Linux-Development-Sys Digest #753, Volume #6     Sat, 29 May 99 01:14:16 EDT

Contents:
  How soon 'till USB Zip drive support? (ReverendTW)
  Re: USB Support (Andreas Dilger)
  Re: Terabite Plus Filesystems (David T. Blake)
  Re: Terabite Plus Filesystems (Nick Manka)
  Re: loading version nonsensitive module into version sensitive kernel? (Jamie Walker)
  How to access SQL Server using C language in Linux ??? ("Henrique Eichler")
  Re: Terabite Plus Filesystems ("Al in Seattle")
  Canon Printer Users...Please Read ("Deborah K")
  Re: Terabite Plus Filesystems (Leslie Mikesell)
  Re: CVS (Alexander Viro)
  Re: Terabite Plus Filesystems (Christopher Browne)
  Re: Can I call system function directly ? (Christopher Browne)
  Re: Rebuilding SRPMs (Christopher Browne)
  Re: IBM PS/2 Microchannel SCSI device driver ? ("Arne K. Haaje")

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

From: ReverendTW <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.hardware
Subject: How soon 'till USB Zip drive support?
Date: Fri, 28 May 1999 19:45:40 +0000

   I'm currently running kernel 2.3.2 and the USB support is fairly
dismal, from what I can tell.  First of all, the USB section is labled
as "USB drivers - not for the faint of heart" (which actualy kinda makes
me want to use it even more ;) ) and secondly it looks like there's just
support for a couple USB busses (thankfuly mine seems to be supported),
USB keyboards/mice and USB sound.  I didn't even know you *could* run
sound off USB until I saw that ;)  Anyway, I'm willing to beta/alpha
test any drivers someone may be working on :)


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

From: [EMAIL PROTECTED] (Andreas Dilger)
Subject: Re: USB Support
Date: 28 May 1999 20:51:31 GMT

In article <7imh3l$76$[EMAIL PROTECTED]>,
James R. Van Zandt <[EMAIL PROTECTED]> wrote:
>I hope people are working on scanners, too.  The USB-only scanners
>seem to be less expensive than the SCSI ones.

That's because the USB scanners are the same as the parallel-port ones
(at least this is the case for the Umax scanner).

Cheers, Andreas
-- 
Andreas Dilger   University of Calgary  \"If a man ate a pound of pasta and
                 Micronet Research Group \ a pound of antipasto, would they
Dept of Electrical & Computer Engineering \   cancel out, leaving him still
http://www-mddsp.enel.ucalgary.ca/People/adilger/       hungry?" -- Dogbert

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

From: [EMAIL PROTECTED] (David T. Blake)
Crossposted-To: 
comp.os.ms-windows.nt.admin.misc,comp.sys.sun.admin,comp.sys.hp.misc,comp.os.linux.hardware
Subject: Re: Terabite Plus Filesystems
Date: 28 May 1999 13:02:08 -0700

Jake Maizel <[EMAIL PROTECTED]> writes:

>We are building a system that needs to handle a huge number of files
>that are 500KB-1MB in size (1-2TB total).  Our only constraint right
>now is the desire to use intel-based hardware for the host computers
>for cost purposes.  My question really is regarding which OS would
>best handle a filesystem of this size.

I see an inconsistency here in someone trying to design a real
enterprise class system, but having a budget that only allows
$2-3k/server. Right now I would choose a Tru64Unix, Solaris, HPUx, or
even Irix solution as adequate on non-x86 hardware. Any of those would
need a $5-10k/server to work reasonably well.

-- 
Dave Blake
[EMAIL PROTECTED]

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

From: [EMAIL PROTECTED] (Nick Manka)
Crossposted-To: 
comp.os.ms-windows.nt.admin.misc,comp.sys.sun.admin,comp.sys.hp.misc,comp.os.linux.hardware
Subject: Re: Terabite Plus Filesystems
Date: 28 May 1999 21:14:50 GMT
Reply-To: [EMAIL PROTECTED]

In article <[EMAIL PROTECTED]>,
        Jake Maizel <[EMAIL PROTECTED]> writes:
> We are building a system that needs to handle a huge number of files 
> that are 500KB-1MB in size (1-2TB total).  Our only constraint right now
> is the desire to use intel-based hardware for the host computers for
> cost purposes.  My question really is regarding which OS would best

NT shouldn't be in the running for this.

I also wouldn't predispose myself towards wintel hardware, as the
price of the disk array will likely surpass the host system by
large margin :>

You may want to consider one of the various dedicated storage machines
like EMC, Network Appliance, and Auspex make.

We have a NetApp, and though the disk array is only a paltry 120G,
we've been very pleased with it, and the sales guy rarely fails to
remind us that the machine can be upgraded to a terabyte array :>
It's a lot nicer in many important respects (notably the ease of
growing the filesystem) than our Digital UNIX disk servers, and a
lot faster too.



> HP LPr hosts connected to a AL-FC RAID system (probably HP).  We would
> want to pick either HPUX, linux, NT or Solaris x86.   Any experience

I would suggest getting a "real" system like a decent HP-UX or SGI
or DECpaq UNIX machine. If you must use Intel hardware, I would
sugges Net or FreeBSD instead of Linux, FFS is a lot more tested
and reliable than ext2 and with the recent "soft updates" dependency
analysis and reordering it's just as fast if not faster.


-- 
All dictators are grey in the dark.

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

From: Jamie Walker <[EMAIL PROTECTED]>
Crossposted-To: fa.linux.kernel,revue.linux-kernel
Subject: Re: loading version nonsensitive module into version sensitive kernel?
Date: Fri, 28 May 1999 17:49:24 +0100

In article <[EMAIL PROTECTED]>, Rolf Welde
Skeie <[EMAIL PROTECTED]> writes
>Hi!
>
>We have a driver which (unfortunately) is not open source.
>We deliver this driver in binary format (as an rpm) to customers.
>Customers sometimes add patches or use newer kernels than what
>we compiled with. With versioning (which is default on and *mostly* a
>good thing in my opinion) symbol names change and our driver module does
>not load. We do not wish to tell our customers to disable versioning.
>Is there any way of getting around this?
I have often wondered whether it is feasible to ship a module in two
parts - an open source part that hooks into the kernel, and a closed
source binary which is called by the open source part. I.e. the open
source part would just be functions that hook into the kernel and call
specifically defined points in the binary file.

It may even be possible to write a script to put the real work of the
driver into inline asm blocks in a c source file, and then compile this
into the kernel - thus removing the need for the binary part. 

In this way, companies/individuals that are under NDA would be able to
write the bits they are allowed to disclose in the standard way, and
still be able to ship a driver that will compile with any (ok, most)
kernel release. I know this is nowhere near as good as having the whole
source code for the driver, but this (if it worked) would let me, for
example, use the SBLive! driver from Creative on a 2.2.9 kernel
(although, I haven't tried this, but I know that under the present
situation of a binary only module, it is not guaranteed to work).

Don't think I explained that very well, but does anyone that actually
writes code for the kernel and/or modules want to comment?
-- 
Jamie Walker,                     http://www.howgarth.demon.co.uk/
LaL Computers: http://lal.rvx.net/ <- for cheap computer hardware.
"You're one of those condescending Unix computer users!"
"Here's a nickel, kid.  Get yourself a better computer" - Dilbert.

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

From: "Henrique Eichler" <[EMAIL PROTECTED]>
Subject: How to access SQL Server using C language in Linux ???
Date: Fri, 28 May 1999 20:13:11 -0300





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

From: "Al in Seattle" <[EMAIL PROTECTED]>
Crossposted-To: 
comp.os.ms-windows.nt.admin.misc,comp.sys.sun.admin,comp.sys.hp.misc,comp.os.linux.hardware
Subject: Re: Terabite Plus Filesystems
Date: Fri, 28 May 1999 16:30:42 -0700

I don't see where money is an issue in his original mail.
Other than the fact that you folks all use Unix based systems that are
recommending Unix based system, what technical reason are you siting for not
using an NT based system?

Some of the quotes:
"I feel that if your data is important and you want a file server that comes
up and stays up, you should discount NT immediately. I have heard some
horror stories about NT with very large directories "  no basis in fact
here.

"PCs are just not
built to the same standard as most of the "real" Unix boxes from Sun, HP,
IBM, SGI, etc. The one exception that comes to mind would be the Sequent
range."     pure bs. It simply depends on what you are willing to spend.
Compaq and others have totally capable boxes if you want to spend the same
kind of money that the Unix crowd delivers.

I would continue your quest to price out the system consistently on various
box types. Your expertise as either an NT or Unix admin will ultimately be
probably the main factor here.

al in seattle



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

From: "Deborah K" <[EMAIL PROTECTED]>
Subject: Canon Printer Users...Please Read
Date: Fri, 28 May 1999 20:32:16 -0500

To all,
    I wrote canon about a driver for my printer that would work with Linux
and here is there reply!
*********** Your inquiry as follows: ******
*****
Is there a model that supports linux right now? If not do you plan to
support linux operating system on any of your models?

===== Original Message =====
From: <[EMAIL PROTECTED]>
Sent: Thursday, May 27, 1999 10:22 AM
Subject: Re:Color BJ Printer - BJC-610


> *********** Your inquiry as follows: ******
> *****
> This is more of a question. I am going to be installing Linux on my second
hard-drive and want to know if you know of a driver that will work for my
printer under linux?
>  Thanks ahead of time Deborah Gooch
> *********** CCSI response as follows: *****
> ******
> Dear Deborah,
>
> Thank you for your inquiry.
>
> Canon does not currently support the Linux operating system. There are
currently no plans for future Linux operating system driver development for
this model.
>
> Please keep all past inquiries/replies with the future inquiries/replies.
Thanks.
>
> Best Regards,
> Tom
> Canon Computer Systems, Inc.


*********** CCSI response as follows: *****
******
Dear Deborah,

Thank you for your inquiry.

Canon does not currently have any printer models that support Linux. There
are currently no plans for Linux driver development.

Please keep all past inquiries/replies with the future inquiries/replies.
Thanks.

Best Regards,
Tom
Canon Computer Systems, Inc.



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

From: [EMAIL PROTECTED] (Leslie Mikesell)
Crossposted-To: 
comp.os.ms-windows.nt.admin.misc,comp.sys.sun.admin,comp.sys.hp.misc,comp.os.linux.hardware
Subject: Re: Terabite Plus Filesystems
Date: 28 May 1999 21:36:27 -0500

In article <7in9f3$iac$[EMAIL PROTECTED]>,
Al in Seattle <[EMAIL PROTECTED]> wrote:
>I don't see where money is an issue in his original mail.
>Other than the fact that you folks all use Unix based systems that are
>recommending Unix based system, what technical reason are you siting for not
>using an NT based system?
>
>Some of the quotes:
>"I feel that if your data is important and you want a file server that comes
>up and stays up, you should discount NT immediately. I have heard some
>horror stories about NT with very large directories "  no basis in fact
>here.

I have a real horror story about NT with a very large directory.
It crashed occasionaly and kept taking longer and longer to finish
the chkdisk before coming back up as more files accumulated.
Eventually it never did finish the chkdisk - or at least I gave up
after 3 days and replaced the machine.  I don't actually know how
many files had accumulated but the disk space was not full.

  Les Mikesell
    [EMAIL PROTECTED]

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

From: [EMAIL PROTECTED] (Alexander Viro)
Subject: Re: CVS
Date: 28 May 1999 20:29:15 -0400

In article <[EMAIL PROTECTED]>,
Geunhyung Kim  <[EMAIL PROTECTED]> wrote:
>Hello!
>
>Could I explain what the CVS is and what is the benefit of that and the
>differencce between that and others ?

Difference between CVS and researcher unable to RTFM is that the former
is useful.

Please, do yourself a favor - get relevant documentation and read it.

>Research Staff
>Network Technology Team
>Telecommunication Network Lab.
>Korea Telecom

-- 
"You're one of those condescending Unix computer users!"
"Here's a nickel, kid.  Get yourself a better computer" - Dilbert.

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

From: [EMAIL PROTECTED] (Christopher Browne)
Crossposted-To: 
comp.os.ms-windows.nt.admin.misc,comp.sys.sun.admin,comp.sys.hp.misc,comp.os.linux.hardware
Subject: Re: Terabite Plus Filesystems
Reply-To: [EMAIL PROTECTED]
Date: Sat, 29 May 1999 00:38:07 GMT

On Fri, 28 May 1999 11:58:43 -0700, Al in Seattle
<[EMAIL PROTECTED]> wrote: 
>You might want to check out the Terraserver project by Microsoft. They seem
>to be doing what you are wanting to do with NT and existing hardware
>environments. This is all NT, NTFS, SQL Server 7.0, etc.
>
>Database Statistics
>The Terra-Server database has 999.3 GB of user data stored in 344.3 million
>records. About 75 GB of additional space is consumed by overhead (about
>25%). The remaining space is used for indices, catalogs, recovery logs, and
>temporary storage for queries and utilities. The database has a formatted
>capacity of 1.1 TB.

Unfortunately, the material substantially represents "big blobs" of data
that could probably be about as effectively handled by a series of
filesystems as by a SQL DBMS.  (Note that the more modern journalled
filesystems share considerable features with journalled databases, so
there is considerable convergence in the ideas.)

At any rate, this sort of application tends to stress the "database"
part of RDBMS functionality as opposed to the "relational" side of it.
Other database systems involving on the order of terabytes of data tend
to involve *huge* TP analysis, which is a vastly different sort of
application involving much finer granularity of "objects."

If the application *is* similar enough, Terraserver is worth comparing
to, to be sure.  Just be aware that care must be taken in comparisons,
as there are smaller DBs that may prove considerably *more* challenging
to manage... 

-- 
The *Worst* Things to Say to a Police Officer:  Hey, is that a 9 mm?
That's nothing compared to this .44 magnum.  
[EMAIL PROTECTED] <http://www.ntlug.org/~cbbrowne/lsf.html>

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

From: [EMAIL PROTECTED] (Christopher Browne)
Subject: Re: Can I call system function directly ?
Reply-To: [EMAIL PROTECTED]
Date: Sat, 29 May 1999 00:39:37 GMT

On Thu, 27 May 1999 20:06:58 GMT, Juergen Heinzl
<[EMAIL PROTECTED]> wrote: 
>In article <7ik16f$s1q$[EMAIL PROTECTED]>, Hui-ung Kwon wrote:
>>Now I write daemon program.
>>System function call via system call increases overhead.
>>So I want to call system function( ex). sys_open, vfork) directly in my
>>application.
>>Can I call system function directly in my application for reducing overhead?
>
>Yes, but (a) you'll have to take care of errno and such yourself and (b)
>a system call requires a context switch and this is the most expensive
>part anyway. In short ... forget it.

In effect, if the overhead is too much of a cost, it is probably
necessary to turn the daemon into a kernel module, pushing it into the
kernel, and thereby eliminating some context switch costs. 

Of course, that immediately kills portability to other systems, which is
undesirable, may tie you to particular kernel versions, and forces
substantially different programming techniques that will feel rather
restrictive... 

Those "undesirables" having been mentioned, it still may be worthwhile
to do this for a real-time application, and the "kernel module" approach
is essentially what the RT-Linux extension provides.

I'd look at whatever other options I could find first before considering
this, though... 
-- 
If only women came with pull-down menus and online help.
[EMAIL PROTECTED] <http://www.ntlug.org/~cbbrowne/linuxkernel.html>

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

From: [EMAIL PROTECTED] (Christopher Browne)
Crossposted-To: comp.os.linux.misc
Subject: Re: Rebuilding SRPMs
Reply-To: [EMAIL PROTECTED]
Date: Sat, 29 May 1999 00:40:49 GMT

On 28 May 1999 09:20:56 -0400, Johan Kullstam
<[EMAIL PROTECTED]> wrote: 
>what i am trying to say is this
>
>1) if it's mission critical - compile specifically for the target host
>   processor.  recompile for each machine architechure.
>
>2) it it's *not* mission critical, then it just doesn't matter does
>   it?  just use a happy middle such as i486 target.

The majority of the "value" in this is likely to come in compiling code
that is CPU-intensive and that runs a lot. 

Major libraries (e.g. - libc, compression-related libs, libs that
manipulate graphical images) would be good candidates; if you were to
just do these, you'd get the *vast* majority of potential benefits.

I expect that tuning GIMP would reap benefits, which doesn't matter if
you're not using GIMP heavily.

Compiling the kernel to be "MMX-optimized" will be worthwhile in the
specific case where you're running SCSI RAID, and can use MMX to make
RAID5 checksums run faster.  That is probably the only *significant*
thing that could benefit in the kernel, and is only helpful if you're
using SCSI and RAID5.  

That kernel optimization likely won't expose any Pentium-versus-PPro
differences. 

Beyond that, I would be quite skeptical that there would be
*substantial* performance improvements available to be reaped from
specially compiling code for IA-32 variations.

The BSD Ports system likely would be a good way of handling this sort of
customization of "compiler options;" I also hear rumor that a
rearchitecture of Debian's apt-get scheme may provide a similar ability
to automatically download packages as source, compile and install. 

Frankly, I wouldn't worry too much about the different compiler options.
If your processes are I/O bound, which is more common than them being
CPU-bound, then the "peephole optimization" of object code won't be
terribly helpful anyways... 
-- 
"Some sins carry with them their own automatic punishment.  Microsoft is
one such.  Live by the Bill, suffer by the Bill, die by the Bill." 
--  Tom Christiansen
[EMAIL PROTECTED] <http://www.hex.net/~cbbrowne/linuxpkgs.html>

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

From: "Arne K. Haaje" <[EMAIL PROTECTED]>
Subject: Re: IBM PS/2 Microchannel SCSI device driver ?
Date: Tue, 25 May 1999 22:33:25 +0200

selious wrote:
> 
> Thomas Hassel wrote:
> 
> > Is there anyone who had Linux installed on a IBM Server 9585-XXX ? I'm
> > searching for the driver to use the built-in SCSI Microchannel-Adapter with
> > Hard-disk and cd-rom.
> 
> I have the model 76i, it uses the same SCSI card...
> 
> Use the standard Future Domain TMC-16x0 SCSI driver !!
> 
> It is supported on the Debian 2.0/2.1 installation floppy, and should
> auto-probe !!

Slackware has a MCA-bootdisk. My newserver runs on a 9585. Uptime today
is 91 days!

Arne
-- 
================================
Arne K. Haaje   | 
Enebakkveien 2  | M: 92 88 44 66
N-1825 Tomter   | [EMAIL PROTECTED]
================================
The system needed Windows 95
or better, so I installed Linux!

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


** FOR YOUR REFERENCE **

The service address, to which questions about the list itself and requests
to be added to or deleted from it should be directed, is:

    Internet: [EMAIL PROTECTED]

You can send mail to the entire list (and comp.os.linux.development.system) via:

    Internet: [EMAIL PROTECTED]

Linux may be obtained via one of these FTP sites:
    ftp.funet.fi                                pub/Linux
    tsx-11.mit.edu                              pub/linux
    sunsite.unc.edu                             pub/Linux

End of Linux-Development-System Digest
******************************

Reply via email to