the average K-Mart buying linux user),

"People" should upgrade their installed software as their distribution vendor tests and certifies it as stable and secure (to the best of their knowledge) and released it to the "stable" branch. Else, they should upgrade as new functionality, drivers, driver bases, etc. is desired (GigE, ACPI/APM enhancements, etc.)

People who run Linux as a hobby "should" upgrade as they feel neccesary, but they'll probably have atleast one workstation on the bleeding edge anyways, so they're a moot point.

People who run servers, casually or professionally, "should" track updates and understand their impact on the remainder of their systems. Their professional, corporate, mission-critical, or otherwise important (generally in the context of "to paying customers", "management" or "other employees") "should" be updated on a semi-regular basis as neccesitated by (potentual/actual) stability or security issues, and after a sufficient testbed period and impact analysis.

In general, an upgrade as major as a kernel (major revision) should be taken with all due consideration. Mostly I'd reccomend a distribution version upgrade (ie; an all-encompassing upgrade procedure that will account for the C library and compiler, all system libraries, shells, and related utilities, as well as the userland).

If an "average K-Mart buying linux user"{sic} wishes to keep their computer's performance and features up to par with what's currently available, they should keep their packages, including their kernel, up to date. With modern GUI-based upgrade procedures as simple as scan, select, apply, ignore, the kernel is just another package, and pre-compiled binary kernels will often come with all the requisite modules and configure itself to become bootable. Minimal input is required by the user, except for perhaps "This upgrade didn't work, I'll boot the previous option and revert."

I know that my workstations are noticeably more responsive since upgrading to 2.4, and I make extensive use of much of the added/enhanced functionality. However, if what you're running works for you, does the job, is stable and secure, and still being maintained (which kernels as far back as 2.0.x still are), there's no definite reason to upgrade. If it ain't broke, and all that. That's the very same reason I still service a lot of Windows'98(SE) customers, and the very reason I'll often reccomend that they stick right where they are. To Linux, FreeBSD, etc. users I'll offer the same advise. Anyone doing otherwise is offering a disservice.

Re:Linux's new target market

[Osty]

I consider myself a pretty big geek, being thoroughly integrated into the Slashdot hive mind and all. I'm on top of CVS gaim, xine/totem, Mozilla, and GNOME (hoping it won't suck eventually) because they seem to become noticeably better by the day. But the kernel? Could someone please explain why?

Like I said, I was on that kernel upgrade treadmill, and I think it's a bit of an addiction. You want the bragging rights to say that you're running whatever the latest and greatest version of the kernel happens to be. For most people, it's about showing off and making themselves feel superior because they're running the latest stuff. I think it goes the same for the rest of the stuff you mentioned as well. Once those projects get to the "good enough" point, what do you gain by using the CVS nightly drop vs. a released version? Why not Mozilla 1.1 (or whatever Mozilla is at now) rather than CVS? I used to do the same thing with Microsoft stuff. I ran various betas of Internet Explorer (IE4, IE5, IE6), and even had various RC releases (all legally acquired) of XP and Win2K (even back before Win2K was Win2K). Now, though, I'd rather get my work done instead of futzing with my system.

Re:Linux's new target market

[Erik Hollensbe]

I think the argument is (since long, long ago, I was doing tech support too), that sometimes it's hard enough to get people to POINT AND CLICK over the phone, much less, type things.

Read your shell script (or even better, a perl script) out loud, and that might just give you the slightest on how hard it might be to do something like that over hte phone.

Of course, to provide a counter-argument to this one, something that I do with a couple of linux-using friends is set them up with sshd and an account for me, tell them to get on the net and discover their ip address, and do it myself.

Re:Linux's new target market

[Jedi Alec]

Ah, if it were only as easy as double-clicking an rpm...speaking from the perspective of a Windows user with a couple of years tech-support experience who happens to have a spare harddrive on which I installed Linux "for fun" please allow me to tell you it just isn't that simple. I downloaded some rpm's left and right to install stuff like Wine for example and double-clicking the buggers just didn't do the trick. Opening up the command line and executing the rpm's by hand revealed that a couple of command line switches had to be added in order for it to work properly, but in order to find that out I already had to go through obscure text files, and check a couple of websites. So far I've been able to find 1(one) piece of software that under Linux installs as easily as it does under Windows, and that's Mozilla.

Changelog

[SexyTr0llGal]

Found the changelog here. It reads:

Linux 2.2.24-rc5

* Fix n_hdlc globals pollution (Paul Fulghum)
* Fix initialisation of sk->sleep (Holger Smolinksi)
* Handle init_ethdev returning null in tulip (Neale Banks)
* Backport rtc wildcard fix to 2.2 (Paul Gortmaker)
* Correct wireless config help (Neale Banks)
* Fix smc9194 build (me)

Re:Changelog

[Kourino]

Remember, that's just the changelog from -rc4 to -rc5. There's more changes between .23 and .24, unfortunately Alan doesn't seem to keep as accurate changelogs in 2.2.

why 2.2? simple answer

[StandardDeviant]

If it ain't broke, don't fix it.

There's some nice things in 2.4, certainly (e.g. USB support that works). My home machines and laptop run it. But many of the servers I admin have been humming along fine with the 2.2 tree for quite some time, so I see very little reason to upgrade (indeed, with the hairiness in the 2.4 tree's virtual memory handling, I can several reasons NOT to upgrade beyond just change management). (Some will say that you should upgrade to 2.4 for the new firewalling features, but I prefer to put firewalling onto a dedicated openbsd machine or an appliance like a netscreen so the issue is moot for me.)

I'm running ...

[B3ryllium]

kernel 1.0.1

It kind of itches a little.

Re:Firewalling

[Student_Tech]

I had a machine (on its 3rd motherboard, last 2 died, 1 of which had been purchased specificly for it) that is running 2.2.10 because I couldn't get the ppp stuff upgraded to work with 2.2.12 (clue to the last time I played with it). It is doing a firewall between home network and ISP (over 56K modem) and routing between the 10 Mb/s and 100 Mb/s networks at home (got some 10 only cards and a 100 only hub about 3.5-4 years ago).

Re:Firewalling

[darnok]

Me too - I've got 2.2 running a firewall on a 100MHz Pentium box, and the box does exactly what it's supposed to do, all the time, and has only ever been rebooted in the last 4 years when there's been blackouts in my area.

A 2.4 box would be "cooler", and would probably even have some extra capabilities that I might find useful, but the simple rule is that you don't change something that works perfectly.

My main reason for keeping 2.2 around...

[samrolken]

I don't really have a choice. It's well-known that the 2.4 kernels can't compile properly for the 32-bit sparc architecture. http://www.rocklinux.org/mailing-list/rock-ports/2 001-7/5.html -- Sam Kennedy

Re:My main reason for keeping 2.2 around...

[Tester]

I use 2.4.20 on debian stable on a SparcClassic and s SparcStation 5 and it works very well. I never had any problem with it.. And with the speed of the disk on that thing, I really need ext3, because when my roomate pulls the plug I can't really wait 3 hours before my firewall is back up.. SparcClassics make really nice firewall especially if you find a scsi-1 hardware that's not too noisy..

Re:My main reason for keeping 2.2 around...

[ader]

I have a Classic running RHL 6.2 because that was the last supported release on SPARC. It's been extremely stable (KDE/GNOME apps run like dying dogs though). However, my new SS5 will run Aurora Linux 1.0, which is based on RHL 7.3 and has a 2.4 kernel.

Ade_
/

Need 2.2 for microsecond packet timing.

[Chmarr]

I've written a little application around libpcap that needs the microsecond resolution for packet arrival times. 2.2 has that. 2.4 only gives me 10 millisecond resolution.

Re:Need 2.2 for microsecond packet timing.

[Effugas]

Chmarr--

Try this:

int immediate = 1; ...
ioctl(pcap_fileno(pcap), BIOCIMMEDIATE, &immediate);

Does screw with some nonblocking modes, though.

Another quick tip: __attribute__ ((packed)); after your structure declarations will make structs vastly nicer to apply against raw packets in a cross platform manner.

Whatcha trying to write?

Yours Truly,

Dan Kaminsky
DoxPara Research
http://www.doxpara.com

Is this the long awaited

[infonography]

Sanders Kernel? I always know to trust the Kernel.

Linux 2.2.x?

[JWSmythe]

People still use 2.2.x?

Just kidding. :)

As I recall, I had some old old Slackware machines . I don't even remember the version number, but I think they started out with a 2.0.x kernel. On most of our machines, I didn't really want to take them down til they died of old age or whatever (usually we wanted faster machines over time), but kept upgrading the kernels on some occasionally for new features.. As I recall, we just couldn't get the 2.4.x kernels to even compile on them, without library upgrades, which I wasn't prepared to do (and probably mess up) on a whole bunch of machines. :)

But, I'm sure there are plenty of people out there with 2.2.x still, who haven't had a need to upgrade. I was just working on a machine a few days ago, that is, and there's no need to upgrade, it works fine.

> uname -a
Linux foo.bar.com 2.2.13 #3 Sun Nov 21 18:45:36 EST 1999 i586 unknown

That machine is still running strong. We just upgraded the CPU, motherboard, and memory, but it was all compatable with the drivers that were compiled in back in 1999.. :) The CPU went up from an AMD 300Mhz to AMD 450Mhz.. hehe. Free upgrades are the best, right?

2.2 kernel...

[dennism]

I've got a Redhat 6.2 machine running 2.2.21... and I doubt if I'll even update it to .24. The only reason it was upgraded to .21 was because I swapped out the 10mbit network card for a 100mbit.

It's been running fine pretty much non-stop for 4 years now. The only time it ever is rebooted is when the power goes out (yeah, I know...)

It's rock solid, and serves as my apache and SSH server to the outside and my FTP, MySQL, NFS, Telnet and AppleTalk server to the inside. To top it off, it's running SETI@Home and it's on a P120 with 32 megs of RAM, with no swap space being used. Not too bad, in my opinion.

Unless I suddenly get a big outgoing bandwidth upgrade, and an increase in activity to go along with it, I don't think I'll ever be upgrading this machine :D

torn

[awing0]

I don't know, I'm always torn between losing my uptime and installing the latest kernel. Its a lose/lose situation. If slashdot wouldn't announce the new 2.2 kernel I could just go on believing I had the latest code and it wouldn't bother me. But now that I know there's a new kernel, I'll have to compile it and bear the pain of losing my (not much, but around 130 days now) uptime. Thanks again for dragging me out of my sheltered update life.

Embedded devices

[Anonymous Coward]

There seems to be an assumption that we are only talking about workstations/servers when it comes to if the 2.0 or 2.2 kernel is still in use. The reality is that the Linux kernel is in use on embedded devices and it is not always desirable to try to be bleeding edge with such devices. Once you get a 2.0 or 2.2 kernel to fit withen the design limitations of an embedded device, upgrading may mean rethinking the hardware and starting from scratch. For example, I would be willing to bet that the majority of Agenda Computing/VR3 devices will never be upgraded to version 2.4 of the kernel.

Re:Embedded devices

[kieltux]

The Agenda VR3 runs 2.4.0-test9. And the Agenda Community is working on Kernel 2.4.19/20 for VR3.

http://sourceforge.net/projects/agos

Re:why 2.2? simple answer

[GRH]

If it ain't broke, don't fix it.

No kiddin. My headless P100 MP3 player/server has been 2.2.19 since that kernel was released and it's never gone down (except for having to move the machine a few times).

It's behind my 2.4.20 firewall, so I'm not too concerned about security updates or patches on the old box.

So, in the end, a 2.4 upgrade would provide nothing, and waste a day of my time. There's your reason.

Re:Debian Kernel

[ocelotbob]

Wrong. Debain applies a few patches to their kernels as well. While not as many as the level of patches Redhat et al apply, they still apply some. Looking at the README.Debian included with 2.4.20, the big ones I can see are the bigmem patch, cramfs support, NFS performance tweaks, and a few dozen other patches.

On the flip side of the coin..

[XaXXon]

I just upgraded to 2.5.64 with Linus's patch ( mentioned yesterday) merged in.

I am running Gentoo and I first installed the gentoo-optimized 2.4.20 kernel. When I read the article yesterday I decided to make the jump to 2.5.64 + patch. Holy wow, Batman.

I'm running Gentoo under VMware on a dual 2.2 GHz Xeon (only 1 processor makes it through to the virtual machine, though). After figuring out that I needed new modutils, I had everything up and running. I started up a kernel compile with make -j 2 to really try and saturate the system, and moved the mouse around. The mouse was silky smooth, KDE quickly and properly recognized mouse-overs and everything was just so nice. I then booted back to 2.4.20 and ran the same test. Oh the pain! The mouse was chunky, KDE didn't even try and do mouseover animations.. it was horrible. I've switched grub to default to the 2.5 kernel and I'm not going back.

That said, this is a play machine and does nothing important. So if it crashes more often (no crashes yet), then it doesn't really bother me..

Re:On the flip side of the coin..

[Hal-9001]

I'm running Gentoo under VMware [vmware.com] on a dual 2.2 GHz Xeon (only 1 processor makes it through to the virtual machine, though)...I started up a kernel compile with make -j 2...

Exactly what is the point of running make -j 2 if VMWare only takes advantage of 1 processor (except maybe for shits and giggles)?

Re:On the flip side of the coin..

[ameoba]

If you've got dual 2.2G Xeons for your 'play machine', what's your main box look like?

Re:On the flip side of the coin..

[Billly Gates]

His main box is all around you. Its in this very room. When you go to work, when you go to church, when you watch television........ its a prison that you can not taste, see or touch.

Re:On the flip side of the coin..

[ultrabot]

Exactly what is the point of running make -j 2 if VMWare only takes advantage of 1 processor (except maybe for shits and giggles)?

Because it is actually faster than w/o -j 2. I have a single cpu system and use something like -j 8. Compiling with multiple concurrent processes ensures that cpu keeps doing something while blocking on i/o, instead of just waiting. If your system has a lot of ram (>64MB), it can hack it.

Washington University's Robotic Photographer,

[BrodieBruce]

Lewis, which was featured here on /. sometime ago, is still running red hat 6.2. And I don't know of any software upgrades our lab will be giving him any time soon.

keep SW and HW in-sync

[wwwillem]

I have a simple rule. When I have a box of year 19XX, then I try to give it an OS of year 19XX and preferably the applications of year 19XX. (BTW, for the picky ones, same is true for 20XX :-).

The advantage is that you get good performance and that the drivers still understand old hardware like .... (fill in your dots).

So my Advantech 486 / 64MB IPC is still running fine as a firewall / DNS-server, with two ISA-based 3COM 509Cs. And given my rule above, it is of course running a 2.2 kernel, in this case RH 6.2.

My desktop (P5) is running Debian (also 2.2 kernel), but on the other hand my more recent notebook is using RH 8.0 (Linux 2.4).

Performance wise this is all pretty optimal, the only worry that you can have is that those older configurations are not coping with the latest virus attacks. Anyway, so far, so good......

Willem

Better virtual memory management

[dsfd]

According to my personal experience with 2.2, if you do a lot of small memory allocations and bring the machine very close to the limit (including swap space), the system will eventually stop working, perhaps not crashing, but not responding to rsh/telnet.

I remember that I even wrote a small C code to reproduce the problem. In summary: Alloc M bytes of RAM. If they are available, ask for M more. If not, ask for M/2, M/4 ... Repeat that until even 1 byte is not available. Store all the pointers and, between mallocs, access to all the memory correctly allocated before.

The problem disapeared with 2.4. This holds at least for the old versions of 2.2 we used, I don't know if it has been fixed.

kernel 2.0

[mog]

I recently deployed a new dial-up server at my work using Slink Debian, kernel 2.0.38. The machine is a 100Mhz 486 with 8MB of RAM. It's deliriously stable, and runs all the software we need on it (mgetty, atftpd, cron, and some sh scripts). Why would I go for a kernel that requires more memory (Debian 2.2+ wouldn't boot in 8MB of RAM), when the tried and true works great?

Re:archeology

[rsax]

My main box is a typewriter. I have it connected to my friend's typewriter in another apartment with a copper wire and we just type 1's and 0's back and forth. I think I might upgrade sometime soon.

2.0.36

[michael_cain]

Much as I hate to admit it, the household server is still running 2.0.36. IP masquerade and firewall for the household LAN's Internet access via cable modem, Samba print service for the other machines, backup storage for kids' schoolwork, and an antique version of Apache whose main function is to provide access to the Perl scripts that allow multiple people to share an old SCSI scanner. Haven't had a monitor connected to it for years, and the BIOS is old enough that you can tell it to ignore the fact that the keyboard check fails at boot time. The old AT power supply comes back up without any manual action after a power failure -- no idea how many of those it's been through, but the ext2 file system doesn't seem to have ever lost anything.

Kentucky Fried Linux.

[infonography]

ouh ouh, I gotta patent that idea. Thanks AC, if your name was here I would thank you in person. But since your nobody I get to walk all over your rights. ;)

I am sure it would be a Plucky little distro. We could showcase it at a Birds of a Feather meeting in Silicon Valley.

I could relabel the various bits of the OS, Call the whole system 'The Sky' as in 'the skies the limit'and if there was any problems a little yellow baby chicken graphical agent ala Clippy we'll call Chicken Little will run about the screen yelling 'The Sky is Falling! The Sky is Falling!" If you chase him with your mouse and can click on him he won't reboot your system. Kind of like what would happen if you were using a windows box.

If not I could sell it to Jeff Bezo. He could call it One Cluck computing.

LRP - LEAF

[MCZapf]

Though the Linux Router Project (LRP) is somewhat stalled, a lot of the work seems to have moved to the Linux Embedded Appliance Firewall (LEAF).

I used to use LRP on my router. Using such a stripped-down system was a great way to learn things. But eventually I switched to a minimal Debian install (once I got a hard drive for that old box).

2.2 forever..

[PeekabooCaribou]

cd /usr/src/linux
perl -i -e 's/^PATCHLEVEL = 4$/PATCHLEVEL = 2/' Makefile

All the benefits of 2.4, with the beautiful, symmetric numbering of 2.2.

Re:2.2 vs 2.4

[NerveGas]

Wow. 14 days between reboots? 90-day uptimes? Might I respectfully suggest that something on your systems sounds entirely broken?

Out of a pool of about 12 heavily-loaded servers that have been running for 4 years on 2.2 and 2.4 kernels, so far, I have had exactly *one* need to reboot that couldn't be positively traced to hardware problems. And that time I'm not entirely sure that it wasn't hardware-related, I just couldn't *prove* it.

The couple of times there have been hardware problems have been because of things like failed RAID cards or power supplies. I could count the number of incidents on one hand, and have at least one hand left over. A couple of the machines, in the 3 or 4 years they've been in service, have only been rebooted to switch colo facilities (twice) and for batched kernel+critical software (libc) upgrades (two or three times).

The last time I switched colocation facilities, *ALL* of the machines had been running for over a year. The thought of rebooting them never crossed my mind. And while some of them were very robust systems (triple-redundant power supplies, etc.), most of them were plain old commodity machines that I slapped together on my desk.

If you're really having to reboot those machines like that, you probably want to dig deeper and find out what the problems are. Chances are it's not just that one kernel version is more stable than another, it's that one kernel version doesn't exacerbate underlying, pre-existing problems as much as another.

steve

NBD!

[fuzza]

I use NBD (network block device) combined with software RAID1 to give automatic mirroring of data across 2 machines.

Unfortunately NBD in 2.4 simply _doesn't work_ - the client (with the nbd.o module) dies as soon as you try to transfer any significant amount of data (~4Kb). How it could have made it all the way to the stable kernel is beyond me, even with 2.4's reputation.

I've tried unapplying the NBD sections of each relevant patch since 2.2, (all of 2.4 and 2.3 series) to see where it broke, without much luck so far. I've worked out that the current behaviour has existed since 2.4.4. I got all the way back to 2.3.46, prior to which (haven't tried any earlier yet) it doesn't compile properly.

In short, yucky. So much for our backup solution on a RH6.2->RH8.0 upgrade...

Re:2.0.35, uptime 55 days 12 hours!

[DaveAtFraud]

root@fwupdate2:~/
# uname -a
Linux fwupdate2.vericept.com 2.2.16-22 #1 Tue Aug 22 16:16:55 EDT 2000 i586 unknown
root@fwupdate2:~/
# uptime
10:12pm up 153 days, 14:51, 1 user, load average: 0.00, 0.00, 0.00
root@fwupdate2:~/
# cat /proc/cpuinfo
processor : 0
vendor_id : CyrixInstead
cpu family : 5
model : 7
model name : Cyrix MediaGXtm MMXtm Enhanced
stepping : 4
fdiv_bug : no
hlt_bug : no
sep_bug : no
f00f_bug : no
coma_bug : no
fpu : yes
fpu_exception : yes
cpuid level : 2
wp : yes
flags : fpu msr cx8 cmov 16 mmx cxmmx
bogomips : 88.27

root@fwupdate2:~/

Why do we make such a big thing out of upgrading?

[Dthoma]

When kernel 2.6 comes out, no doubt we'll all cootchy-coo over it and quite a number of us will run to download it simply because it has a lot of improvements and because it's the most functional kernel. And yet in four, five or six years' time those same people would probably recoil in horror if they found out that someone is "still" using 2.6 because "everyone knows" that some newer kernel is "so much better".

If something works now, why won't it work in a few years time with the same hardware? If stability is important to you, isn't it better to stick to something tried and tested?

Don't upgrade to 2.4!

[Arandir]

Don't upgrade to 2.4! If you do, then Linux will never beat FreeBSD in the uptime department!

quite happy with 2.2.x for a dedicated server

[yahkah]

I run 2.2.18 on an ancient sparcstation used as a small traffic web and cvs server. It just passed its 397th day of uptime.

Bridging firewalls!

[Krellan]

I run a 2.2 kernel, with the patch for Ethernet bridging and firewalling.

http://bridge.sourceforge.net/


http://www.math.leidenuniv.nl/mailman/listinfo/bri dge


It seems that the new firewalling technique of 2.4 (iptables) does not play well with Ethernet bridges.

I have a DSL connection to a small subnet of static IP addresses (/29). The problem is that the DSL uplink, out of my control and unfirewalled, is on one of the addresses in my subnet! It's as if there is a fox in the henhouse.

There is no proper routing subnet, as there should be. This is no doubt because of the IP address shortage. The DSL uplink must exist on the same subnet as my machines, giving me only 5 usable addresses for my machines. Broadcasts must be passed correctly, or the machines won't be able to ARP each other. Proxy ARP is not an option, because of the need to keep the DSL uplink on the same subnet.

So, I run Ethernet bridging with firewalling. I bridge two Ethernet cards together, passing broadcast packets between them (filtering out externally generated "smurf" broadcast packets, of course). I also implement my firewall at this point. The network is one logical LAN, but partitioned into two physical LAN's, with the firewall machine in between them. The firewall makes sure that unwanted packets from the DSL uplink never reach my machines.

It's not perfect (there is no stateful connection filtering), but it has worked well for me. Probes come in at least every hour, and no successful breakins to my knowledge.

And another reason not to upgrade? The machine's uptime is now at 326 days, I'm going for the year :)

Old VMWare license

[karlm]

I dual boot 2.2 and 2.4 kernels. My Fuji FinePix digital camera and IBM USB keyboard appear to not interface with kernels in the 2.2 series. (I'm the only person I know to run dual keyboards.) However, I bought a VMWare 2.x license and don't want to shell out $300 for the latest VMWare version until there's a version that supports the Linux 2.6 kernels. I've so far resisted the temptation to grab a VMWare keygen or cracked version.

2.0.35, Baby!

[vogon jeltz]

What do you mean, 2.2? Seriously, the "server" in my parents not-so-small medical cabinet connects to 5 serial terminals and a couple of printers and card readers. I set it up in 1998 (downgrading from SCO ;-) and it's been running ever since, litterally 24x365. Not one single crash. It runs 2.0.35.