*BSD News Article 47646


Return to BSD News archive

Path: sserve!newshost.anu.edu.au!harbinger.cc.monash.edu.au!simtel!swidir.switch.ch!newsfeed.ACO.net!Austria.EU.net!EU.net!howland.reston.ans.net!spool.mu.edu!sdd.hp.com!svc.portal.com!news1.best.com!blob.best.net!not-for-mail
From: dillon@best.com (Matt Dillon)
Newsgroups: comp.unix.bsd.freebsd.misc
Subject: Re: FreebBSD 2.0.5-R crashes every 2 DAYS!!
Date: 23 Jul 1995 12:35:52 -0700
Organization: Best Internet Communications, Inc. (info@best.com)
Lines: 98
Distribution: world
Message-ID: <3uu8ao$4ts@blob.best.net>
References: <3urlmb$7co@ucsbuxb.ucsb.edu> <3us8mc$17j@agate.berkeley.edu>
NNTP-Posting-Host: blob.best.net

:In article <3us8mc$17j@agate.berkeley.edu>,
:Jordan K. Hubbard <jkh@violet.berkeley.edu> wrote:
:>In article <3urlmb$7co@ucsbuxb.ucsb.edu>,
:>Loren Koss <loren@beauty.ucsb.edu> wrote:
:>>to hit ^C, then it will hang after the sendmail.  I have to usually 
:>>reboot three times before it actually gets back to working.
:>
:>You have to reboot 3 times to get it working??  This spells one thing
:>to my mind:  Hardware problems.  It may have run DOS before, but that
:>doesn't prove much as we've already well established that FreeBSD
:>will make mincemeat of marginal hardware where DOS simply doesn't use
:>enough of the machine's capabilities to be worth mentioning (if you
:>don't believe me, take a look at the relative performance of DOS
:>and FreeBSD file I/O).
:> 
:>>P.S. 2.0.5-R isn't very stable.. is it?
:>
:>Well, a number of people have reported excellent stability with it and
:>our incredibly busy ftp.cdrom.com machine runs it with very good
:>reliablity.  This is not to say it's bugless, but stable?  Sure, unless
:>you're beating the heck out of it in interesting ways like BEST seems
:>to be (and we're working very actively with them to fix each problem they
:>find) but they're a fairly rare case.
:>
:>					Jordan

    <GRIN>

    We've seen our share of hardware problems.  I have one suggestion:
    never buy memory without parity.  We had a bad 32MByte Simm
    a few weeks ago which crashed the machine about once a day, and
    if it were not for the parity NMI, we would never have known it
    was the memory.

    Other problems we've seen: disk drives going bad, of course, but
    the most common one seems to be sticky ethernet cards.  Sometimes one
    of the ethernelink III's stops dead in its tracks, solved only by
    ifconfig'ing it down and then up again.  So far I have yet to find
    a card that operates perfectly in the face of constant stress.

    As far as OS bugs go:  Well, all OS's have them.  I have yet to
    find an OS that cannot be crashed.  When you drive a machine into
    the ground, *ALL* OS's have a tendancy to show their worse aspects.
    On the whole, I find commercial OS's such as IRIX and SunOS (to
    name two in my direct experience) to be no more reliable then
    FreeBSD.  That isn't to say that OS's don't have their bad days,
    but on the whole I would take statements such as "go with commercial
    OS *BLAH* rather then this free crap" to be an indication that
    the person making the statement has been brainwashed by the written
    word rather then by any real experience.

    I can tell you, frankly, that I get a whole lot better response from
    the FreeBSD folks then I do with the BSDI or SGI folks.  The BSDI
    folks seem focused on giving me *exactly* the level of technical 
    support I pay for and most of the people I talk to anyway know less
    then I do!  The SGI folks dwell on their trade secrets so much (for
    no good reason, I am not all that impressed by the IRIX core)
    that it is impossible to have a conversation or even get to someone
    who knows the answers to my questions without a lot of hassle.  The 
    SUN folks have been focused on Solaris, which I wouldn't touch with 
    a ten foot pole, and until very recently pretty much ignored the much 
    more stable SunOS, plus the OS is full of security holes.  I spent a 
    week trying to secure a SunOS machine once and a friend stepped in 
    and broke root in less then 10 seconds from an unprivilaged account.  
    Sigh.  Also, unless you pay huge amounts of money, you are stuck with
    releases and updates that are few and far between.  In another
    unrelated project about a year ago, I was forced to drop NextSTEP
    as a viable OS because they never fixed the more serious OS bugs!

    (But people should keep in mind that I am not asking dumb-user questions
    of them either.  An OS-illiterate could very well get better technical
    support service from one of the commercial OS's)

    So, that leaves FreeBSD.  The source code is available and compileable,
    the people responsible for the OS (which is everyone :-)) are amoung
    the best K-Programmers in the world, and so on.  The only fault.. the
    only thing that possibly gives the commercial OS's an edge beyond their
    brandname recognition, is that the commercial OS's have a more 
    formalized release system and maintain a tree for bug fixes to the
    most recent release independant of any new work.  The Linux and
    FreeBSD folk tend to combine current-release bug fixes with new work,
    so one must be careful when one merges in a FreeBSD-current tree
    into one's production tree so as not to introduce new bugs when fixing
    existing bugs.

    Linux isn't bad, but it is so informal that it is extremely difficult
    to merge in only bug fixes due to the large number of people experimenting
    on the main tree.  Ignoring performance aspects, I find I have to
    discount Linux as a production OS on those grounds alone, but I still
    think Linux is an excellent choice for a home machine.  Taking
    performance into account makes the situation even less tenable.  While
    I admire all the work that has gone into Linux, it is simply too
    inefficient for a production system:  Disk I/O and network performance
    is abysmal and I could never load it down as heavily as I load down
    our FreeBSD machines.  But, again, for a home system, Linux is fine.

					-Matt