Rock solid: NOT if linux freezes daily in real world!

Rock solid: NOT if linux freezes daily in real world!

Post by Yan Xi » Mon, 28 Aug 1995 04:00:00



Just we are gaining confidence at the Linux as
a "real" OS, we are recently experiencing OS frozen
on a daily basis.
(Isn't it ludicrous to worry about uptime being too long!!)

We need your help to find out why our Linux box crashes.


==================================
Symptons:  won't accept any input (even contrl-alt-delete).
When:   1) sometimes a http connection
        2) sometimes when key pressed on the console
        3) about once a couple of days.  Occasionally more frequent
Debugging effort so far:
        1) checking /var/adm/debug:  nothing found unusual
        2) checking /usr/local/etc/httpd/logs/access_log: often
           a connection just prior to the time of crashes.
        3) no file loss, yet.

System: 1) Linux 1.2.12
        2) modules: 3c509, dummy
        3) NCSA httpd version NCSA/1.4.1, with patch to run multiple servers
        4) The machine functions as two IP hosts (through dummy module).
        5) Hardware: 3Com509/32 MB RAM/486 DX from Dell

History: used to run 1.2.0, no IP aliasing,
         with a record of 90 days of uptime (no freezes/crashes whatsoever)

Hypotheses of reasons of crash:
        1) Faulty kernel compilation
        2) Buggy dummy module
        3) Kernel incapable of handling high network traffic
           (no local bridge, machine connected directly
           to backbone with network mask: 255.255.0.0)
        4) httpd bad
        5) dummy driver mis-installation

Help wanted:
        finding out the cause(s) to system freeze, and solutions

 
 
 

Rock solid: NOT if linux freezes daily in real world!

Post by William E. Powe » Mon, 28 Aug 1995 04:00:00


: Just we are gaining confidence at the Linux as
: a "real" OS, we are recently experiencing OS frozen
: on a daily basis.
: (Isn't it ludicrous to worry about uptime being too long!!)

: We need your help to find out why our Linux box crashes.

Just a suggestion, but try going back to your earlier kernel and see if the
problem persists.  Or try a distribution kernel and see if that exhibits the
same problem.  Isolate the kernel from the rest of the system.  If the kernel
is the problem, back up to 1.2.8 or 1.2.11 (which have worked well for me.)



 
 
 

Rock solid: NOT if linux freezes daily in real world!

Post by Mark Swans » Tue, 29 Aug 1995 04:00:00


I can't run anything higher than 1.2.8.  1.2.12 seemed to crash for me too.
All the 1.3.x kernels up to 1.3.14 crashed for me too...:-(
I'm sticking with 1.2.8 untill a little later in the 1.3.xx series .

: Just we are gaining confidence at the Linux as
: a "real" OS, we are recently experiencing OS frozen
: on a daily basis.
: (Isn't it ludicrous to worry about uptime being too long!!)

: We need your help to find out why our Linux box crashes.


: ==================================
: Symptons:  won't accept any input (even contrl-alt-delete).
: When: 1) sometimes a http connection
:       2) sometimes when key pressed on the console
:       3) about once a couple of days.  Occasionally more frequent
: Debugging effort so far:
:       1) checking /var/adm/debug:  nothing found unusual
:       2) checking /usr/local/etc/httpd/logs/access_log: often
:          a connection just prior to the time of crashes.
:       3) no file loss, yet.

: System:       1) Linux 1.2.12
:       2) modules: 3c509, dummy
:       3) NCSA httpd version NCSA/1.4.1, with patch to run multiple servers
:       4) The machine functions as two IP hosts (through dummy module).
:       5) Hardware: 3Com509/32 MB RAM/486 DX from Dell
:      
: History: used to run 1.2.0, no IP aliasing,
:        with a record of 90 days of uptime (no freezes/crashes whatsoever)

: Hypotheses of reasons of crash:
:       1) Faulty kernel compilation
:       2) Buggy dummy module
:       3) Kernel incapable of handling high network traffic
:          (no local bridge, machine connected directly
:          to backbone with network mask: 255.255.0.0)
:       4) httpd bad
:       5) dummy driver mis-installation

: Help wanted:
:       finding out the cause(s) to system freeze, and solutions
:        

 
 
 

Rock solid: NOT if linux freezes daily in real world!

Post by Perry Donh » Wed, 30 Aug 1995 04:00:00



    William> gaining confidence at the Linux as : a "real" OS, we are
    William> recently experiencing OS frozen : on a daily basis.  :
    William> (Isn't it ludicrous to worry about uptime being too
    William> long!!)

    William> : We need your help to find out why our Linux box
    William> crashes.

    William> Just a suggestion, but try going back to your earlier
    William> kernel and see if the problem persists.  Or try a
    William> distribution kernel and see if that exhibits the same
    William> problem.  Isolate the kernel from the rest of the system.
    William> If the kernel is the problem, back up to 1.2.8 or 1.2.11
    William> (which have worked well for me.)

I experienced similar problems until I disabled the level 2 CPU cache
RAM. Symptoms included apparently random system hangs and segment
violations. The system would not stay up for more than a day or so
without*. After disabling cache, the machine has been running
nearly two months without even a murmur.

Perry

--
----------------------------------------------------------------------

                      PGP encrypted mail welcome
            Visit KidPub!  http://www.veryComputer.com/

 
 
 

Rock solid: NOT if linux freezes daily in real world!

Post by Dave Fenwic » Wed, 30 Aug 1995 04:00:00




>I can't run anything higher than 1.2.8.  1.2.12 seemed to crash for me too.
>All the 1.3.x kernels up to 1.3.14 crashed for me too...:-(
>I'm sticking with 1.2.8 untill a little later in the 1.3.xx series .

>: Just we are gaining confidence at the Linux as
>: a "real" OS, we are recently experiencing OS frozen
>: on a daily basis.
>: (Isn't it ludicrous to worry about uptime being too long!!)

>: We need your help to find out why our Linux box crashes.


>: ==================================
>: Symptons:  won't accept any input (even contrl-alt-delete).
>: When:     1) sometimes a http connection
>:   2) sometimes when key pressed on the console
>:   3) about once a couple of days.  Occasionally more frequent
>: Debugging effort so far:
>:   1) checking /var/adm/debug:  nothing found unusual
>:   2) checking /usr/local/etc/httpd/logs/access_log: often
>:      a connection just prior to the time of crashes.
>:   3) no file loss, yet.

>: System:   1) Linux 1.2.12
>:   2) modules: 3c509, dummy
>:   3) NCSA httpd version NCSA/1.4.1, with patch to run multiple servers
>:   4) The machine functions as two IP hosts (through dummy module).
>:   5) Hardware: 3Com509/32 MB RAM/486 DX from Dell
>:  
>: History: used to run 1.2.0, no IP aliasing,
>:    with a record of 90 days of uptime (no freezes/crashes whatsoever)

>: Hypotheses of reasons of crash:
>:   1) Faulty kernel compilation
>:   2) Buggy dummy module
>:   3) Kernel incapable of handling high network traffic
>:      (no local bridge, machine connected directly
>:      to backbone with network mask: 255.255.0.0)
>:   4) httpd bad
>:   5) dummy driver mis-installation

>: Help wanted:
>:   finding out the cause(s) to system freeze, and solutions
>:    

Until I recently upgraded (ahem) to  the 1.3.x kernel, I regularly had
uptimes on my 1.2.1 box of 30-50 days.  Now, I can't get the ethernet
interface on my new box to be recognized in the 1.3.x kernels.  I just
upgraded my hardware to a Pentium 100 from Gateway with an SMC EtherPower
(Tulip) card, and it just gives me bogus SIOCSIFFLAG errors.  Can't figure
it out unless it's a mixture of old and new include files.  I have noticed
that the kernel will include files from the /usr/include directory where
files in the $DIST/include/linux directory should be used.  After a lot of
painful reconfiguration, I am currently recompiling my kernel with the files
that I believe are the right include files to see if the problem goes away.
If not, then I have to  go back to the drawing board and see if I can get
anything to work.

I ABSOLUTELY need the Adaptec 2940 driver.  I now have a Pentium and a PCI
SCSI controller with the old disk drives.  I could probably live with the
old ISA ethernet card, but that doesn't work either.

They should build some type of include file updater in the kernel installation.
It would make it so much easier to have the new files automagically updated.

--

Senior Systems Engineer                           (202) 761-4442 x119
Simulation Technologies, Inc.                      FAX (202) 761-5352

 
 
 

Rock solid: NOT if linux freezes daily in real world!

Post by Daniel Panca » Thu, 31 Aug 1995 04:00:00




>I can't run anything higher than 1.2.8.  1.2.12 seemed to crash for me too.
>All the 1.3.x kernels up to 1.3.14 crashed for me too...:-(
>I'm sticking with 1.2.8 untill a little later in the 1.3.xx series .

Move past 1.2.8 to at least 1.2.9.  1.2.9 fixed the Sudden Death Syndrome
problem.

Dan

--------------------------------------------------------------------------
Daniel Pancamo      (713)286-0399 voice       InfoCom Networks (ISP)

19.95 Month PPP     http://www.infocom.net    Houston, Texas 77259-0343

 
 
 

Rock solid: NOT if linux freezes daily in real world!

Post by Steffen Muell » Thu, 31 Aug 1995 04:00:00




: (Tulip) card, and it just gives me bogus SIOCSIFFLAG errors.  Can't figure
: it out unless it's a mixture of old and new include files.  I have noticed
: that the kernel will include files from the /usr/include directory where
: files in the $DIST/include/linux directory should be used.  After a lot of
: painful reconfiguration, I am currently recompiling my kernel with the files
: that I believe are the right include files to see if the problem goes away.
: If not, then I have to  go back to the drawing board and see if I can get
: anything to work.

This is a problem many of us also encountered. Alax Cox gave an answer immediately concerning problem with the include files where illegal redifinitions may occur.

I believe the 1.3 kernel tree really isn't the way to go in case you need a stable system. As far as I can remember the 2940 patches also apply to 1.2 production kernels.

--

NTG Netzwerk und Telematic GmbH  \   /        fax:   +49 721 9652 210
Geschaeftsbereich Xlink           \/          phone: +49 721 9652 211
Vincenz-Priessnitz-Str. 3         /\ L I N K  http://wwwi.xlink.net/~steffen/
D-76131 Karlsruhe, Germany       /__________

 
 
 

Rock solid: NOT if linux freezes daily in real world!

Post by Alan C » Thu, 31 Aug 1995 04:00:00



>upgraded my hardware to a Pentium 100 from Gateway with an SMC EtherPower
>(Tulip) card, and it just gives me bogus SIOCSIFFLAG errors.  Can't figure

How do you know they are bogus, you dont even say what they are. Bet its
because you have the card on an IRQ already in use.

Quote:>it out unless it's a mixture of old and new include files.  I have noticed
>that the kernel will include files from the /usr/include directory where
>files in the $DIST/include/linux directory should be used.  After a lot of
>painful reconfiguration, I am currently recompiling my kernel with the files
>that I believe are the right include files to see if the problem goes away.
>If not, then I have to  go back to the drawing board and see if I can get
>anything to work.

The /usr/include files are not used for the kernel building. They
/usr/include/linux files are used so programs accessing kernel features
directly get the right structures. This external interface changes very
rarely.

Quote:>They should build some type of include file updater in the kernel installation.
>It would make it so much easier to have the new files automagically updated.

The kernel comes with all its include files.

Alan

--
  ..-----------,,----------------------------,,----------------------------,,

Redistribution of this message via the Microsoft Network is prohibited
<A href="file:/dev/mouse">Click here to disable mouse.</A>

 
 
 

Rock solid: NOT if linux freezes daily in real world!

Post by bill davids » Thu, 31 Aug 1995 04:00:00



| I experienced similar problems until I disabled the level 2 CPU cache
| RAM. Symptoms included apparently random system hangs and segment
| violations. The system would not stay up for more than a day or so
| without*. After disabling cache, the machine has been running
| nearly two months without even a murmur.

This is a useful diagnostic tool, but the idea of running production
in that mode strikes me like taking a wheel of a car if the tire is
flat.

On my 486/P5 systems I see about a 30% drop in throughput when L2 is
off, and I've never run production that way beyond the time to get a
repair.

 
 
 

Rock solid: NOT if linux freezes daily in real world!

Post by Martin Viss » Fri, 01 Sep 1995 04:00:00



> Until I recently upgraded (ahem) to  the 1.3.x kernel, I regularly had
> uptimes on my 1.2.1 box of 30-50 days.  Now, I can't get the ethernet
> interface on my new box to be recognized in the 1.3.x kernels.  I just
> upgraded my hardware to a Pentium 100 from Gateway with an SMC EtherPower
> (Tulip) card, and it just gives me bogus SIOCSIFFLAG errors.  Can't figure
> it out unless it's a mixture of old and new include files.  I have noticed
> that the kernel will include files from the /usr/include directory where
> files in the $DIST/include/linux directory should be used.  After a lot of
> painful reconfiguration, I am currently recompiling my kernel with the files
> that I believe are the right include files to see if the problem goes away.
> If not, then I have to  go back to the drawing board and see if I can get
> anything to work.

I'm having probs with the new kernel as well. I think they have changed
the IOCTL interfaces , as my old ifconfig, route didn't work. I got the
new net-tools but I am having problems compiling this as some typdefs and
#defines don't appear to be in my inc files. Maybe I need to upgrade libc
as well?

Martin

--
            : Martin Visser
    /\/\    : BHP Steel - Slab & Plate Products Division
   / / /\   : Engineering Technology - Computer Systems
  / / /  \  : PO Box 1854 Wollongong NSW 2500, AUSTRALIA
 / / / /\ \ : A.C.N. 000 019 625
 \ \/ / / / : Phone     +61-42-753852
  \  / / /  : Fax       +61-42-754141


            : WWW   (Work) http://wwwet.sppdwol.bhp.com.au/people/mvisser
            :       (Home) http://www.ozemail.com.au/~mvisser

 
 
 

Rock solid: NOT if linux freezes daily in real world!

Post by Yuan Jia » Mon, 04 Sep 1995 04:00:00


A friend of mine is starting up an ISP business.  I tried to get him
to use Linux because it's "rock solid" and free.  He was not moved
and gave me two reasons: 1) "all the other ISPs" use BSDI and 2)
the Linux software ftp site froze every two weeks.

I've never used BSDI.  Can any one share their experience and opinion
on it?  If "all the other ISPs" use BSDI, there must be a reason.

Indeed, my free software ftp site froze every two weeks until I compiled
a new wu.ftpd.  If an application, ill function maybe, can freeze the
whole system, it makes you doult whether it's "rock solid".

 
 
 

Rock solid: NOT if linux freezes daily in real world!

Post by Jeff Mcfarli » Tue, 05 Sep 1995 04:00:00



>A friend of mine is starting up an ISP business.  I tried to get him
>to use Linux because it's "rock solid" and free.  He was not moved
>and gave me two reasons: 1) "all the other ISPs" use BSDI and 2)
>the Linux software ftp site froze every two weeks.

>I've never used BSDI.  Can any one share their experience and opinion
>on it?  If "all the other ISPs" use BSDI, there must be a reason.

>Indeed, my free software ftp site froze every two weeks until I compiled
>a new wu.ftpd.  If an application, ill function maybe, can freeze the
>whole system, it makes you doult whether it's "rock solid".

I run Linux at home (1.3.13), but I run BSDi 2.0 at work on the servers.
Reasons? Cuz it's more stable, faster, and other reasons. Other than that I
don't want to get into a OS bashing war cuz I love both OS's for different
reasons. :-)

_,.-=~'`^`'~=-.,__,.-=~'`^`'~=-.,__,.-=~'`^`'~=-.,__,.-=~'`^`'~=-.,__,.
| Jeff Mcfarlin             |         Finger for Public PGP Key       |

| http://galaxy.neca.com/   |          saying to place here.'         |
_,.-=~'`^`'~=-.,__,.-=~'`^`'~=-.,__,.-=~'`^`'~=-.,__,.-=~'`^`'~=-.,__,.
|      NECAnet: Eastern Connecticut's Internet Service Provider       |
_,.-=~'`^`'~=-.,__,.-=~'`^`'~=-.,__,.-=~'`^`'~=-.,__,.-=~'`^`'~=-.,__,.

 
 
 

Rock solid: NOT if linux freezes daily in real world!

Post by Nathan Stratt » Thu, 07 Sep 1995 04:00:00


: A friend of mine is starting up an ISP business.  I tried to get him
: to use Linux because it's "rock solid" and free.  He was not moved
: and gave me two reasons: 1) "all the other ISPs" use BSDI and 2)
: the Linux software ftp site froze every two weeks.

I run a NSP, and we have 6 linux servers, we have had no problems with
the OS. We also have used BSDI, but droped it for linux.

: I've never used BSDI.  Can any one share their experience and opinion
: on it?  If "all the other ISPs" use BSDI, there must be a reason.

Run, Linux.

: Indeed, my free software ftp site froze every two weeks until I compiled
: a new wu.ftpd.  If an application, ill function maybe, can freeze the
: whole system, it makes you doult whether it's "rock solid".

--

Nathan Stratton           CEO, NetRail, Inc.    Your Gateway to the World!
---------------------------------------------------------------------------
Phone   (703)524-4800                          NetRail, Inc.
Fax     (703)534-5033                          2007 N. 15 St. Suite B-5

WWW     http://www.netrail.net/                Access: (703) 524-4802 guest
---------------------------------------------------------------------------

 
 
 

Rock solid: NOT if linux freezes daily in real world!

Post by Michael Dill » Mon, 11 Sep 1995 04:00:00




>A friend of mine is starting up an ISP business.  I tried to get him
>to use Linux because it's "rock solid" and free.  He was not moved
>and gave me two reasons: 1) "all the other ISPs" use BSDI and 2)
>the Linux software ftp site froze every two weeks.

Lots of ISP's use Linux. Send a subscribe request to

cheap NE2000 clone network cards. Get a 3com 3C509B and you won't have
the problem.

Quote:>I've never used BSDI.  Can any one share their experience and opinion
>on it?  If "all the other ISPs" use BSDI, there must be a reason.

We are an ISP. We used BSDI. We no longer use BSDI but we continue to use
Linux. BSDI had a great advertising campaign for Internet servers that
got them a lot of ISP customers, that's why so many ISP's use BSDI but
most ISP's use Linux for some jobs if not for everything.

--
Michael Dillon                                    Voice: +1-604-546-8022
Memra Software Inc.                                 Fax: +1-604-542-4130

 
 
 

1. Rock-solid modem for use with Linux, suggestions?

I want to set up my Linux box, which is currently connected to the
network, so that users have dial-in access to it and ppp services.  This
will be a single phone line.  I've lurked on this newsgroup on and off and
know that a common theme is "help, modem model xyz doesn't work right with
Linux".  I also gather that, in general, external modems are preferable to
internal modems as all the handshaking signals are present and work in a
stable fashion if the serial cable is correct.

Has anyone out there found a 28.8 modem that they were exceptionally
pleased with in terms of compatibility with Linux, stability, line noise
performance, etc.?    

Andy

2. AIX on Intel x86?

3. Cheap rock solid video board

4. automatisches Login mit NET.EXE ??

5. Needed - rock solid Netscape

6. ISDN for Linux (again)

7. Kill your Mag DX17: rock solid 1280x1024 with ATI Mach64

8. Very strange problem with bash in an xterm

9. Linux still rocking the business world.

10. bttv driver freezes Linux solid

11. Linux freezing solid while compiling

12. Another example of Microsoft not living in the real world:

13. Real World Linux Security