Linux unstability

Linux unstability

Post by Marc Jord » Wed, 06 Feb 2002 13:07:22



Okay, let's start : I believe that Linux is the
more stable OS for my x86 architecture out there,
and based on this, I can suppose it is equally
stable on another architectures too.

   So this said, software, given its nature, tends
to be unstable at least during some of its evolution
towards a mature product. So does it means that if
I use Linux all my software will run perfectly ?
Of course it does not mean this. We all know about
Segmentation Faults, coredumps and even the crontab
to delete old core's from the hard disk.

   Then why do I like Linux and prefer it over another
OS's (aka Windows flavors) ? Well, first of all,
developers tend to label 1.0 versions when they've
tested them as much as possible, and new versions do
in fact try to correct old bugs without redoing things
from the start and introducing a lot of new ones.

   And second and very important, a single process or
program failure NEVER leads the system unstable at all,
you can leave it there or kill (sometimes needed due
to CPU overload) it, but the whole system will keep on
working until eternity. So sometimes I have heard
comparations between Linux core and segfaults and Windows
blue screends and unexpected errors : well, these last
ones ALWAYS leads the system to a unstable state, that
before or late will make you reboot the machine (if it
is even able to do this correctly). In fact, these kind
of error under Windows (any version) are just a polite
form of telling you : please, try to reboot and good luck,
'cause we are *ed and we don't know why the Hell.

 
 
 

Linux unstability

Post by David Edne » Wed, 06 Feb 2002 13:45:32




Quote:>    And second and very important, a single process or
> program failure NEVER leads the system unstable at all, you can leave it
> there or kill (sometimes needed due to CPU overload) it, but the whole
> system will keep on working until eternity.

I'll be perfectly honest, and admit that this isn't 100% true.  I've had
linux freeze up on me, but it's extremely rare, nothing like Windows.  In
fact, basically every single time which it has frozen has been while
emulating windows :) (well, technically not emulating, but WINE anyway)

--
GCS/J d s+:-- a--- C++++ UL++++S+ P+ L+++(++)>+++++$ E---- W++ N+ o? K?
w+>--- O !M-- V-- PS+(-) PE- Y-- PGP- t+>++ 5>+ X R>+ tv+ b++(+++)
DI++++ D++ G e>+++ h! r-- y?

 
 
 

Linux unstability

Post by Adam Warne » Wed, 06 Feb 2002 14:17:53



>    And second and very important, a single process or
> program failure NEVER leads the system unstable at all, you can leave it
> there or kill (sometimes needed due to CPU overload) it, but the whole
> system will keep on working until eternity.

Unfortunately some people are experiencing significant virtual memory
problems at the moment Marc. This thread's an interesting read:

http://groups.google.com/groups?hl=en&th=684ed074e42abca&rnum=1

I've not experienced any of this but I'm not pounding my boxes hard
either. It seems clear that Alan Cox is rather annoyed with the
whole situation.

I wonder if Marcelo is considering swapping the VM system back to
Rik's model?

It's a telling indictment when Alan writes:

   The only kernels you are likely to see that not happen on are

   -    The 2.4.9 kernel with Rik's patches that Linus didnt take
        (Red Hat 2.4.9-*)
   -    2.4.17/18pre with the rmap11/rmap12 patches
   -    2.4.17/18pre with the -aa patched VM
        (which I believe is also in the SuSE kernel packages)
   -    2.2

Note the reference to the 2.2 kernel series. 2.2 appears to be a better
workhorse kernel than the current stock 2.4.17 kernel. Let's hope
this is resolved soon.

Regards,
Adam

 
 
 

Linux unstability

Post by Anonymou » Wed, 06 Feb 2002 14:37:48



> Unfortunately some people are experiencing significant virtual memory
> problems at the moment Marc. This thread's an interesting read:

> http://groups.google.com/groups?hl=en&th=684ed074e42abca&rnum=1

> I've not experienced any of this but I'm not pounding my boxes hard
> either. It seems clear that Alan Cox is rather annoyed with the
> whole situation.

I'm having a similar problem with lots of swapping.
It's very irritating, practically the whole system
freezes up until the jobs are done.

I'm using kernel 2.4.7-10.

-----=  Posted via Newsfeeds.Com, Uncensored Usenet News  =-----
http://www.newsfeeds.com - The #1 Newsgroup Service in the World!
 Check out our new Unlimited Server. No Download or Time Limits!
-----==  Over 80,000 Newsgroups - 19 Different Servers!  ==-----

 
 
 

Linux unstability

Post by Adam Warne » Wed, 06 Feb 2002 14:57:04




>> Unfortunately some people are experiencing significant virtual memory
>> problems at the moment Marc. This thread's an interesting read:

>> http://groups.google.com/groups?hl=en&th=684ed074e42abca&rnum=1

>> I've not experienced any of this but I'm not pounding my boxes hard
>> either. It seems clear that Alan Cox is rather annoyed with the whole
>> situation.

> I'm having a similar problem with lots of swapping. It's very
> irritating, practically the whole system freezes up until the jobs are
> done.

> I'm using kernel 2.4.7-10.

Well you know what to do. Try one of the kernels Alan mentioned. You
might be interested in this recent post:

Date: Mon, 04 Feb 2002
From: Rik van Riel

Due to field circus duty I forgot to announce 12b ... so here is 12c

The third maintenance release of the 12th version of the reverse
mapping based VM is now available.
This is an attempt at making a more robust and flexible VM
subsystem, while cleaning up a lot of code at the same time.
The patch is available from:

           http://surriel.com/patches/2.4/2.4.17-rmap-12c
and        http://linuxvm.bkbits.net/

My big TODO items for a next release are:
  - auto-tuning readahead, readahead per VMA

rmap 12c:
  - fix small balancing bug in page_launder_zone          (Nick Piggin)
  - wakeup_kswapd / wakeup_memwaiters code fix            (Arjan van de Ven)
  - improve RSS limit enforcement                         (me)

...

Regards,
Adam

 
 
 

Linux unstability

Post by Richard Stein » Wed, 06 Feb 2002 14:54:26


Here in comp.os.linux.advocacy,

Quote:>Okay, let's start : I believe that Linux is the
>more stable OS for my x86 architecture out there,
>and based on this, I can suppose it is equally
>stable on another architectures too.

What are you comparing it to?

Subtract the sometimes problematic WorkPlace Shell from OS/2, and I'd
put it against a Linux installation anytime.

Some of the BSD variants also aren't bad, and I suspect that the more
specialized OSes (QNX, etc.) might have a say as well.

There's a lot more to the PC world than Linux and Windows, even though
the press might have you believe otherwise.  :-)

(Your other comments about Linux are good, but many of them are not
unique to Linux, but apply equally well to other well-engineered OSes.)

--
 -Rich Steiner >>>---> http://www.visi.com/~rsteiner >>>---> Eden Prairie, MN
   OS/2 + BeOS + Linux + Solaris + Win95 + WinNT4 + FreeBSD + DOS + PC/GEOS
            + Fusion + vMac + Executor = PC Hobbyist Heaven! :-)
                But what if I'm a figment of MY imagination?

 
 
 

Linux unstability

Post by Marc Jorda » Wed, 06 Feb 2002 14:55:57


   For desktop purposes, I have had no problems until
now (now I am 2.4.17), but I do not use to push my
system to the limit. Production servers that are under
my control do run under 2.2.20 nowadays.

   Anyway, my Windows home machine does show
the behaviour I described without pushing it to the
limit :-)

 
 
 

Linux unstability

Post by Anonymou » Wed, 06 Feb 2002 15:16:57


Can't the virtual memory manager
be turned into a device driver?
This way, you can just use insmod
rmmod to delete the virtual memory manager
and insert a new one.




> >> Unfortunately some people are experiencing significant virtual memory
> >> problems at the moment Marc. This thread's an interesting read:

> >> http://groups.google.com/groups?hl=en&th=684ed074e42abca&rnum=1

> >> I've not experienced any of this but I'm not pounding my boxes hard
> >> either. It seems clear that Alan Cox is rather annoyed with the whole
> >> situation.

> > I'm having a similar problem with lots of swapping. It's very
> > irritating, practically the whole system freezes up until the jobs are
> > done.

> > I'm using kernel 2.4.7-10.

> Well you know what to do. Try one of the kernels Alan mentioned. You
> might be interested in this recent post:

> Date: Mon, 04 Feb 2002
> From: Rik van Riel

> Due to field circus duty I forgot to announce 12b ... so here is 12c

> The third maintenance release of the 12th version of the reverse
> mapping based VM is now available.
> This is an attempt at making a more robust and flexible VM
> subsystem, while cleaning up a lot of code at the same time.
> The patch is available from:

>            http://surriel.com/patches/2.4/2.4.17-rmap-12c
> and        http://linuxvm.bkbits.net/

> My big TODO items for a next release are:
>   - auto-tuning readahead, readahead per VMA

> rmap 12c:
>   - fix small balancing bug in page_launder_zone          (Nick Piggin)
>   - wakeup_kswapd / wakeup_memwaiters code fix            (Arjan van de Ven)
>   - improve RSS limit enforcement                         (me)

> ...

> Regards,
> Adam

-----=  Posted via Newsfeeds.Com, Uncensored Usenet News  =-----
http://www.newsfeeds.com - The #1 Newsgroup Service in the World!
 Check out our new Unlimited Server. No Download or Time Limits!
-----==  Over 80,000 Newsgroups - 19 Different Servers!  ==-----
 
 
 

Linux unstability

Post by rapska » Wed, 06 Feb 2002 16:11:12


Error Log for Tue, 05 Feb 2002 07:17:53 -0500: segfault in module "Adam
Warner" - dump details are as follows...


>>    And second and very important, a single process or
>> program failure NEVER leads the system unstable at all, you can leave
>> it there or kill (sometimes needed due to CPU overload) it, but the
>> whole system will keep on working until eternity.

> Unfortunately some people are experiencing significant virtual memory
> problems at the moment Marc. This thread's an interesting read:

> http://groups.google.com/groups?hl=en&th=684ed074e42abca&rnum=1

> I've not experienced any of this but I'm not pounding my boxes hard
> either. It seems clear that Alan Cox is rather annoyed with the whole
> situation.

> I wonder if Marcelo is considering swapping the VM system back to Rik's
> model?

> It's a telling indictment when Alan writes:

>    The only kernels you are likely to see that not happen on are

>    -       The 2.4.9 kernel with Rik's patches that Linus didnt take
>    (Red Hat 2.4.9-*)
>    -       2.4.17/18pre with the rmap11/rmap12 patches -   2.4.17/18pre with the
>    -aa patched VM
>    (which I believe is also in the SuSE kernel packages)
>    -       2.2

> Note the reference to the 2.2 kernel series. 2.2 appears to be a better
> workhorse kernel than the current stock 2.4.17 kernel. Let's hope this
> is resolved soon.

> Regards,
> Adam

I have experienced this a couple of times on my Enigma box (2.4.10-17)
with a moderate load and uptime, though it seems to stabilize after
seemingly swapping out the entire memory contents.

My Potato router/server (2.2.16) hasn't done this yet and it has far less
resources than the previously mentioned system.

I really don't know if the other two do it as I don't monitor/use them on
a regular basis, though my wife hasn't made a help call since I swapped
her out.

--
rapskat  -  9:05am  up 1 day, 21:54,  5 users,  load average: 0.65, 0.29, 0.27

The game isn't over till it's over. -- Yogi Berra.

 
 
 

Linux unstability

Post by Kenneth Down » Thu, 07 Feb 2002 02:32:00




>>    And second and very important, a single process or
>> program failure NEVER leads the system unstable at all, you can leave it
>> there or kill (sometimes needed due to CPU overload) it, but the whole
>> system will keep on working until eternity.

> Unfortunately some people are experiencing significant virtual memory
> problems at the moment Marc. This thread's an interesting read:

> http://groups.google.com/groups?hl=en&th=684ed074e42abca&rnum=1

> I've not experienced any of this but I'm not pounding my boxes hard
> either. It seems clear that Alan Cox is rather annoyed with the
> whole situation.

> I wonder if Marcelo is considering swapping the VM system back to
> Rik's model?

> It's a telling indictment when Alan writes:

>    The only kernels you are likely to see that not happen on are

>    -  The 2.4.9 kernel with Rik's patches that Linus didnt take
> (Red Hat 2.4.9-*)
>    -  2.4.17/18pre with the rmap11/rmap12 patches
>    -  2.4.17/18pre with the -aa patched VM
> (which I believe is also in the SuSE kernel packages)
>    -  2.2

> Note the reference to the 2.2 kernel series. 2.2 appears to be a better
> workhorse kernel than the current stock 2.4.17 kernel. Let's hope
> this is resolved soon.

> Regards,
> Adam

Check out www.byte.com, there is an interesting article there about
intentionally pushing the kernels until they crack.

--
Ken
oSigBlock = createobject("sigblock")

 
 
 

Linux unstability

Post by Terry Port » Thu, 07 Feb 2002 04:26:05


Marc Jordan went insane and wrote the following
on his bedroom wall with an old crayon:

<snip>

Quote:>   Then why do I like Linux and prefer it over another
>OS's (aka Windows flavors) ? Well, first of all,
>developers tend to label 1.0 versions when they've
>tested them as much as possible, and new versions do
>in fact try to correct old bugs without redoing things
>from the start and introducing a lot of new ones.

<snip>

A nice advocacy piece Marc, and based on this article,
I'd have to say I've been wrong about you being a Wintroll,
and would like to offer you my sincerest apologies.

Welcome to Cola :)

--
Kind Regards from Terry
My Desktop is powered by GNU/Linux. Debian 2.2 kernel 2.2.20  
Free Micro burner: http://w3w.arafuraconnect.com.au/~tp/burn.html          
** Registration Number: 103931,  http://counter.li.org **

 
 
 

1. Unstability

Has anybody observed the following: in a Linux alpha machine (164LX)
used as an X-terminal (XFree SVGA server, now v. 3.3.5, Matrox
Millenium ) to connect to a Linux Intel host, windows randomly
die. Especially, the KDE panel often disappears. There were also
problems with scwm, so I don't think this is KDE-specific. And
everything is stable in another X-terminal running on Linux Intel.  I
don't get any log or error messages. We're running nfs and X
connections in the other direction and they work perfectly.

The alpha uses RH 5.2 originally, now partially upgraded to 6.0. I
have tried various X servers up to the version 3.3.5-9 in RedHat's
rawhide directory, various kernels (now 2.2.14pre3), two compilers for
the kernel (now egcs-2.91.66 19990314/Linux (egcs-1.1.2 release)), and
upgraded libc to RedHat's glibc-2.1.1-7.

Has anybody experienced this? Should I suspect a hardware problem and
start experimenting with various memory combs, network cards etc.?

--
Janne

2. RAMDISK I/O error

3. Causes of unstability under stress

4. Reducing log files

5. 100Mb-TX 3Com-3c905 unstability, x86 Solaris DU7

6. apache 1.2.1: inconsistencies in <Directory>,<Location> etc.

7. comp.os.linux, comp.os.linux.misc, comp.os.linux.security, comp.os.linux.networking

8. catopen/catgets on SuSe6.4 fails

9. uninstall REDhat linux uninstall REDhat linux uninstall REDhat linux uninstall REDhat linux uninstall REDhat linux uninstall REDhat linux uninstall REDhat linux

10. Linux in Linux in Linux in

11. proposal: feed smaller mailing lists into linux-kernel, add linux-kernel-core and linux-kernel-bugs

12. DHCP Linux to Linux problem, Windows to Linux works using same PC!

13. linux ppc vs linux alpha vs linux x86?