Follow Slashdot stories on Twitter

 


Forgot your password?
Close
typodupeerror
×
Linux Software

Linux 2.2.11 Released 180

Posted by justin++ from the linuxworld-present?-:) dept.
kuro5hin writes "Yet another in the stable 2.2 series, linux 2.2.11 is out and on the mirrors. Download, patch, compile, repeat as necessary. " Seems Alan has blessed us with a present for LinuxWorld. Please use a mirror to download it.
This discussion has been archived. No new comments can be posted.

Linux 2.2.11 Released More

Linux 2.2.11 Released

Comments Filter:
  • Most of the addresses, ftp.us.kernel.org in particular, point to several machines, and which one you get is random. If you find a mirror that hasn't updated yet, just disconnect and try again. Same deal if you hit a really loaded server. Eventually you'll find one you like. I had no trouble just now.

    And, of course, always get the patch(es), if you can.
  • Actually, since Alan Cox is now the official maintainer of the stable tree we might see kernels coming out at this time more often, considering he lives in England! :)
  • I just have to whine - just when I FINALLY got some ISA ISDN cards this morning, what happens? Almost simultunaeosly comes 2.2.11 with support for those teles pci -cards too..
    --
  • Really, people into radio have thought in terms of UTC (formerly GMT) for years - remember where the International date line is and that on any day, when someone in Hawaii is just stumbling out of bed to get the paper and make breakfast, someone in Japan has already come home from work!
    It's always sometime somewhere.

    Chuck
  • Maybe what we need is a view of economics where value is expressed as a complex number w/ both real and imaginary components; then products and services can have a phase angle as well as magnitude, and could go a long way toward pleasing both the cluel^H^H^H^H^Hlayperson user and the technical specialist at the same time, instead of one at the expense of the other. Where's my application form for a research grant...

    But I may be hallucinating again.

    Chuck
  • So am I. Matter of fact, I just woke up (2:23 PM EST). Nice to see my story got posted! That's the first time.

    BTW, the reason I knew about the new kernel was because of the linux-kernel-patch email list. Just send a message to majordomo@vger.rutgers.edu with "subscribe linux-kernel-patch " in the body of the message. They email out changelogs and, if the patches are small enough, complete patches whenever there's a fresh kernel.
    ----------------------
    "This moon-cheese will make me very rich! Very rich indeed!


  • Make sure your local machine name is listed in your hosts file. This will solve that problem.


  • Do we have DVD+USB support in main kernels yet?
    SOrry haven't been following kernel devel recently *ouch*

    I am planning to get a laptop soon, so I need to know which DVD drives are supported (if at all!) to make a decision.

    linuxlover

  • My provider offered 128Kbit for $30 bucks,
    or 1.5Mbit for $60/month. (it's an apartment complex and they run few T1s in it, then go over
    phone cables) I decided to save money.
    Nevetherless, I ALWAYS get around 80 Kbyte
    download/upload to my office machine and under 14ms ping.
    I think there are actually no 2 speed options -
    it all sits on the same pipe. It just a scam.

    P.S. It is amusing how many people on the subnet run stock Win95 with file sharing (to the whole world to see and write)... I thought about mailing them with a warning, but decided not to bother. What do you think, should I warn them? They are good people in this neibourhood.
  • It defaults to your old configuration, and prompts you about new options only.
  • Wow, I didn't know that. I always figured they were the same server. Thanks. Still, it irks me to find out that some servers (notably, ftp.uk.kernel.org) seem to lag behind for days (I think 4 days is the longest I remember). I think it makes more since to call these "photograph" sites.
  • Actually most of us do it for fun.
  • What's the point. Everytime we make big hype about a public release, something allways goes wrong, and kernel.blah.blah+1 shal be released within a few days. What's the point of using slashdotters as dogfood. heh. save the linux-kernel list the spam
  • Just compiled the 2.2.10 kernel on a slow 486 last friday. I'm very tempted to pretent that the new kernel is not there at all.....
    *sigh*
    Well, time to read up the new features/bug-fixes before I decide whether to download it or not... might not need it this time.
  • He lives in Wales - see his wife's diary at http://roadrunner.swansea.linux.org.uk/~hobbit/ind ex.html.
  • Nah. "Australia".
  • Oh, but I always read your Kernel Newsflash [csiro.au], so nothing can happen to me!!

    But somehow I don't you've signed with your real name.

    Anyway, I do have a clue about the Linux kernel, and no, I haven't read the patch. For whatever reason, I do trust that Alan Cox did check all of it.

    Sure, one day a stable kernel will come out with a bug that will cause filesystem corruption for lots of people. It'll catch a few hundred people. And one day my harddisk will crash as well.

    But a trojan in the stable kernel is just totally unlikely. Trojans are much more likely to show up in the huge amounts of other software that can be grabbed from the net. And do you think that kernel developers check all the (non-kernel related) sources that they compile?
  • i hope
  • ...that news got posted at this time of night (or is it morning?). Props to Justin, for serving slashdotteres at 3:26 AM. Or maybe I'm just being silly and not thinking about the other coast of America.

    I'll just quietly go and get 2.2.11 before the servers get busy.
  • Where is the beer then???
  • Sendmail looks up your hostname before starting up, and it takes a long time for it to find if it does not succeed (for example, you blew up your eth0, erased your /etc/hosts.conf or your /etc/resolv.conf).

    Try looking up any other thing, you should have the same problem - nslookup will seem to hang forever until it dies. The same if you try to ping any machine without specifying its IP.

    A word of warning - There have been ENORMOUS changes between RedHat 5.2 and the current Linux versions. Kernel 2.2.x will NOT work correctly with 5.x RHs if you don't install a bunch of RPMs, including the new glibc (which may force you to rebuild many programs).

    My advice? Either downgrade, or (if possible) get an updated CD.

  • Those guys don't rest do they?

    *Yawn*

    Guess I'll go get me some source then =)

    - dc.

  • >Mind you, I'm not sure if doing a compile on an NFS-mounted filesystem will be much faster.

    It has been my experience that it is much faster. I'm running a 486/100 w/16 mb as my masq/gateway box, I only have a 200mb HD on it now and no monitor, kbd - so I only compile from my PII on a nfs-mounted FS on the 486. This works just fine, compiles much faster than the 486 ever did (used to run this as a stand-alone box before I got cable) and it seems to complie just about as fast as a native filesystem on the PII does. (although I'm running a 10/100 mbps network, so your mileage could vary if you are using 10mbps) ;-)

    I used to run an older kernel on this box because I hadn't gotten around to figuring out how to cross-compile from another macine, and the old kernel was fine. When I wanted to start port-forwarding, I moved up to 2.0.36 and patched it with port forwarding and recompiled from the PII. No sweat.

  • Anybody else notice that there is a serious shortage of moderator points being allotted lately?

    Used to be that offtopic crap used to get hammered down to -1 where I didn't have to see it. Now I guess I'll have to start browsing at a score of 1 to filter out the AC's. :-)


  • Only 12:35 on the west coast... Who doesn't check /. before going to bed :), but then again... The night is young... the DSL fast... 20 K/sec off the US Mirror. This will do nicely. -SyniK
  • Posted by Justin:

    bedtime = 3am-noon
    waketime = noon-7pm

    i heard about this an hour ago and gave the mirrors time to sync. thanks for the kind words, though :) and yes, i do live on the east coast, btw.
  • And so is the 4.0-19990809-CURRENT iso image.

    Is it really necessary to announce every kernel patch? It's not like Alan Cox isn't going to release 8 more next week. Like we really need to know that 2.2.11-ac12 is out...
  • Actually, 2.2.11 is the first kernel patch in recent memory to fix a pertinent, widespread problem, namely the massive FS corruption that seemed endemic in 2.2.10. How bad was it? It scared me into buying a tape drive. I'm quite glad to see 2.2.11.

  • Patching your system beyond recognition is not the path to job security. There is no such thing as true job security, no-one is indespensable. Anyone who kids themeselves that they are is sadly deluded.
  • Then run windows :)

    This is a definition of "fun" of which I was not previously aware...
  • The convicts who got caught went to Australia.

    Those who got away went to the americas.

    All the white-collar criminals and corrupt nobles stayed in Britain.

    Just a pessimistic view of the heritage of the english-speaking world :)
  • morning: 0-4 hours after I wake up
    afternoon: 5-8 hours after I wake up
    evening: 9-12 hours after I wake up

    :-)

    ---
    Yust because you're paranoid
    Doesn't mean THEY are not out to get you
    ---
  • Well, at the time of writing, my local mirror
    (ftp.uk.kernel.org, at HENSA) had it, in its full
    glory.

    Plenty fast enough for us UKers cowardly enough
    to want the stable series.. :-)

    Remember, some mirrors are faster than others-
    balance speed of update with mirror load in making
    your choice. There are a LOT of mirrors.

    Enjoy..


  • DOOM clone is in Excel 95. You might be amused by the flight simulator in Excel 97, although it is admittedly somewhat less interesting.

    Here's how to get to it:

    (courtesy of eeggs.com:)
    1. On a new Worksheet, Press F5
    2. Type X97:L97 and hit enter
    3. Press the tab key
    4. Hold Ctrl-Shift
    5. Click on the Chart Wizard toolbar button
    6. Use mouse to fly around - Right button forward/ Left button reverse

    Drop by eeggs.com to find some of the other, less "official"-looking (read: likely non-sanctioned) eggs.
    ---
  • If you're supporting Microsoft, then there's something that you should be aware of. If linux has 6 minutes remaining, then Microsoft's current flavors of Windows (9x, 2000, NT 4 & 5) have about 3 left. The Next Best Thing(tm) will probably be out in about 3-4 years, and most geeks will flock to it. Those who use Linux then may be called the lamers, those who cannot think for themself, who refuse to change with the times, those who refuse to embrace the Next Best Thing(tm). Those who still use Windows will be living in a technological stone-age, no longer influenced by the current technologies (much like all incarnations of DOS, and even the older Apple II and Commodore are now). Those who then continue to use that software will no longer be ridiculed by those on the bleeding-edge of technology, their attention will be left with the users who still use Linux.

    Unfortunately, such is the computer's lifecycle. It is inevitable that Linux will no longer be the "geek's OS" in the near future.

    Also IMO, this could all be triggered by Red Hat's IPO. Money = Power ==> Greed ==> Corruption ==> Indifference to anything but the making of more $$. If you need proof, look at Microsoft. While Linux as a whole is far from being controlled by any one distribution, if one company gets enough power, then they're the one calling all the shots.

    I would not be suprised if within the next year, there become real, tangable rivalries between distributions, fueled by greed and power.

    Of course, this is all my opinion, I may be wrong.
  • yeah i do. but the server is inside a large intranet - who's going to bother? Anyway, the bug causes a reboot, upgrading the kernel is a reboot: so why reboot to prevent a bug from rebooting the machine when the machine hasn't rebooted because of that bug?

    and i have 2.2.7-ac compiled on it and ready to go should the machine ever need a reboot. I'll wait a week or three, and if 2.2.11 turns out to be stable i'll compile it and replace the unused 2.2.7-acX.


  • A question: What money?!?! Did you pay any money at all to Linus?

  • Anyone isolated cases/hardware where the fs bug seems to happen? I run several linux machines for personal testing/general productivity, and I've never had a fs problem.... typing this on a 2.2.10 system that's doing fine in it's file system. Is it on some scsi drives/controllers, odd ide timings (mb has via82c586 chipset, option enabled, no problems) raid controllers? Basically, any common elements to the fs corruption?

    BTW, rest of the boxes just have generic ide... no special chipset support... no scsi cause ide has never been too slow yet.. always too many other things going in the the background I call life...
  • We sent the convicts to Australia

    America got the religious nutters
  • This release has had a lot of prereleases, so it should be pretty well tested. If you missed those prereleases it's probably because they weren't announced here.
  • Oh, well maybe it's just me that's got it wrong then ;)
  • Aye, and that's why I be patchin' tonight! I too fear for the corruption, as I got a minor hit from it in recent weeks. It seems I had some null/unprintable characters in my Modelines in my XF86Config files. very odd indeed.
  • take a look at the article in linuxtoday.

    there's a surprising amount of fixes. (smp races, ext2fs problems, driver problems)

    this one's worth the time to upgrade
  • look at the article in linuxtoday

  • I looked all over last night for this, but didn't find it until tonight.

    Apparently there are some significant SECURITY UPDATES in this one.

    Take a look at http://www.linux.org.uk/VERSION/ relnotes.2211.html [linux.org.uk] for details.

    "
    It fixes security holes in the kernel so should seriously be considered as an important update for all Linux 2.2 systems."

  • I am normally not one to advocate "up"(?)grading just to have the largest available version number, but this isn't a case of something that is "not broken."

    "
    It fixes security holes in the kernel so should seriously be considered as an important update for all Linux 2.2 systems."

    Anyone considering not upgrading to this kernel should go read http://www.linux.org.uk/VERSION/ relnotes.2211.html [linux.org.uk] and reconsider.

  • Smart users will use the pgp signature to verify the package! I do it all the time.

    You can also maintain oldversions of your kernel to immediately go back if you have problems.
    ---------------------------
    ^_^ smile death approaches.
  • There is wisdom in staying with the you already trust. Every kernel for every OS has bugs that can be exploited or take down your machine.
    Stable means that the kernel works well enough on enough machines and the amount of bugs are negligable to the user.
    There is wisdom in upgrading for these reasons:

    Kernel Exploits found.

    More efficient code.

    More traps for GPFs.

    Workarounds for Hardware errors like F00F.

    Better filesystem support.

    More EYES the less bugs.


    The list goes on. I advise to backup your old kernel, give new kernels a try and if your computer crashes go back to the old one
    (Don't do this for mission critical servers). A petty crash won't fry your machine (I hope).
    Also complain about any anomolys you may find. The more eyes the better linux gets. That is the heart of linux's stability.
    ---------------------------
    ^_^ smile death approaches.

  • Stupid troll... Go back to AOHell.
  • This is not a 2.2.11-ac12 type of release, but the release of stable kernel that has had weeks of testing. Not incredibly big news, but still noteworthy. The last stable release was about 8 weeks ago.
  • Isn't this on the list of /. trolls? "BSD 343982839-238 is out"

  • Uh, I do check the checksum. If Linus proves to be ineffective as kernel maintainer, then he will be replaced. He has so far, however, proven himself to be fairly good at knowing what goes in and what stays out.

    That's one of the things maintainers do: filter out all the crap code. That's why projects have maintainers. That's why we don't have a bazillion trojans in the code.

  • >nfortunately Linux is buggy like an old rug, and those kernel >upgrades are in fact *needed* to keep production servers alive >(actually, no sober sysadmin would run Linux on a production server - >but there's no such thing as a sober sysadmin anyway).
    Hmmm. Don't you reallly mean Windows 2000 rather than Linux? In case you didn't know as of 8/10/1999 at 7:48 am the Windows2000test.com site has gone down yet again....Doesn't say much for you and your microsoft buddies, does it?
  • Either do a make oldconfig or make sure that you have selected the maximum physical memory size you want in the config. It's a new option that's in this release.

    Using make xconfig it's on the same page as the choice for SMP MTRR etc.

    Iggy
  • Eat a beaver:)

  • > no sober sysadmin would run Linux on a production server

    Hey Rob, I think it's time to change AC from Anonymous Coward to Astroturfing Chump.

    ps - I'll send you the money for a sixpack if you do, even temporarily.

  • AC==Chicken Little

    The "downfall" of Linux? Hardly. *If* it ever happens a few folks will get they're fingers burned, and the rest will be a bit more prudent with the checksum sig.

    Back under the bridge troll...

    Krakken
  • indeed, upgrading kernels for the sake of it is plain stupidity. (anyone upgrade from 2.2.5 or so to 2.2.8 just for the sake of it? see what i mean?)

    i have a machine with uptime of ~80days running 2.2.2-ac7. Previous uptime was ~90 days on 2.0.36.
    The reboot was to upgrade the kernel. And i only did that cause 2.2 had a couple of things i wanted. I tested the new kernel on non esential machines first for a while before upgrading the server.

    you don't have to upgrade to every new kernel. In fact doing so is silly on anything else apart from your desktop linux 'play' box.

  • As of 11:30 GMT I don't see anything on the list to indicate file system corruption. Do you have a reference for this?

  • Swansea, Wales. Also check out Alan's diary [linux.org.uk]. Not quite as funny though
  • Posted by Synsthe:

    this will be the downfall of linux.

    You're kidding right? Do you know _anybody_ who actually reads the MS Windows changelog when upgrading from 95 tp 98, or 9x to NT, etc? Does MS even provide changelogs?

    This is indicative of nothing more than the simple fact that your kind (trolls and pranksters) are alive and well on the internet, and aren't showing any signs of going away.

    Please, quit your job at Microsoft and get real work. Spreading FUD only serves to get a big red "troll" stamped on your forehead. ;-)

    --
    Mark Waterous (mark@projectlinux.org)
  • Too long to compile on an 8MB 386DX33? Then use your kick-a$$ machine to cross-compile the kernel for your 386 machine. :) Ah the wonders of free software. :)
  • kernel 2.2.11 is running on both linuxgod.net and ws2.linuxgod.net. After 1 hour of screwing around i finally got the damn thing to compile my custom work for ATAPI/SCSI emulation. If you download it, download from ftp://ftp.us.kernel.org. I got a pretty fast download off a 56k taking 33 minutes of my wasted free time. http://www.linuxgod.net ( if im online )
  • What do you mean? There is no trojan in the linux kernel? Phew. Anyway, can you point at where I attack anyone? Maybe you should learn to read (and write, or find a place where euthanasia is cheap).
  • I thought that's how they came out with the Windows 2000 Betas..
  • Yeah, in comparison, Microsoft's expediency with fixes to bugs is phenomenal. You have to wait 6 months for a service pack that fixes a bug that exists but supposedly hasn't affected anyone.
  • You guys just don't get the concept of tongue in cheek humour, do you?
  • some servers (notably, ftp.uk.kernel.org) seem to lag behind for days

    I believe ftp.uk.kernel.org is actually four servers - Demon and three on JANET. Luckily JANET has better LINX connectivity than it used to :-)
  • Some people Just Don't Get It.

    The more people who use a kernal, the more bugs they find, the more bugs get fixed, the better the OS gets. Upgrading is the _whole point_.

    Now, if you've got a production server you don't want to risk (a reasonable concern) then you slam the upgrade on your test server (you _do_ have a test server, don't you?) and run it there for a while.

    Upgrade! It's not just fun, it's part of the evolution process!
  • Watch it, slag. Linus has made a couple of questionable decisions in the past, but he has basically been a very responsible maintainer.

    I grepped the CREDITS and MAINTAINERS files for "Anonymous Coward", but found no mention of you - perhaps you have no kernel coding credentials either.

    Matthew.

  • I bet it's more likely that one of the thousands of anonymous Microsoft employees puts a Trojan in Windows 2000 than that Linus and/or Alan put one in a kernel.
  • There is a difference between K/sec which means Kilobytes per second, and kbps, which means KiloBITS per second. Also take note that KB is kilobyte, and kb is kilobit. I doubt that in New York you can get 4 meg per second constantly. His dsl is about 160 kbps, which compare much better to you average. But still, 20K for dsl? BellAtt gets around 40 -50 average.
  • Actually, any version of Linux is infinate in value. If value is coolness/cost and cost=0, then now matter what the coolness is, value is infinity. That also means that if windows were free (it is, how many people actuall BOUGHT a copy of Windows 98?) it would also be infinite in value. Must be something wrong with my equation. Oh Well.

  • I bet it's more likely that one of the thousands of anonymous Microsoft employees puts a Trojan in Windows 2000 than that Linus and/or Alan put one in a kernel.

    That's a sobering thought...

    In an environment where people can be fired for inserting easter eggs et al in the product, you still see pinball games, flight simulators and even entire DOOM CLONES in Microsoft software. And these people are just screwing around for fun... what happens when you get someone with an agenda working on these projects? Say, a psychopath or a terrorist?

    How well do companies like Microsoft screen their employees?


    ---
  • actually Win2K was created in a stadium sized room packed with thousands of monkeys punching away at computers...

  • Nothing wrong with your equation (Score:3)

    by Gleef ( 86 ) on Tuesday August 10, 1999 @06:46AM (#1755513) Homepage
    be-fan wrote:

    Actually, any version of Linux is infinate in value. If value is coolness/cost and cost=0, then now matter what the coolness is, value is infinity. That also means that if windows were free (it is, how many people actuall BOUGHT a copy of Windows 98?) it would also be infinite in value. Must be something wrong with my equation. Oh Well.

    There's nothing wrong with your equation. People get Windows three ways: Retail sales (which are paid for), OEM computer purchases (which are paid for, the cost is just hidden in the cost of the system), and illegal copying (which are free, so infinite in value). In addition, the coolness of Windows is imaginary, so the value of any costly copy of Windows is also imaginary. The only Windows with real value are pirated copies :-).

    Likewise, the free developer's release of Be is infinite in value, but the users' versions are not (but they do have real value, not imaginary). Downloading Linux is far more valuable than getting a CD, but only if you have a flat rate internet connection. I think I should stop now.

    ----
  • Except until recently, Ireland was *not* part of the "English" speaking world. But the English tried to kill the Gaelic World by shipping off any people who protested, and then killed off many other Gaelic speakers by percipitating the Potato Famine.

    ttyl
    Farrell
  • really... make oldconfig
    I've never heard of that. What does this do as opposed to make config?
  • ask netscape...
    they say: its spelt "N-E-T-S-C-A-P-E" but pronunced as "Mozilla"

    go figure...

    also i think E checks for the Beers for you during compile... but i don'tdrink alcohol, so it never found any in my fridge :)

  • The ac kernel patches should usually be available from ftp.*.kernel.org/pub/linux/kernel/alan (Replace the * with a country code, ie lu for Luxembourg or no for Norway) Unfortunately I didn't find any homepage for them.

    As for them being better than the 'normal' ones, I have absolutely no idea because until now I only used the standard kernels. I was told that ac contains newer drivers but is a little more unstable. (Sounds like something between 2.2 and 2.3 ;)

    Anyway, keep compiling!
  • Is that stable in the sense of " constantly being updated because we tweak stuff" or "constantly changing because we fix broken stuff" or "completely stable but we release even more completely stable versions every three weeks" or "It's stable because it doesn't crash, but you need to reboot every three weeks to upgrade the kernel anyway" or, rather is it stable in the sense of:

    "Not stable, but in fact, constantly changing."

    Only 50% tongue in cheek....


  • ...I've been told that the latest "ac" kernel is really the one to have...

    It has been shown that support costs increase dramatically with every deviation from a standard operating system release. You may know that your system was patched, why and how it was done. Your successor on the job will not. Beware.

  • How about "Stable, but who cares? All the cool nifty's are in the devel kernels" =)

    Seriously, Stable is boring. On a mission critical server, that may not be the case, but for me, well I just want to have fun.

    Stable means a feature freeze for bug fixes. But when you feature freeze, you don't get the new cool gizmos =) Like USB and what not.

    So the moral of the story? Stable is all well and good because it probably helps the devel kernel ppl fix some of the underlying problems :-D
  • Hmm still wondering why /dev/changer can't be put into the kernel tree, even the 2.3.x tree seems to stable for it...

    See /dev/changer [unfix.org]

    Maybe when it's in the experimental 2.3.x tree, maybe that more people would try it which would speed development up as more hardware can be tested... There is a multiple-cdrom-status in 2.3.x which is stated as "beginning of multimount-support" while /dev/changer is working since 19 May 1999...

    Greets,
    Jeroen Massar

    (/dev/changer maintainer/coder)
    --
  • Come on dont start the same story again.

    The development models of Linux and FreeBSD
    are completely different. It sounds strange for Linux Folks that B*BSD is unchanged for months and for the BSD folks that Linux has a new Kernel release every two weeks.

    But stop the stupid comments about craps and
    'going with the times'. Go use NT then!
    I have F*BSD and Linux dual boot and I love them both.
  • I'm getting the same error message. Haven't bothered to look it up yet.

  • My production fileserver is 2.0.34 (behind a firewall, keep your panties on)...

    the web/mail/news/radius/ldap/db server is 2.2.1 I believe (too far away to keep rebooting)... I need to get around to getting a new kernel on this because there's a DoS which takes down servers 2.2.3 I think. It's got 2.2.5 ready to come in on next reboot so if it does get shut down it comes back fine. :-)

    My masq server at home: 2.0.33 (it takes too damn long to compile on an 8M 386DX33!) But then my toy boxes are running 2.2.9 and 2.3.something.

    Stable? damn right. that 2.2.1 box has almost a half year of uptime and there ain't nothing protecting its ass but itself on a 10mb backbone 60km away from where I'm typing. The 2.0.34 fileserver had some decent uptime but something weird is happenning either with the EATA RAID controller or the SCSI Zip drive... sometimes accessing the zip drive causes a bus hang (EATA reports mbox # is full, resetting controller). If anyone has any insight, please let me know!

    anyway enough ranting. The point was that not everyone is hell-bent on the latest and greatest. :-)
  • I have redhat 6.0 (upgraded from 5.2), and when I went to 2.2.11, sendmail takes a long time (2 mins) to start up. Time to downgrade I guess.

  • 20 K a sec? Yech...

    Western New York cable modems get 500K/sec. ;)

    -Jeff

  • I'm sure most of the replies regarding job security and such were tongue-in-cheek, but the real point of keeping a standard kernel is to help manage your own workload. Doing massive modifications means keeping track of it somewhere - that means either in your head (good luck) or doing extensive documentation (good luck). What really happens is that you end up spending lotsa time backtracking over your old code and mods to try to figure out what the hell you were doing!
    Caveats: don't get anal about this - mods and patches are needed. But the more you can mods and patches, the easier your job becomes (use the time to manage more systems, do kernel research, or play Solitaire FAIC).
    f
  • Rather than doing an enormous diff and browsing through thousands of changes to see if the problems we *do* have on our server might be fixed, is there a place where a ChangeLog can be found? The Documentation/Changes file is just changes for 2.2 in general.

    -Lars
  • Oops...
    s/more you can mods/more you can avoid mods/

    f
  • Its been said over and over, but should keep being said:

    Avoid to "must upgrade" urge, especially on production servers. If it works, it doesn't need fixing.

    The idea that these are "stable" kernels is a misnomer, they're just tweaks and updates on the stable tree, and there's no guarantee they're actually stable. Example in point, one of my servers was experiencing pretty serious corruption using MySQL in the databases, whenever the load on the database got too high. System was running 2.2.5 SMP. Upgraded to 2.2.9/SMP and the problem went away. Pretty indicative of a problem in 2.2.5.

    If you've got a production server, you run the same risk that 2.2.11 might cause a problem that you weren't having before. So think first before upgrading, and only do it if you're having specific problems. We don't want people to start claiming Linux is as instable as NT because of the occasional bad kernel release (like some of the pre-2.0.36 kernels!).
  • i think you are talking about 2.2.11-preX - there was one note about FS corruption under 2.2.11-pre4 (AFAIR) and also i personaly got FS corruption under 2.2.11-pre7 (after 40 minutes).

    now i'm running on 2.2.11 for about 5 hours and nothing like FS corruption yet.

  • One thing that has been bothering me is the software RAID support patches from ftp.kernel.org. Does anyone know whether the stock kernels are going to be supporting the 0.90 raid tools in future versions? (For all I know, they already are...) What I am talking about are the patches from ftp://ftp.us.kernel.org/pub/linux/daemons/raid/alp ha/ . Sometimes I wonder whether or not I should be installing some of the RAID patches on a kernel since it almost seems as if the change might have been merged in already, but not quite...
  • Actually there is something wrong w/the equation. If value is coolness/cost and cost=0, then no matter what the coolness is, value is undefined. Therefore, if it is Linux we're talking about i would say that value is near infinity. However, as far as ms winblows goes, we all know that cost is not = to 0 so value will never be as high as Linux. As for free copies, the undefined value would have to be a negative.
  • Once upon a time, I started using this little operating system called Linux. I was working at a start-up where everybody had 1.2.13 kernels patched for ELF support. But everybody knew these kernels were rock solid. A year or two went by while work on 1.3.x progressed, and then 2.0.0 was released.

    A flurry of 2.0.x kernels came out. Finally, 2.0.18, the "last" stable kernel was released. No, wait, there was 2.0.19. Then 2.0.29. And so on, up to 2.0.37. Which we think is stable.

    Consider, for a moment, the system used by Gnus [gnus.org], the singing, dancing mail and news reader for Emacs. Gnus has alpha releases with names ("September", "Red", "Quassia", "Pterodactyl"). These aren't announced to the general public, though they are discussed on a mailing list. These are essentially analagous to the a.n.x Linux kernel releases, for odd n. Then there are beta releases, which have version numbers like a.n.x for even n. These are probably stable enough to use, but if you care a lot about your mail you'll wait. Finally, there is a single final release with a version number of the form a.n for odd n (5.3, 5.5, 5.7).

    The problem with the Linux kernel development model is that it has lots of "alpha" and "beta" releases, but no "final" release. So there's really no way to tell if a particular even-numbered kernel release is "stable" or not, aside from by reputation. At some point there needs to be a "final" we've-done-all-the-testing-we-can-possibly-think-o f-and-fixed-every-single-bug-which-is-wh y-we're-working-on-the-next-version release.

  • If you can't get replaced you can't get promoted.

  • I find it amazing that, even though a kernel release has no features of bugfixes which directly impact a user, that user feels compelled into upgrading.

    People actually argue that Linux releases too many kernel upgrades too often. As if someone stands behinds them, points a gun at their head and forces to them compile and install the kernel.

    I think this is probably one of those reasons FreeBSD is perceived to be more stable than Linux. It's because Linux people reboot their damn boxes for kernel upgrades every week (or 4) for no reason other than to have a spiffy new version number.

    If you have a production server, don't fix what's not broken. Kernel upgardes for the sake of kernel upgrades is not a good reason :)

    --

Slashdot Top Deals

The rule on staying alive as a program manager is to give 'em a number or give 'em a date, but never give 'em both at once.

Close