Slashdot is powered by your submissions, so send in your scoop

 


Forgot your password?
Close
typodupeerror
×
Red Hat Software Businesses

Redhat 5.2 2.2-Kernel Update 88

Posted by CmdrTaco from the stuff-to-play-with dept.
An anonymous reader "Red Hat has a set of update rpms that are apparently intended to make 5.2 fully compatible with the 2.2 kernels. (I've been using 2.2.1 since it was released with no ill effects anyway...) " I'm posting this just so people stop emailing me and asking. Besides, the rpms seem to be worki
This discussion has been archived. No new comments can be posted.

Redhat 5.2 2.2-Kernel Update More

Redhat 5.2 2.2-Kernel Update

Comments Filter:

  • No Subject Given (Score:1)

    by Anonymous Coward
    As one who bought 5.1 shortly before 5.2 was released, I'm not particularly pleased that upgrading 5.1 to kernel 2.2 is as difficult as it is. I d/led several .src.tar.gz updates but not one installed :(

    Hopefully a mate of mine is letting me borrow his 5.2 CD tomorrow so then I'll upgrade.

    Pity there's no kernel-2.2 directory in the 5.1 updates area on ftp.redhat.com :(

  • Out of curiosity, is eth0 an eepro100 card?

    Does doing `ifconfig eth0 promisc' make the problem go away?
  • It looks like it's got ipchains, which I don't think was in the regular disto. Also has updated samba (2.0) in case you want it.
    I've been using 2.2 at home since the release on my 5.2 with no ill effects.
  • Posted by skitzo:

    Welp, I have this system which has Red Hat 5.1 installed, and the most upgrading I've done is to take the kernel up a step from 2.0.35 -> 2.0.36. What other shtuff do I gotta do to take it to 2.2? Is there a list of software I hafta download or is it just the RPMS from RH [redhat.com]? Thanks...
  • 5:40pm up 8 days, 51 min, 5 users, load average: 1.26, 1.35, 1.30

    Linux obsidian2 2.2.1 #4 Thu Jan 28 20:28:30 EST 1999 i586 unknown

    running Red Hat 5.2. not to sound arrogant, but the only reason i've had to reboot was because i hooked up a SCSI CD-R drive. also, i run a heavily modified version of Red Hat, with CVS code flying around everywhere..
  • I was getting some updates last night from my favourite local mirror when I saw this directory there. So stop hammering updates.redhat.com and get thee to thy local mirror!
  • contrib.redhat.com contains RPMS submitted my lots of people in order to make up for the deficiency in RPMS supported by RedHat. If you were using Slackware you wouldn't wait for a program to be included in the distribution, you would just go get the source. If even contrib doesn't have an RPM then make one yourself and submit it, that way other people can use your work of compiling it to save them time.

    I've been using RedHat since 4.0 on multiple computers and I have never unintentionally destroyed my libraries with an RPM, not once. Maybe it's because I'm very careful about using --force, or maybe it's because I resist the temptation to just make and install stuff without using RPMs so I don't have noncontrolled files sitting around (except in /usr/local and /opt, which no RPM should touch), but I've never noticed a problem with this.

    Please don't think that I'm an RPM bigot. It definately has it's shortcomings, and they get very noticable, very quickly on a large installation (lots of NFS storage starts making problems with putting way too much on the /usr drive), but I definately prefer RPMs to compiling and installing. I've never looked at the Debian nor the Stampede package management system, so I don't know how good they are comparatively (nor how many packages are available in them). There is no reason to repeat work that someone else has done for you.


  • Only use -9 as a last resort. The -9 flag cannot be trapped by the process, which means it can't clean up after itself.

    A simple "kill" will suffice (I believe it sends a TERM(inate) signal by default, which is friendly), and that's what's used on my system to drop the connection. It's worked flawlessly under 5.0 and 5.2 with 2.2.1. Pppd catches the TERM signal and shuts down the connection in a friendly fashion.

    Someone else mentioned "ifdown ppp0". Perhaps that's the "correct" way of doing it; I don't know.
  • By "stable" of course I assumed you meant applications ran on it without problems or library incompatibilities.

    As far as *stability* goes (i.e. crashing), Red Hat's never crashed on me once (when I wasn't using an early 2.1 kernel). Applications that failed to work with glibc were upgraded (the "correct" way to solve this problem), and everything works perfectly now.

  • While I don't like it, I have to say (Score:3)

    by BadlandZ ( 1725 ) on Tuesday February 16, 1999 @04:28AM (#2013987) Journal
    While I personally prefer to compile things on my own, I do believe that RPM's for things like a kernel make remote administration much easier for some people (eg multiple servers, some of which have low cpu power despite high bandwidth). Seems that the RPM's are intended to make things more "2.2.x" friendly, but, they don't have a generic 2.2.x kernel in RPM yet. Anyone know why?

    I have rpm'ed a few 2.0.x kernels in for upgrades on older boxes (like a 386 that takes 18 hours to compile a kernel on, but rpm's in in 2 minutes), so I do understand thier value. So, I guess I want to know, will the new rpm's break anything in 2.0.x systems, and if they have rpm's to upgrade the stuff needed for 2.2.x, why don't they have a 2.2.x kernel in .rpm?

    I do applaud Red Hat for so far waiting, and not jumping to release a 6.0 version with 2.2.x kernel. If they can hold just a little longer, they might be able to use glibc 2.1, gnome 1.0, and kernel 2.2.x... If they do widespread testing before distributing it, it might be cool. Of course, this is going to scream "buggy" despite any amount of beta testing, due to the "new-ness" of so many of the parts... But, wooo... come 6.1, it will probably really rock (if they wait like I hope for 6.0).

  • "rawhide a couple of days ago..."?

    I haven't been able to get into rawhide.redhat.com for more than a week. I get "ICMP 13 Unreachable from gateway router.redhat.com (199.183.24.225) for icmp ..." How did you get in?

  • 2:04am up 56 days, 11:19, 4 users, load average: 0.00, 0.00, 0.00

    That's my uptime running redhat 5.2 The only reason that machine has ever gone down was because of power outages.
    Anyways, I think this anonymous coward is probably one of the many people in the linux community that think poorly of redhat because it has made linux easy (yet still stable and secure) enough for even a windows95 user.

    If these people would take the time to examine the redhat distribution you would find a solid distribution of packages.
    Joseph Elwell.
  • I had this problem when I ran netatalk on
    an EtherExpress Pro 100 network card.
    I fixed it by changing the line
    max_multicast_filter = 64
    (in drivers/net/eepro100.c), to
    max_multicast_filter = 3
    (as suggested by someone on netatalk-admin)
    Now it works fine!
    (Apparently you can also do this by passing
    an argument max_multicast_filter=3 when you
    load the driver module.

    Danny.
  • I believe Redhat has said that 6.0 will definitely have glibc 2.1, gnome 1.0 and a 2.2 kernel in it
  • I had that problem as well. It was a very silly and stupid problem. I accidentally made the main ATAPI/IDE function a module because it let me. I was then unable to mount root. So I read the help for that, and it explains that you can't modularize that option if your root filesystem is on an IDE drive. Make it Y, recompile, reboot, and it worked for me. If this is not the problem, can't help ya...
    -mickey
  • Actually, it's real easy to get from RedHat 2+
    to RedHat 5.2... I've been running 5.1 on my
    laptop for a while (after running 4.2 and maybe
    5.0), and just upgraded to 5.2.

    I'd tended toward complete "clean" reinstalls,
    since I was upgrading the drive, repartitioning,
    or things like that, but the hardware end of
    things is now pretty stable... okay, VERY stable,
    so I tried the "Upgrade" function. I basically
    told the laptop what its IP address was, and
    where the FTP server was, and went and worked
    on other stuff for an hour, and wham-o, I've
    got a 5.2 system.

    I'm a little more wary of dropping kernel RPM's
    in, but I might try it while I'm at it... :)

  • Woo hoo! (Score:1)

    by emag ( 4640 )
    Funny how just last night I started an FTP of kernel 2.2.1...

    On a related note, can anyone give a brief synopsis of what the patches @ www.kerneli.org [kerneli.org] add to the kernel?
  • Any word on whether the RPMs are sufficient to bring RH 5.0 up to speed? I rather expect they would be, but I'd prefer to hear from someone "in the know."
    Thanks, cb

  • (d'oh! previous attempt got munged)

    Problems running 2.2 on RH5.2? Buh? I never had a wink of a problem...well, other than needing to hypermodularize everything. Such a vast range of support offered that it's tempting to dig up old hardware and spider it off of my workstation, just to watch it run for the first time in years. :)

  • I bet that the executables are in different directories. Check /usr/bin and /usr/sbin. If they're there, either move/copy/symlink them to where they are expected to be.

    -Dave
  • 11:54pm up 168 days, 13:40, 1 user, load average: 0.12, 0.04, 0.01

    This is my RedHat5.1 (with 2.0.35), quite stable I think...

  • Anyone who has experienced problems with the sound card on kernel 2.2.0 & 2.2.1?

    I have the most odd problem. The sound card (SBAWE32) works fine as long as I either use my tv-card or play cd music. However when I really start using to soundcard (e.g playing mp3) then the whole computer crashes and the only thing to do is to press the reset button. This is quite annoying because it means no mp3...buhuuu!
    (There are no memory/irq conflicts)

    Someone told me that there was a bug somewhere in the kernel soundblaster code which (one some computers) caused the soundcard to generate thousands of interrupt which caused the processor to crash. Anyone know if this is true - or if there are some other explanation for this?

    Just as a final comment. There were no problems whatsoever on kernel 2.0.34. From 2.0.36 they started to appear.

    I know this is a bit off topic..:)
  • I couldn't get anything to compile with my 2.2.1
    kernel until I updated my C library to the latest
    GLibC. When I did that, StarOffice stopped working, and I had some other miscellaneous problems. These are all known problems, I believe, but does anybody have any workarounds? Anybody got 2.2.1 working (and stuff compiling) with libc 2.0.7, or have a workaround to make StarOffice work with the new GLibC?

    Thanks for any pointers,

    -----Scott.
  • Once upon a time, after getting fed up with my first distribution (slack 3.3 -- so maybe I am a youngin'!), I switched over to RH5.1 thinking RPMs would make my life generally easier. After a few months, I have come to realize, they are a PAIN IN THE POSTERIOR. I have nothing against the use of RPMs for some things, but if you can upgrade your kernel without RPMs then you should probally learn to upgrade its required programs without RPMs too. Not that RPMs don't work just fine most of the time, but it is a pain having to wait on RedHat to release seemingly vital components a few weeks after the new kernel is out. Plus, using precompiled binaries is sometimes a risky situation. (Lord knows how many times I destroyed my dynamically linked libraries with RPMs before weening myself off of them...)
  • I'm almost happy about this, being a redhat user. a redhat 5.1 user... argh... I tried to run an ftp install/upgrade, and it wouldn't work, as always. Why is that? I don't know.
    So, I'll be waiting for 6.0 and upgrade from there, since there's no way I'm going to buy the 5.2 CD.
  • anybody notice that the new util-linux is missing /bin/mount, umount, swapon, and swapoff?

    i was running util-linux-rhcn-2.9h-1 and made the mistake of just uninstalling it and installing the new one. now i can't get past single user mode because i have no mount. what's the easiest way out of this predicament (you can tell i'm no expert)?

    tia,
    marc
  • I get the same problem between a 2.2.1 box running 2.0.0 and a 2.0.36 running 1.9. something. The older version mounts 2.2.0 fine but not vice versa. No matter which version of samba I'm running under kernel 2.2.1, I can't mount samba shares (although I can mount windows shares).
  • Ahum.. Wsn't it something like the Samba team tried to fix a not-too-lethal bug from 2.0.0 to 2.0.1 and ended up with an even more lethal bug, hence "brown paper bag release" 2.0.2 ?

    Perhaps you might want to hold on a bit concerning the Samba upgrade from these RPM's..?
  • At least initscripts can't be upgraded; I tried to upgrade my 5.1 and it said 'redhat-release =5.1 conflicts with initscripts-yaddayaddayadda'.

    As for the rest; I guess they can. Try it out - it'll tell ya it it won't cut it.
  • It's no big issue keeping a 5.0 up to speed - upgrading the RPMS as necessary will keep it alive & happy for a long time (I'm running 5.0 on my workserver at home, but with _lots_ of rpm upgrades).

    I would like to upgrade to a 'native' 5.2 (or 6.0 when it's there), but there's just too darn much reconfiguration for me to care for that on that particular machine :)
  • has anyone tested the 2.2 kernel for linux sparc?
    Last time It wouldnt compile because of some syntax problems.

  • What I was hoping to find here would be RPMs of the kernel itself. I installed all the RPMs from the tango project, then tried make menuconfig and so forth with 2.2.1, and when I rebooted, my system wouldn't even boot...and when I tried to go back to the old kernel, I found it had munged my modules directory somehow to the point where I had to do a full reinstall before I could get back into 2.0.36!

    I really would like to upgrade, so I can maybe get my sound card working (it's a SB Live), but I've never had any success with a kernel upgrade any time I've tried...if anyone can help me, please let me know?
  • Are these rpms required? I am running 2.2.1 on a number of RedHat 5.2 boxes with no problems whatsoever. Are there any docs that describe exactly what these rpms are supposed to do? I don't like the idea of simply updating things with no indication of what the update is supposed to do (other than the fact that the filename references the program being updated)
  • I haven't upgraded to 2.2, but I did notice that in the SRPMS ftp directory, there is an update to dhcpcd. You might want to try upgrading that, and probably the rest of the packages there too.
    ----------
  • Though I don't have as impressive uptimes as the previous person (mine is only 27 days), I can tell you that I've used Windows 98 for the past three hours, and I've had to reboot FOUR times! I've never had my Linux box tell me to reboot, and the only times I do are when I rebuild and reinstall the kernel. If I change simple network or multimedia settings under Windows it forces me to restart everything, which even on a P2-300 takes forever.

    Linux on my old P-90 may take a long time to boot up, but I only have to do it once a month.
  • I have a RedHat 5.0 box IP Masquerading my DHCP cable modem to several Windoze boxes. It uses 2 3Com 3C503 combos. I updated everything on the box to RedHat 5.2 using RPMS and then used the Russian Tango RPMS that were posted here last week and compiled 2.2.1 from kernel.org. Everything seemed to work fine until it tried to find eth1. It gave some errors about being unable to find the module (which is the same as eth0) ... I backed out to 2.0.36 from RedHat 5.2. Has anyone seen anything similar to this?
  • Greetings,
    When 2.2.1 came out I'd JUST installed my first post 0.12 version of Linux off of a RedHat 5.2 CD. So I followed the directions, got all the necessary upgrades as per the upgrade pages, recompiled the kernel, upgraded everything to a new glibc, etc., etc... And Samba stopped working.

    I had gotten it configured and working perfectly under the stock RH5.2 config, my home NT box was seeing itself as part of the Linux machines network, the two were mounting each other in a veritable orgy of internetworking, and all was right with the internal net.

    Now I get a mount version error...? (Says it needs Mount version 6 or something? Linux box is at home, I'm at work.) I upgraded to Samba 2.0.2 and it didn't fix it. My *GUESS* is that something changed in an internal version number in the kernel and something of mine didn't get upgraded... (OR the default 'make install' type command for Samba 2.0.2 doesn't install into the same spot as the RH5.2 rpms...!)

    Any clues from the clueful slashdotters? For what it's worth, I can use the SMBFTP-type program to connect between them and transfer files, but that's just UnPleasant compared to transparent file communication.

    Thanks muchly!

    Cyberfox

    p.s. To those who would say 'Ditch NT, go to Linux only!' Yes, I need NT at home. How ELSE can I develop Java 1.2 applications right now...? C'mon Blackdown, get on the case! If I had a 1.2 JVM, I'd be that hair away from being all Linux.
  • Greetings,
    D'oh! My problem exactly! Please tell me if you find a solution to that... Using the ftp-ish Samba client to transfer files bites! (Geez, you'd think they'd at LEAST include 'readline' support in it!)

    Cyberfox

  • Summary: Soundblaster Live does not work in Linux

    I accidently bought an SB Live since I'd grown so used to Linux supporting most common PC hardware. Mistake.

    The last I heard (and I can't recall the source, so take this with a large grain of salt...) was that Creative Labs wasn't releasing the device specification right away.

    The OSS [opensound.com] folks claimed 4Q99 support when I tried to install the eval version of OSS.

    Since this was a bit longer than I wanted to wait, I went out and purchased a Sound Blaster PCI128, which works great with the es1370 driver included with the 2.2.1 kernel. I just had to hack the /etc/conf.modules a bit to get it to autoload properly.

    Given Creative Labs' spotty history of support for Linux, I'm strongly tempted to abandon my 8-year-old brand loyalty. I'll have to see how the Live support comes.

  • You learned your lesson the same way I did. I always compile at least one monolithic (no modules) kernel and keep it in lilo.conf forever. That way my system will be bootable even if I hose up my modules. Again. ;->

  • I have been running the two together for a week or so with no problems at all. No tweaks, no problems or esoteric reconfiguring whatsoever. I will download the RPMs anywho as it prolly won't hurt.
  • agreed - indiscriminate use of rpm can hose things but I don't consider that rpm's fault. Like any tool it needs to be used properly.

    I see more danger in mixing tarball-compiled stuff in the filesystem and not respecting the proper guidelines: non-rpm stuff should be installed under /usr/local. (Rpms will not install there unless they have been created by someone who doesn't follow the rules) The main reason for this is because rpm has no knowledge of it's existance. Hence it can be easily clobbered by a package...

    For what it's worth, I am a big proponent of source rpms - you can upgrade them fairly painlessly in most cases, they can be built on the target system, and they are a neat way to distribute source.
  • I upgraded 5.2 with all packages suggested on www.linuxhq.com. After recompiling kernel 2.2.1 I get VFS error message "can't mount root filesystem". Has anyone else had this problem?
  • I had a similar problem. In my case, when I played mp3's, my SCSI CD-R drive went crazy. It turns out there was a memory conflict between my Plug 'n' Pray SB64 and my ethernet card (don't ask me how my CD-R drive got envolved, I don't really have any idea). Anyway, I too was absolutely certain that there were no io/irq conflicts, until I logged in at console and tried using mpg123 (I think I actually logged in as root), and I got this bizarre console message indicating that there was a problem. Anyway, you might want to double and triple check for io conflicts. Note I had the same problem under both '95 and NT at one point or another.
  • I BELIEVE I read on Kernel Traffic that they'd identified this as a problem, and had a fix in 2.2.2pre4.

    Hope that helps..

    Orv - W6BI -

Slashdot Top Deals

How many hardware guys does it take to change a light bulb? "Well the diagnostics say it's fine buddy, so it's a software problem."

Close