Skip site navigation (1)Skip section navigation (2)
Date:      27 Jul 1998 17:26:38 +0200
From:      Peter Mutsaers <plm@xs4all.nl>
To:        sas@schell.de
Cc:        freebsd-questions@FreeBSD.ORG
Subject:   Re: FreeBSD or LINUX??? - Which one should I choose?
Message-ID:  <87n29v2u35.fsf@muon.xs4all.nl>
In-Reply-To: Sascha Schumann's message of "Mon, 27 Jul 1998 02:12:51 %2B0200 (MET DST)"
References:  <Pine.LNX.4.00.9807261214020.589-100000@guerilla.foo.bar>

next in thread | previous in thread | raw e-mail | index | archive | help
>> On Mon, 27 Jul 1998 02:12:51 +0200 (MET DST), Sascha Schumann
>> <sas@schell.de> said:
    SS> I'm running FreeBSD-stable (2.2.7) now for four days and it
    SS> crashed three times on me. The first time, I copied a 2MB file
    SS> to a clean ext2fs partition - the system hang (I could still
    SS> switch between terminals), but the partition was mixed up -

You're right: ext2fs is unstable and causes the system to crash. I
just tried it last night again with -current and I got instantaneous
crashes. Otoh, Linux doesn't even offer a read/write FreeBSD
filesystem, only a read-only one. Alas, communication between FreeBSD
and Linux is only possible via tape or FAT.

    SS> My own Linux installation here crashed twice - in about 3
    SS> years. The first time was while running Quake with wrong
    SS> settings and the second time, it was a known KFM bug which I
    SS> "tested" out...

Well, my Linux installation (running 2.0.35) crashed 4 times in the
past week. 2.1.110 crashed several times too. I think (for both Linux
and FreeBSD) it depends on what you do with it and what hardware you
have (ie what drivers do you use and how is their quality). In general
though (looking at the mailing lists) I got the impression that
FreeBSD -current is about as stable as Linux -stable (2.0.35) and
Linux 2.1.x is much less stable.

    >> >> - Linux scheduling algorithm is poor on high system load
    >> >> (THUS - to run an internet server I'd prefer FreeBSD!)
    >> 
    SS> Which of the scheduling algorithms do you mean? There are
    SS> actually three I remember now (rr, fifo, other). And there is
    SS> Linux-rt (realtime support).  Does FreeBSD have this?

Do you mean processes with real-time priorities? Yes, about like Linux
has it (not hard real time, but a real-time priority class for
processes). See rtprio (maybe only -current, I don't know for sure).

    >> 
    >> This is really true, and one of the reasons I'm annoyed right now. I
    >> ran FreeBSD, now use Linux and am upset about the (relative of course,
    >> Windows is much worse of course) poorer scheduling I experience.

    SS> Can you prove this with some numbers? 

No, as I mentioned it is how the system feels. I didn't do any
scheduling benchmarks. On my system I experience clearly that FreeBSD
runs smoother and faster, but on other hardware it may be vice versa
(depending maybe on the type of SCSI or IDE controller you have).

There's an easy test you can do: Try creating very large directories
(like with >10000 files in them) and then delete all files. You'll see
that FreeBSD handles such cases much better (using async or
softupdates for mounts of course).

    SS> Just do hdparm -d 1 /dev/hd? and your UDMA will work. If you
    SS> are lurking on linux kernel so hard you have probably seen the
    SS> messages about UDMA support in the last 24 hours...

No, it did't work. Yes, I do know that Linus disabled UDMA support in
2.1.111 (which I didn't try yet) because someone got file corruption
while using UDMA.

The UDMA wouldn't be activated because Linux didn't recognize my IDE
chipset. I could put it in 32 bit mode with readahead though. That
helped a bit, but not much.

    >> 3.0 has been supporting it for a long time. In the Linux world, being
    >> much less conservative (or more careless) 3.0 would have been the
    >> stable production release already.

    SS> You know it probably, but you don't name it: New features go
    SS> into the dev kernels (e.g. 1.3.x, 2.1.x) are tested there and
    SS> if they prove stable/bugfree they make it into stable
    SS> releases. I don't know why anyone could call this behavior
    SS> less conservative (it makes sure everything works!!) or more
    SS> careless (I always thought that would include doing something
    SS> dangerous...)

I know, but the question is: when do you move from 1.3 to 2.0. Was 1.3
really so stable at 1.3.100 (or whatever) when it moved to 2.0. What I
meant to say is that FreeBSD probably would have waited much longer
before declaring a branch stable, which explains why new features
appear later in the -stable releases. Whether this is wise is the
question. It is a very safe approach, but sometimes to be successful
one should take more risk at move faster (look at Microsoft, they
release what is outright alpha or beta code, just to get some new
stuff to the market fast).

    SS> I don't want to blame FreeBSD here for anything. I'm *very*
    SS> new to it. I'm Linux biased. I'm open to new things - but not
    SS> to Linux bashing.

I wasn't bashing Linux. I've used Linux since 0.98, (i.e. much earlier
that *BSD) and would never bash it. But having quite a bit of
experience with both I think I'm entitled to give my opinion on each
advantages and disadvantages. All in all I prefer FreeBSD and think it
is a more efficient and nicer environment, but Linux also has it's
advantages (such as more available software and faster support for
more and newer types of hardware). But IMO the best situation would be
if all effort going into Linux drivers etc. would be spent on FreeBSD
instead.

regards

-- 
 /\_/\
( o.o ) Peter Mutsaers  |  Abcoude (Utrecht), |  Trust me, I know
 ) ^ (  plm@xs4all.nl   |  the Netherlands    |  what I'm doing.

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-questions" in the body of the message



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?87n29v2u35.fsf>