Linux-Hardware Digest #411, Volume #13 Sun, 13 Aug 00 06:13:04 EDT
Contents:
Re: Defrag in Linux? (sideband)
Re: YAMAHA sound card (Nick Chalk)
Re: HP LaserJet 1100 print delays (James Stafford)
Re: Simple question about CD-Writing for Linux (Duane)
Re: Defrag in Linux? (Prasanth A. Kumar)
Re: promise fast trak 66 and suse-linux - followup (Benjamin Grimm)
Vibra 128PCI (Terence Chan)
dead monitor - help!! (paul)
Re: Coppermine SLOW PERFORMANCE... (John Beardmore)
Re: Partition Size Advice (John Beardmore)
Re: Partition Size Advice (John Beardmore)
Re: Partition Size Advice (John Beardmore)
Re: Partition Size Advice (John Beardmore)
----------------------------------------------------------------------------
From: sideband <[EMAIL PROTECTED]>
Subject: Re: Defrag in Linux?
Date: Sun, 13 Aug 2000 02:28:04 -0400
Elliott wrote:
> Thank you all for answering my 8 questions in one. Just out of curiosity, how
> is it that defragmentation is almost never goes over 8%?
> Back to performance, here is the output of my free:
>
> [root@localhost /root]# free
> total used free shared buffers cached
> Mem: 62992 61528 1464 60664 1532 26340
> -/+ buffers/cache: 33656 29336
> Swap: 313228 4632 308596
>
> Wow! I didn't realize that much ram was used! I thought I had 64megs of ram....
> 62.992 is close enough for me. If I bought 64 more megs of ram, do you guys
> think that my performance would increase significantly? thanks!
> J Bland wrote:
>
> > >Filesystem Size Used Avail Use% Mounted on
> > >/dev/hda8 387M 317M 50M 86% /
> > >/dev/hda1 23M 2.6M 19M 12% /boot
> > >/dev/hda7 2.9G 938M 1.8G 34% /usr
> > >/dev/hda5 387M 31M 337M 8% /var
> > >/dev/hdd 390M 4.6M 365M 1% /mnt/drive3
> > >/dev/hdb1 393M 16k 373M 0% /mnt/drive2
> > >
> > >all the partitions appear to be fragmented. I would think, that "size"
> > >minus "used" should equal "Avail". I checked my partition tables with
> > >fdisk, and all cylinders and heads are accounted for. Thus leading me to
> > >think that I really need to defrag. (then what the heck is auto-defrag?)
> > >True, for a web server or fully networked computer, defragmenting may
> > >not make a difference, but for my computer, I think it would. So what
> > >program should I use? I have looked high and low, and I can't find one
> > >(except for auto-defrag, whatever that is). Any help would be nice
> > >[EMAIL PROTECTED]
> >
> > Ext2fs (which you are almost certainly using) does not fragment in the same
> > way that windows filing systems do. Ext2fs is intelligent and arranges its
> > data in such a way as to reduce fragmentation. The only time this really
> > breaks down is when your disc is almost full, and all your partitions are
> > quite a way from this.
> >
> > The small amount of fragmentation (of the order of a % or two) that you get
> > on a standard ext2fs system will not produce a system performance drop that
> > you could see. As for df; it only reports disc usage, there is no indication
> > of fragmentation in its output so I wonder where you think it is saying your
> > partitions are fragmented. Any differences between the allocated size of a
> > partition and the reported size in df are due to:
> >
> > a) Space taken by formatting
> > b) by default 5% of a partition is reserved for the root user only. So even
> > when the disc is supposedly 5% full there is still some space for root to
> > play around in. A godsend when users fill your disc accidentally.
> >
> > If you are experiencing performance drops it is much more likely to be a RAM
> > factor, or more the usage of it. Once you start using up your RAM and hit
> > the swap partition or file, performance nosedives.
> >
> > What are you running? How much RAM do you have? What does 'free' report?
> >
> > Linux doesn't 'degrade' with time like certain other systems. If there's a
> > problem with performance it's to do with what you're doing, not the state
> > the OS has got itself into.
> >
> > Frinky
> >
> > --
> > John Bland MPhys(Hons) GradInstP Webmaster and Sys Admin.
> > http://ringtail.cmp.liv.ac.uk/ Condensed Matter Group
> > Email: j.bland at liv.ac.uk Liverpool University
> > "And it can suck a monkey through 30ft of garden hose!!"
>From the looks of it, if your "free" output shows your typical memory usage, then,
yes, 64 more megs will probably improve your system's speed.
HTH.
-SSB
------------------------------
From: Nick Chalk <[EMAIL PROTECTED]>
Subject: Re: YAMAHA sound card
Date: 12 Aug 2000 21:56:03 +0100
Stuart Fotheringham wrote:
>> Does anyone know how to get the Yamaha sound
>> card working under RH6.2
Flukezero <[EMAIL PROTECTED]> replied:
> Please provide the model of the card (ie.
> YMF-724), and in my experiences with yamahas you
> can't use them unless you buy OSS with the
> yamaha modules.
Recent ALSA releases support the Yamaha YMF7*
family - try http://www.alsa-project.org/
Nick.
--
Nick Chalk ........................ Radio Designer
Confidence is failing to understand the problem.
------------------------------
From: James Stafford <[EMAIL PROTECTED]>
Subject: Re: HP LaserJet 1100 print delays
Date: Sun, 13 Aug 2000 06:39:03 GMT
Daniel Armstrong wrote:
>
> Hello,
>
> I just hooked up a HP LaserJet 1100 to my Linux box running Mandrake
> 7.1. Using <printtool>, after a little fiddling I configured the
> necessary settings using the "LaserJet 4/5/6" driver (the closest
> match). Everything now prints correctly - except that there is a 20
> second delay between each page that is printed. The printer is rated at
> 8ppm, and a friend that has the same printer with the stock memory of
> 2mb under Windows 98 does not experience such delays. What can I do to
> speed up my print jobs under Linux (besides the last resort of upgrading
> printer memory)?
>
> Cheers,
>
> Daniel
I have the same printer, and experience the same delays. I don't know
what your friend is doing, but I have the same delays under Windows. The
only thing you can do is get more RAM for the printer. You should be
able to find 16 MB for about $43.00 +/- if you look around.
jamess
--
"On the side of the software box, in the 'System Requirements' section,
it said 'Requires Windows 95 or better'. So I installed Linux."
-Anonymous
------------------------------
From: Duane <[EMAIL PROTECTED]>
Crossposted-To:
linux.apps.cdwrite,comp.os.linux.questions,comp.os.linux.redhat,comp.os.linux,comp.os.linux.setup
Subject: Re: Simple question about CD-Writing for Linux
Date: Sat, 12 Aug 2000 20:04:07 -0700
Jimi Aleshin wrote:
>
> Here is my /etc/fstab file:
>
> /dev/hdb4 / ext2 defaults
> 1 1
> /dev/hdb1 /boot ext2 defaults
> 1 2
> /dev/cdrom /mnt/cdrom iso9660 noauto,owner,ro
> 0 0
> /dev/hdb3 swap swap defaults
> 0 0
> /dev/fd0 /mnt/floppy ext2 noauto,owner
> 0 0
> none /proc proc defaults
> 0 0
> none /dev/pts devpts gid=5,mode=620
> 0 0
>
> I'm wondering where my CD-Burner should go or if anything should be
> renamed, and where it should be mounted, etc. I use cdrecord 1.8.1. I
> tried everything what everybody else said in the past 2 messages and
> nothing works at all.
>
> I turned on all scsi support, emulation, etc. I run RH Linux 6.1 kernel
> 2.2.12-20, and I just rejumpered my cdrom to slave and cd-writer to
> master. And right now, my cdrom doesn't work in linux and my cd-burner
> now works, but not detected to BURN cd's. Like I try "cdrecord -scanbus"
> and it says, can't detect scsi driver. And plus I read the HOW-TO doc
> too. I have no idea what to do?
>
> Any other ideas??
>
> Jimi Aleshin
Are you saying you recompiled the kernel? Are you mounting data CDs as
/dev/scd0 or /dev/hdc? If you have not changed either /dev/cdrom or
/etc/fstab, then you are likely using /dev/hdc. If that is so, then the
fact that it still "works", which I assume means you can mount disks,
means that you did not recompile the kernel correctly.
With a standard RH 6.1 installation, there is no need to recompile the
kernel. If you still have the original kernel, you can go back to it and
do this:
In your /etc/lilo.conf, in the section where you are booting the
original kernel, put the lines:
append="hdc=ide-scsi"
append="hdd=ide-scsi"
This tells the kernel to use IDE SCSI for both CD drives (I am assuming
they are hdc and hdd). Sometimes there are problems if you try to run
one drive with ide-cd and one with ide-scsi.
After making those changes, execute:
# /sbin/lilo
That tells lilo to use the new settings.
Now go to /etc/conf.modules and add the line:
alias scsi_hostadapter ide-scsi
That tells the module autoloading deamon to use the ide-scsi module. Now
reboot, and try rerunning:
# cdrecord -scanbus
It should now work. If not, post the contents of /var/log/dmesg, or at
least the lines that look like they have something to do with IDE and
SCSI. Note that using this technique, the CDROM automounting will no
longer work. You will have to manually mount data CDs (of course, you
never mount audio or blank CDs).
The only advantage of recompiling the kernel is that it allows you to
use the automounter, but I personally think the automounter is evil (at
least the version in RH 6.1) and should be avoided.
If you really want to use a recompiled kernel, then make sure you have
compiled these:
In Block Devices
Select N for include IDE/ATAPI CDROM support.
Select Y for SCSI emulation support.
In SCSI support
Select Y for SCSI support
Select Y for SCSI CD-ROM support
Select M for SCSI generic support
Select N for Probe all LUNs on each SCSI device.
With this method, absolutely no extra entries are needed in either
/etc/lilo.conf or /etc/conf.modules. If you have put some there, they
probably won't hurt anything, but they really should be removed.
If you have done this already, and it still doesn't work, again, post
the contents of /var/log/dmesg. And as much detail as you can of what
you have done. Your messages have been very vague so far.
With either method, CDs are now mounted using /dev/scd0 for the primary
drive and /dev/scd1 for the secondary drive. You probably want to make
/dev/cdrom a symbolic link to /dev/scd0. For recording CDs, cdrecord
uses the devices /dev/sga and /dev/sgb (or /dev/sg0 and /dev/sg1, which
are the same thing) but that is transparent to you - just use the device
nomenclature given by cdrecord -scanbus.
By the way, please reconsider whether cross posting to six news groups
is really necessary.
--
My real email is akamail.com@dclark (or something like that).
------------------------------
Subject: Re: Defrag in Linux?
From: [EMAIL PROTECTED] (Prasanth A. Kumar)
Date: Sun, 13 Aug 2000 07:05:40 GMT
Elliott <[EMAIL PROTECTED]> writes:
> Thank you all for answering my 8 questions in one. Just out of curiosity, how
> is it that defragmentation is almost never goes over 8%?
> Back to performance, here is the output of my free:
>
> [root@localhost /root]# free
> total used free shared buffers cached
> Mem: 62992 61528 1464 60664 1532 26340
> -/+ buffers/cache: 33656 29336
> Swap: 313228 4632 308596
>
> Wow! I didn't realize that much ram was used! I thought I had 64megs of ram....
> 62.992 is close enough for me. If I bought 64 more megs of ram, do you guys
> think that my performance would increase significantly? thanks!
<snip>
The actual amount of memory used is 33656K, not 61528K. This is
because the second number includes memory alloted for buffers and
cache for which there was no other use at that instance in time.
If you want to see if adding more memory will help, run the system
with stuff like X windows, netscape, shell that you typically do. At
the same time, run 'vmstat 1 1000' on a terminal and watch the 'si'
and 'so' values. If they are mostly 0 then there is not much swapping
so extra ram wouldn't help much.
--
Prasanth Kumar
[EMAIL PROTECTED]
------------------------------
From: Benjamin Grimm <[EMAIL PROTECTED]>
Subject: Re: promise fast trak 66 and suse-linux - followup
Date: Sun, 13 Aug 2000 09:48:21 +0200
Neil Golstein wrote:
> Well, here's the info. Fortunately since I actually HAVE Redhat I can
> switch kernels pretty easy, just rpm them in and out. You would have to
> build from source I guess. Anyway, the ftmod61b modules does load in the
> 2.2.12 kernel, at least the uniprocessor kernel, and at least the one RedHat
> calls "i386" haven't tried the "i686" kernel yet but may. The smp kernel
> doesn't work by the way, I tried it.
>
> Unfortunately, about 10 seconds after the driver loads (seemingly
> successfully), I get a kernel panic, with oops! and aiee! messages, system
> freezes, much numbers and etc. displaying on screen, none of which get into
> the logs, but what does get into the logs is the following:
> Aug 12 11:25:55 neil kernel: FastTrak Driver v1.03 (21.APR.2000) :
> Aug 12 11:25:55 neil kernel: scsi1 : FASTTRAK
> Aug 12 11:25:55 neil kernel: scsi : 2 hosts.
I tried to patch Suse 6.4 / kernel 2.2.14 with the patch supplied for RedHat
6.1. The patching and recompiling of
the kernel wasn't a problem, but when I boot up the system I just got the
messages over there (scsi: 1 host found),
none of the following messages or a detected drive array.
Aug 12 11:25:55 neil kernel: Vendor: Promise Model: 2+0 Stripe/RAID0
> Rev: 1.10
> Aug 12 11:25:55 neil kernel: Type: Direct-Access
> ANSI SCSI revision: 02
> Aug 12 11:25:55 neil kernel: Detected scsi disk sdc at scsi1, channel 0, id
> 0, lun 0
> Aug 12 11:25:55 neil kernel: SCSI device sdc: hdwr sector= 512 bytes.
> Sectors= 58673538 [28649 MB] [28.6 GB]
> Aug 12 11:25:55 neil kernel: sdc: sdc1
>
> Note that I am running the latest bios for the promise, and yes the size of
> the array is correct. Also I loaded an Initio SCSI driver first which is
> why you see sdc1, but I also tried it just loading the promise and it set
> sda1 for the array, also froze the system, so it's not a conflict with the
> other scsi driver.
>
> And NO, in the ten seconds I didn't quick mount the array to see if I could
> really access files. Maybe I will though, one last chance is to try the
> i686 kernel (I never did know what the difference was). But I suspect it's
I already choose a new compiled kernel and got no kernel panic or so on.
>
> a hardware or system bios conflict or something like that. I have a P3500
> on a BX board.
>
> Also since I have a genuine contact at Promise maybe I will email them and
> ask for help but not counting on it. I would like to know if they are
> continuing to work on this driver.
>
> By the way, since I installed the Fasttrak lilo is totally broken, I only
> use loadlin now. Interesting. Yet BeOS has no problem
I got no problems with lilo, I'm using a 20 gb boot-disk without raid etc.
------------------------------
From: Terence Chan <[EMAIL PROTECTED]>
Subject: Vibra 128PCI
Date: Sun, 13 Aug 2000 15:24:53 +0800
I've installed RH 6.2 on my machine. I tried to set up my Vibra 128PCI
sound card with sndconfig. It detect the sound card as "Ensoniq | Es1371
Audio PCI-97]. But hear nothing when testing. What can I try now?
--
-(^)-(^)- ! Terence Chan
b ! mailto:[EMAIL PROTECTED]
--- !
------------------------------
From: paul <[EMAIL PROTECTED]>
Subject: dead monitor - help!!
Date: Sun, 13 Aug 2000 08:04:48 GMT
one of my monitors just died - instead of powering up the monitor
continualy "click click clicks" as if its trying but failing to power
up. Any ideas what the problem is, and whether it can be fixed or just
dumped?
monitor is a Hyundai DeluxScan 17b.
Thanks a lot for any advice.
-paul
------------------------------
From: John Beardmore <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.setup
Subject: Re: Coppermine SLOW PERFORMANCE...
Date: Fri, 11 Aug 2000 22:58:57 +0100
In article <[EMAIL PROTECTED]>, Johan Kullstam
<[EMAIL PROTECTED]> writes
>while cpu speeds have been growing by leaps and bounds, the rest of
>the system hasn't been growing nearly as fast.
I was thinking about that, but my first system was a 286/8 with a 30 and
70 meg disks.
My latest is a PIII/550 with 17 and 75 gig disks so I guess that's about
1000 times more storage. Transfer rates gone from about 180 k/sec to 18
meg / sec, only about 100 times better.
> if you are ram bound
>and have pc100 sdram everywhere, you won't see any increase.
Ram seems to be the worst of all. OK, it's gone from 3 meg to 512, but
ignoring the fact that the bus got wider, I doubt the speed has improved
by more than a factor of 10 from the original 150 ns DRAM.
Cheers, J/.
--
John Beardmore
------------------------------
From: John Beardmore <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.setup,linux.redhat.install
Subject: Re: Partition Size Advice
Date: Sat, 12 Aug 2000 11:33:10 +0100
In article <[EMAIL PROTECTED]>, sideband <[EMAIL PROTECTED]>
writes
>> 6.4GB Hd...
>> /usr = 5GB
>> /root = 500MB
>> /home = 500MB
>> /swap = 127MB
Apart from weight of tradition, why do usr, root and home have to be
different partitions ?
( I assume is used raw without a file system so can't share space with
other partitions ? )
If they can share a partition on a single drive, usr, root and home will
be able to share space and there should be less chance of any one of
those partitions running out of space.
I assume this sort of configuration would be illegal though ?
Cheers, J/.
--
John Beardmore
------------------------------
From: John Beardmore <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.setup,linux.redhat.install
Subject: Re: Partition Size Advice
Date: Sat, 12 Aug 2000 11:37:24 +0100
In article <[EMAIL PROTECTED]>, Tom Van Lemmens
<[EMAIL PROTECTED]> writes
>seems ok, although depends on how you are using your system. i would
>suggest to increase the /home partition or create another one. this is
>how it works for me:
>
>- relative small /root (max 300mb)
>- /swap = 128mb
>- /usr = �3gb
>- /home = 1.5gb
>- rest is /incoming
What is the history behind the /home and for that matter the /var
partitions ?
Early UNIX boxes didn't have them and we didn't miss them. I stumble
across them on newer systems and don't know with to do with them !
What's wrong with having /var and /home in your file system but not
having them as separate partitions ?
Cheers, J/.
--
John Beardmore
------------------------------
From: John Beardmore <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.setup,linux.redhat.install
Subject: Re: Partition Size Advice
Date: Sat, 12 Aug 2000 11:42:16 +0100
In article <8m6v3h$k1$[EMAIL PROTECTED]>, Geoff Short
<[EMAIL PROTECTED]> writes
>You CANNOT have too much /usr ... you can have too little of something
>else though :-) Unless there's a definite reason I always try and stick
>to one big partition, then you don't get worried about which partition
>your free space is on.
Staring to make sense !!
> (This is assuming it's a personal machine
>rather than a server)
So what's different about servers ? Why do they need to be partitioned
so inflexibly ?
I have an Alpha 2100 box and was thinking of trying to do a custom RH
6.2 install with a small raid array as /usr and one partition on another
drove for /root and the rest. Would that be legal ?
Cheers, J/.
--
John Beardmore
------------------------------
From: John Beardmore <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.setup,linux.redhat.install
Subject: Re: Partition Size Advice
Date: Sat, 12 Aug 2000 11:48:11 +0100
In article <[EMAIL PROTECTED]>, Johan Kullstam
<[EMAIL PROTECTED]> writes
>i like
>
>/boot 10-20MB *at front of disk*
>/home 1-2GB or more as you've got space to spare
>/ 4-5GB *including* /usr
><swap> 128MB
What's the rationale for sizing swap partitions ?
In Win32, the rule of thumb seems to be to have an initial swap file
size that is say 1.5 time the size of physical ram, and perhaps to let
it grow bigger if needs be.
How sensible would it be on say a 512 meg ram Linux box to have a mere
64 or 128 meg swap partition ? Can a Linux box be configured without a
swap partition at all ? My gut feeling is that it would be a bit of a
waste of space ! By the time you needed it at all, you'd be more or
less totally out of space !
Cheers, J/.
--
John Beardmore
------------------------------
** 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.hardware) 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-Hardware Digest
******************************
