Date: Fri, 13 Dec 2002 15:12:56 +0100 From: Andy Sporner <sporner@nentec.de> To: "Ronald G. Minnich" <rminnich@lanl.gov>, freebsd-cluster <freebsd-cluster@FreeBSD.ORG> Subject: Re: sharing files within a cluster Message-ID: <3DF9EAE8.9040704@nentec.de> References: <Pine.LNX.4.44.0212121030540.18020-100000@carotid.ccs.lanl.gov>
next in thread | previous in thread | raw e-mail | index | archive | help
Hi Ron,
No harm done! :-) Sometimes this can be a very passionate subject.
Ronald G. Minnich wrote:
>On Thu, 12 Dec 2002, Andy Sporner wrote:
>
>
>
>>Well let's review here. RPC's have many dependencies for correct
>>operation (portmap, etc) regular socket I/O does not--so which one is
>>more reliable???
>>
>>
>
>Actually, that's SunRPC, one variation of RPC, that requires all that
>stuff. RPC is a very simple concept, and what you typically see is people
>implementing things that are actually RPC but calling it socket I/O.
>Pretty much any protocol running over sockets that does a request with a
>"procedure number" and expects a response to that request is RPC.
>
Point taken, then I would have to agree. It would be simpler. I had
planned to put some of this
in the kernel (though I normally hate this) The justification is that
these are proc related things
and why go into user space when all the action is already happening in
the kernel.
>My apologies, the "linuxbios naming" thing really teed me off. I'm also
>sorry to be so darned tactless, a lot of those messages I was just
>pounding out between Pink bringing firefighting or early in the morning
>when I should sleep instead of write email. I apologize to this list.
>
>
I think it takes a better person to acknowledge this. Most people in
the past I have seen
don't so this is a very welcome thing!
>I'm getting really frustrated with this list, however, since people seem
>to be refusing to learn from linux because it's "evil", and at the same
>time proposing stuff that's been tried years ago on linux and found not to
>work. I think it is really important that the FreeBSD community break out
>into something new and really neat. At a minimum, however, I'd like to see
>FreeBSD have equal capability to Linux clustering for HPC. And I think
>before we condemn technologies ("RPC") it's best to make sure it's the
>technology, and not a specific implementation of it, that are the issue.
>
>
Agreed. As i said in my "small flame" point I wanted to port the Bproc
as a learning
exercise, and then to expound on that to do the kind of VM stuff I had
spoken about.
I was not aware of the differences, mainly because having been stung
once (and it was
also at a Government lab--Fermi-labs "CPS project" that I saw
over-featurisms cause
problems--mainly because of RPC's. Since the socket interface isn't too
complicated
and required very few times, I developed my habits away from RPC's. In
the old days
when I worked with SYSV systems they weren't available. These were 3B2
systems
and Pyramid computer (Ok, they had RPC's :-))
>I would really encourage people to get at the literature of SSI, going
>back to Farber's original DCS paper in '72, and try to find an angle that
>is somehow new and uniquely suited to FreeBSD. And "Linux sucks, FreeBSD
>rules" is not going to be the thing that does it. What about FreeBSD can
>somehow make it much better for clustering? That's the problem I've never
>been able to answer -- maybe one of you can.
>
>
I am not sure that there is such an aswer. Religous wars (either
computer or
theological NEVER solved anything--nor will they). I do believe that when
one feels a strength in a system, one tends to use what is more
comfortable. I
in face use Linux on my desktop (as I write this email) because I have
more tools
at my disposal (VMware for instance). For embedded systems I prefer
FreeBSD
because it is easy to work in kernel space compared to Linux. We have a
project
in the house that is using Linux and the developer who was on my project has
been swearing against it because a misplaced printf statement crashes
the kernel.
Where he could have freely put them in Freebsd kernel and got away with it.
But to the point. IMHO, Digital had it right with Vaxclusters and it is
to that
goal that I am trying on my efforts to achieve for FreeBSD. So in terms of
plowing new ground, I am under no illusions at all! :-) I think even
the former
NOW project even had a lot of this in it. I found many others too in
my early
investigation in this. With the time I have left for such projects, I
would have
to be committed if I was going to do this just out of pride.
>This list is sporadically active, but there have not been tons of new
>ideas crossing it -- seems like we see the same stuff over and over again.
>
>
New people come and go all the time :-) I think though we are at the point
of critical mass though (or at least near to it).
>
>
>>If this was such well plowed ground, why does it not already exists in
>>FreeBSD?
>>
>>
>
>Because, and as a former FreeBSD cluster builder I hate to say this, Linux
>won. And the mindshare is in Linux, as are the compilers, 3rd party apps,
>and all the big vendors. There are many 10s of millions of dollars of
>clustering money being spent by IBM, HP, Intel, etc., etc. and absent
>something really new and innovative I don't see FreeBSD breaking in.
>
>The closed-minded nature of the FreeBSD core has not helped. I wrote a DSM
>for FreeBSD 2.0.5 ca. 1994, which was based on simple mods to NFS --
>called MNFS (really it was a port from a SunOS version I also wrote). It
>required a simple extension to the VM system (it required the VM layer to
>tell the VFS layer if a page fault was for read or read/write -- that's
>what SunOS does) and I could not get the FreeBSD core to add this simple
>additional parameter -- it violated "information hiding between layers",
>which is good in theory but in practice the more info the better. (this
>code is still on my web page along with the MNFS papers).
>
>
I have had my share of issues with some of these people as well. This
is the
only list that I have not removed myself from because I can say that people
here are generally polite to one another and I have yet to see a major
flame.
I must also apologize for my response. Fighting fire with fire is also
never
a solution and I felt sort of bad having said the things that I did. So
I must
also ask your indulgence.
>8 years ago both Sandia/Livermore and I were building freebsd clusters.
>It's arguably the better OS. Is it so much better or different from Linux
>that it is inherently better for clustering? Probably not. So we got
>clobbered by the rest of the world. What we got steamrolled by was the
>availability of so many 3rd part apps.
>
>
Totally agreed. My biggest objection to Linux isn't about Linux it's
about some of the marketing engines that are pushing it. I think even
if you were to ask this one who complained about RedHat samba, I think
it was more of a complaint about Redhat than anything else.
I used to work with Redhat before it was traded on the stock market and
I can really say that their attitude has changed. I used to work for a
company that
also went out about the same time. It is important for the stock prices to
make announcements periodically to bolster the price. So if they announce
RedHat Samba for instance, there is a small increase. Does this justify
the development of a separate version (if I understand this correctly), yes
and no. I generally don't agree with doing this unless it can help the body
public first.
I would say only this about the debate so far. That being that a project or
initiative must be viable by itself without consideration of
"one-upmanship".
For my part I work now primarily with FreeBSD because it is comfortable
and that is the only reason. Some years back I had my old company called
BSCsoft (put bscsoft linux in google and you will see it). and we did load
balancing for Linux. We had then a minor offering for BSDi, but at that
time it was the black sheep.
>I'm glad to hear it, sorry for lack of reply. Please get the clustermatic
>ISO at www.clustermatic.org and then we can talk about porting. I would
>LOVE to see this happen.
>
>
You have a deal...
>You can find an initial private name space VFS implementation for FreeBSD,
>called v9fs, I think on my web page.
>
>Sorry, Andy. I'll try to behave.
>
>
So will I ;-)
Best wishes and welcome back...
Andy
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-cluster" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?3DF9EAE8.9040704>