Re: boot and shutown both suddenly getting delayed

1999-10-28 Thread Andrew Hately
Anthony Campbell wrote:
> 
> I've suddenly started experiencing long delays in both boot and
> shutdown.

If accompanied by ide reset messages your hard disk could be dying.

Andrew

mailto://[EMAIL PROTECTED] 
http://counter.li.org/cgi-bin/runscript/display-person.cgi?user=45690


Re: Linux/X and LCD projectors (sort of video beamer)

1999-10-27 Thread Andrew Hately
Joachim Trinkwitz wrote:

> Now I'm looking for a possibility to set down the monitor frequency
> without losing to much of the picture quality.
> 
> Any ideas?

Change your mode lines.

If there are not a hundred well commented mode lines alread in you
XF86Config then read the video timings faq.
If they are there and you want to get to them, rename them to something
special and insert those as your modes of your display.
If all else fails run xf86config.


Andrew
-- 
mailto://[EMAIL PROTECTED] 
http://counter.li.org/cgi-bin/runscript/display-person.cgi?user=45690


RE: Compact version for running on 486/33 or /66 with 8->32 Mb ram

1999-10-27 Thread Andrew Hately
> While worming through the pages I found a reference to
> a compact verison of Debian installation that was
> designed to work on a 486.

Slink works on a 486. I'm running it on two 486 DX2 at home. One with 8MB of
ram, which is too little, one with 20MB which is plenty.

The 20MB ram machine has a 200MB disk of which about 60 is swap. On the rest
is one partition. I installed "basic machine ~25MB" then added the stuff I
needed. /usr/src is mounted over nfs (I always build the smallest possible
working kernel as soon as the machine is up and running). I seem to remember
not installing some of the suggested documentation packages. There is still
free space on the disk, but /home is more or less empty.

The 8MB ram machine has a 2.5GB disk of which 128MB is swap and the rest is
one big partition. Again I installed "basic machine ~25MB" then added things
a bit at a time. On this machine /home is mounted over nfs. Previusly this
machine had 24MB of ram and ran X and numerous applications (e.g. LyX)
satisfactorily. Now with less ram I find it swaps too much to run X, even X
clients are rather hampered - I reconfigured this machine solely to cut
CDroms; which it does successfully from cdwrite but xcdroast produces
coasters due to timeouts from swapping. The 2.5GB disk is mostly empty most
of the time but I want to use it to store iso9660 file system images between
cuts.

As long as you don't need to run the gimp, just install the minimum to do
what you want and avoid unnecessary daemons (apache, x font server, etc) a
486 can be a reasonable linux box.

Andrew


Re: SED question

1999-10-21 Thread Andrew Hately
Ben Lutgens wrote:
> 
> I had to use the back quotes``

Back quotes can't be nested, while $( ) does the same job and can be nested


-- 
http://counter.li.org/cgi-bin/runscript/display-person.cgi?user=45690


Re: small dselect question

1999-10-21 Thread Andrew Hately
jh wrote:
> 
> Hi. If I just want to install 1 program from dselect, would I want to go
> through and put an = sign by the programs that are currently installed so
> it doesn't have to go through all the files when installing? I hope this
> question makes sense.
> 
> Thanks,
> 
> Jeff
> 
> --
> Unsubscribe?  mail -s unsubscribe [EMAIL PROTECTED] < /dev/null


Wouldn't that mean that should a newer and better version of any ='ed file
be available you would not install it?
Sounds like a bad side effect.
If you know you just want to install one thing and dont want to wait for
dselect to run for an hour, try dpkg.

Andrew

-- 
http://counter.li.org/cgi-bin/runscript/display-person.cgi?user=45690


RE: [Linux: SCSI-Controller] LILO with aha152x

1999-10-19 Thread Andrew Hately
...
> If I start with my Rescue-Disk and I use the following parameter
>
> linux aha152x=0x340,11

Are you using a VLbus card with one of these built in? A something-or-other
VL300?
If so what is the ide controller chipset? (other than hopeless)

> But if I boot from harddisk the aha152x is not there.

in /etc/lilo.conf
add this:

append="aha152x=0x340,11"

If there is already an append statement, modify it thus:

append="whatever_was_there_before aha152x=0x340,11"

then run lilo, then reboot.

Andrew


Re: seg faults while building kernel

1999-10-19 Thread Andrew Hately
"Jacob Schmude
> 
> Hello
> 
> When I try to build a kernel the debian way, I get a seg fault. On my 
> previous linux system this never happened.

Was the previou s linux system on this or other hardware?
If other, then I suspect your current hardware. Read the Sig11 faq.
Often a go-faster option that is enabled pushes the machine into instability
that is clearly evident when the machine is compiling a kernel and less
obvious at other times; another cause can be a bug in the motherboard.
A slow process of elimination is required to determine the precise cause of
the problem; I had trouble with one motherboard which I traced to the bios
option "NA# asserted" _or_ to having more than 64MB of ram fitted. Great,
eh?

Andrew
-- 
http://counter.li.org/cgi-bin/runscript/display-person.cgi?user=45690


Re: hdparm - hard disk tuning

1999-10-18 Thread Andrew Hately
[EMAIL PROTECTED] wrote:
> 
> Hi
> 
> Has anyone had any experience using hdparm to increase the performance of
> your ide hard drives?  Does it work and are there any drawbacks?  Is there
> any software which can measure hard drive performance?

Me
Yes, not that I noticed
hdparm itself and (more convincingly) timing a few big transfers between
partitions or drives, eg a cdrom image

Andrew


-- 
http://counter.li.org/cgi-bin/runscript/display-person.cgi?user=45690


Re: HELP: Free space zero no matter what

1999-10-18 Thread Andrew Hately
"Dwayne C . Litzenberger" wrote:
> 
> Okay, I'm back and running, and I figured out my problem. ext2 filesystems
> have reserved blocks, though I don't know what they're for. It seems only
> root can access them, although I haven't really checked this.  tune2fs can
> lower the number of reserved blocks, but here's my question:
> 
> Why, on a 6.4 GB hard drive, were there 300MB or reserved blocks?  What
> are they for, and do I really need them?

Some kilobytes of space is needed on the / and /var so that root can log in.
Free space of about this size was traditionally reserved for root so that
he/she could still log in to take corrective action even if the disk is
full. In the days when 50MB was a big disk, saving a few percent for this
requirement made sense. Applying these same old percentages - often 5% for
most kinds of unix, 10% on hpux, reserves an excessive amount of space;
often for no advantage. On /home and similar partitions it is my belief that
zero percent need be reserved. On / and if it is a separate partiton /var, a
few (<5) megabaytes reseved is plenty. 
This is all purely subjective.
Andrew


Re: Creating/Moving a partition

1999-10-15 Thread Andrew Hately
[EMAIL PROTECTED] wrote:
> 

> > # mv /var /old_var ; mv /new_var /var
> 
> NO, NO, NO, NO, NO!
> 
> Do NOT use mv!  This will change the ownership and protection rights
> of the files.  That could be very, very bad and will most certainly break
> something.

Oops. I missed out:
First make a perfect copy of /var in another partition mounted on /new_var
using tar, cp, cpio or whatever,
Then swap them as described using mv.

However, your approach seems a little more robust :-)

I'll shut up now.
Andrew
-- 
http://counter.li.org/cgi-bin/runscript/display-person.cgi?user=45690


Re: Creating/Moving a partition

1999-10-15 Thread Andrew Hately
"David J. Kanter" wrote:
> 
> I'd like to make a new partition for /var because I don't have enough drive
> space where /var is currently mounted (/) to run apt-get dist-upgrade.
> What's the best way of doing this? I can create a new /var partition because
> I've got plenty of available drive space, but what set-up files will I have
> to modify to make sure things go smoothly?
> 
> I assume I'll have to change fstab. But then do I have to move all contents
> of the current /var to the new /var? Could I create the new partition using
> a temporary name, move all the /var stuff to it, then rename it to /var?

Thats what I would suggest.
Stop everything you can, copy the contents and do the final swapover in one
line, just in case
# mv /var /old_var ; mv /new_var /var

Another alternative is to create a whole new, bigger, root partition, copy
everyting to that, boot it from a rescue floppy and once there 
1) make it bootable, fix up /etc/fstab and /etc/lilo.conf and run lilo
2) remove the original root partition and replace it with an enlarged
verion.
It depends what else you've got on the disk, how much time you have, etc.

Andrew


2.0.38 make install - unable to find lilo footprint

1999-10-15 Thread Andrew Hately
When I make "install" for the kernel 2.0.38 - which I didn't get from a
.deb, I just downloaded from Finland - when it comes to the part where every
other kernel says "shall I run lilo?" this one says it can't find a lilo
footprint.
What is it looking for?
Andrew


-- 
http://counter.li.org/cgi-bin/runscript/display-person.cgi?user=45690


Re: Sun classic

1999-10-11 Thread Andrew Hately
Oki DZ wrote:
> 
> Andrew Hately wrote:
> ...
> > I had intended to use it as an xserver for the rest of the machines on my
> > network but the screen is so dim and unsharp I'm going off the idea.
> 
> How do you set up xservers?

I think I read a note about this on the sparc linux page; something like 
1) make a device for the mouse with mknod
2) install the sun X server binary.
3) startx
4) optionally install olvwm, xdm, etc

> ie: when you are logged in, actually you
> logged in into the server machine (not the one that's running xserver).

sorry?

> Have you tried Java? You can write distributed apps with ease. I think
> Sun's JDK + ObjectSpace's Voyager will help a lot.

I'm more interested in Ada; have you read the distributed systems annex? It
eats other attempts at distribution for breakfast. It also produces binaries
or (soon) java byte code at the flick of a compile time switch.
(See ACT's JGNAT and Intermetric's Applet Magic).

Andrew


http://counter.li.org/cgi-bin/runscript/display-person.cgi?user=45690


RE: Three questions

1999-10-11 Thread Andrew Hately
> > The poor quality of documentation, such as these "how tos"
> is one of the
> > reasons why I recommend when ever I am asked whether a site
> should move
> > to linux to recommend that they don't.
>
> Although I didn't agree with the tone of this message, I
> definitely think something
> needs to be done about Linux documentation, especially if it
> is expected that
> people switch to it from Windows/Mac (my personal opinion is
> that they *should*
> do that) where a lot of things are tucked away and done
> behind their backs.

For me this is a dangerous road to go down. It may be that to "dumb down"
Linux enough that it can be used by a larger group of less computer literate
people might alienate the current users. I believe one flaw in the Microsoft
Windows approach is that the inherent complexity in configuration can be
hidden but never eliminated; the poor user is left wondering what the
computer might be doing and why.
In Linux this complexity is in your face; but in my opinion that's the
safest place for it to be.

But I agree, Linux needs better documentation and some more tools to manage
the configuration in a canonical way; preferable across a network of similar
machines.

Andrew


Re: Sun classic

1999-10-06 Thread Andrew Hately
Oki DZ wrote:
> 
> Ben Collins wrote:
> > IMO, the Debian install is a lot simpler. As I said Debian supports what the
> > kernel supports, which is any sun4c, sun4m, and sun4d (we also support 
> > sun4u,
> > ultrasparc, with the proper kernel). SPARC Classic is a sun4c, the same cpu
> 
> Excellent, so basically I can install Debian on my Sun.
> BTW, uname -a says that the machine is sun4m; what kind of Sun is that?

There is a sun hardware faq maintained by James Birdsall, try searching for
it; google pointed me to http://www.aball.de/~wpv/sun/faq/hwref0.html but
this might be a mirror. From this faq I found out quite a lot about my
sparcstation2 without even opening the box.

Andrew


Re: fax page size

1999-10-06 Thread Andrew Hately
[EMAIL PROTECTED] wrote:
> 
> Hi
> 
> i am using mgetty/sendfax to send faxes but unfortunately it sends the
> faxes in a stretched format... Where abouts can i change it so that it
> faxes as A5?

I think there's a howto somewhere that explains the tif format used for
faxes has a different scales in x and y.
Each pixel is twice as high as it is wide; which could bring about the
stretching you mention.
This odd pixel shape is a form of data compression that is based on the
premise that faxed mesages usually contain printed text. 

Andrew

-- 
http://counter.li.org/cgi-bin/runscript/display-person.cgi?user=45690


RE: Sun classic

1999-10-05 Thread Andrew Hately
> Mine, as I may have said, is a wee bit slow - the 40MHz 
> prcoessor gives 40
> bogomips. That

that being the slowness, not the 40 bogomips.

> is probably down to the 1992 vintage hard disk 


Re: Sun classic

1999-10-05 Thread Andrew Hately
Oki DZ wrote:
> 
> Andrew Hately wrote:
> > I tried both the 2.2.1 kernel rescue disc and the 2.0.36 on my
> > sparcstation2; the newer one didn't seem to be able to eject the floppy when
> > the next one was needed.
> 
> Interesting... then how did you proceed?

ctrl-alt-del is understood by linux on the sun too, amusingly
When I got back to the boot prompt (stop-A might have got me straight there)
I ejected the 2.2.1 based floppy using the boot prompt command "eject" and
tried again with the 2.0.36 floppy

> As far as I know, Suns don't
> have  eject buttons on their floppy drives.

I can confirm that.
 
> > Otherwise, its just like the i386 install, as Ben said.
> 
> It would be different with installing Slackware on x86 I suppose...
> 
> > To find out how to get the sun to boot from floppy, the debain install pages
> > guide you to sun's boot prompt web pages. Something like "setenv
> > boot_device=floppy" is needed; but check this - I did it once, months 
> > >ago...
> 
> Is it for Solaris or Debian? (or both?)

No, this is for the sun's firmware. Like an i86 box's bios setup but using a
command line. You can find an online manual on sun's site.
 
> BTW, why would anyone install the Debian on Suns if the machines come
> with
> Solaris right out the boxes? 

Well linux is said to be faster.
I did it because I got an old sparcstation2 for nothing with a partial
install of sunos 4.1.something (there had once been an external disc pack
containing the rest) and absolutely no documentation and more importantly,
no licence to run sunos. I also had a couple of i86 debian boxes and though
I had a better chance of being able to do something constructive on the
sparc machine if I knew what was installed where.


> The only reason I'm asking about Debian on
> Sun is that I think it's an interesting alternative since the Classic
> where I have an access to doesn't have Solaris on it (the CDs I mean).
> 
> Oki

Go for it.
Mine, as I may have said, is a wee bit slow - the 40MHz prcoessor gives 40
bogomips. That is probably down to the 1992 vintage hard disk and the fact
that it loads lots of quite important stuff over 10base2 ethernet from my
file server.
I had intended to use it as an xserver for the rest of the machines on my
network but the screen is so dim and unsharp I'm going off the idea. My next
plan is to install glade (as in gnat; i.e. the distributed systems annex
implementation by ACT) on it and see if I can write a distributed
application on my hetrogeneous network. But thats after I've done a hundred
other jobs.

Andrew
-- 
http://counter.li.org/cgi-bin/runscript/display-person.cgi?user=45690


Re: Sun classic

1999-09-30 Thread Andrew Hately
Ben Collins wrote:
> 
> On Thu, Sep 30, 1999 at 12:35:40PM +0700, Oki DZ wrote:
> > Hi,
> >
> > I'm interested in installing Debian Linux on a Sun Classic.
> > Any pointers will be appreciated.
> >
> > Thanks in advance,
> > Oki
> 
> There is an install text in /debian/dists/slink/main/disks-sparc. Most
> likely you will want to download the rescue, driver and root images with
> the 2.2.1 kernel image. After booting the rescue disk, the install is
> pretty much the same as an i386 install.

I tried both the 2.2.1 kernel rescue disc and the 2.0.36 on my
sparcstation2; the newer one didn't seem to be able to eject the floppy when
the next one was needed.
Otherwise, its just like the i386 install, as Ben said.
To find out how to get the sun to boot from floppy, the debain install pages
guide you to sun's boot prompt web pages. Something like "setenv
boot_device=floppy" is needed; but check this - I did it once, months ago...
Andrew

-- 
http://counter.li.org/cgi-bin/runscript/display-person.cgi?user=45690


Re: DriveReady SeekComplete Error and DriveStatusError

1999-09-29 Thread Andrew Hately
B. Szyszka wrote:
> 
> Like I said, I can't afford to

OK, but for me it remains the only convincing test.
As it happens I have a couple of old pentium 75 motherboards at home and
I just imagined that maybe you or somebody you knew also had stuff like
that that you could use for a test, temporarily.

Otherwise, I'd start down the "turn off go faster modes until it works
reliably" track as already suggested; the same approach can fix Sig-11s.
Check out the tool hdparm as well as tweaking the dma in the kernel.

Andrew


Re: Serial connection to windoze box

1999-09-28 Thread Andrew Hately
Phil Brutsche wrote:
> 

> The best way of doing things, I think, is ethernet.  I don't know how much
> things cost in Germany, but here in the US you can get a "home networking
> kit" for about US$70.  It includes two PCI cards, a hub, and two ethernet
> cables.

Used ethernet cards cost pennies and Windows 95 usually already has a driver
for them. If you can track down some 50 ohm terminators, these rest is
usually available for next to nothing.

Andrew

http://counter.li.org/cgi-bin/runscript/display-person.cgi?user=45690


Re: hd to hd copy?

1999-09-28 Thread Andrew Hately
Seth R Arnold wrote:
> 
> On Mon, Sep 27, 1999 at 10:27:31AM +0200, Andrew Hately wrote:
> > # ( cd / ; tar cf - bin boot dev lib sbin usr var ) | tar xf -
> xfp -

I forgot etc in the list.

Andrew
-- 
http://counter.li.org/cgi-bin/runscript/display-person.cgi?user=45690


Re: DriveReady SeekComplete Error and DriveStatusError

1999-09-28 Thread Andrew Hately
"B. Szyszka" wrote:
> 
> Well I can't afford to just go out and get a new harddrive, especially since
> the two that I have no have more than enough space. Is there a way to
> run a check on the harddrive that could prove whether or not the harddrive
> itself is the problem?

The surest test would be to use it with another motherboard for a while.

-- 
http://counter.li.org/cgi-bin/runscript/display-person.cgi?user=45690


Re: hd to hd copy?

1999-09-27 Thread Andrew Hately
tf wrote:
> 
> hey guys,
> 
> I might put in a 6 gig drive in place of the 1.6 gig drive I'm using
> now.  My current installation works fine (its me thats "broken"!). is
> there a way that I can copy this hd to the new one?  I only have a swap
> and a root partition on this drive, and will probably want to partion
> the 6 gig drive further...
> 
> or should I just bite the bullet, and do a fresh install?  (I am getting
> some good practice)

I was surprised by all the responses to this - I've done it a few times
without problems.

Get your rescue floppy disk out and keep it handy. Connect the new hard disk
somewhere on your machine somwhere other than /dev/hda. Run fdisk on it and
create the partition structure you want. Make the appropriate partition
bootable (first, generally). Reboot - its safer to always reboot after
running fdisk. Make the file systems. If any are large, make the swaps first
and "swapon" by hand as mke2fs eats memory. Mount your nominated root
partiton as /mnt. Go there. Create any directories where you will mount
other partitions on the same drive. Mount those other partitions. Create the
proc, mnt, tmp and cdrom directories too. Then from /mnt try something like
# ( cd / ; tar cf - bin boot dev lib sbin usr var ) | tar xf -
This takes a while.
(If your linux image is lying around in / instead of being in /boot, you
need to copy that too.)
Change your future /etc/fstab (currently /mnt/etc/fstab) to mention your new
partitions, all relative to /dev/hda. Shut down. Reconnect the new disk as
/dev/hda, boot using the rescue floppy; rescue root=/dev/hda, run lilo,
shutdown, eject the floppy, boot.

However, a reinstall will probably bring less crud with it.

Andrew

http://counter.li.org/cgi-bin/runscript/display-person.cgi?user=45690


Re: awk or sed?

1999-09-23 Thread Andrew Hately
Tim Thomson wrote:
> 
> Hi,
> ... Was it awk or sed, or am I going along the wrong track? ...

Linx Focus this month has an awk tutorial

http://www.linuxfocus.org/English/September1999/article103.html

Andrew


-- 
http://counter.li.org/cgi-bin/runscript/display-person.cgi?user=45690


Re: page layout app

1999-09-22 Thread Andrew Hately
Seth R Arnold wrote:
> 
> On Tue, Sep 21, 1999 at 10:52:50PM -0700, Mark Wagnon wrote:
> > On Wed 09/22/99 01:39AM, Peter Mickle wrote:
> > > does anyone have any recommendations for a page layout/type formatting,
...
> > How about TeX/LaTeX?
...
> *maybe* lyx,
...
LyX is kind of wysiwig - and is happy to include encapsulated postscript or
strings of LaTeX commands for box-arrow-box figures, etc. The included LaTeX
is not rendered; to view it use the "preview" facility.

http://la1ad.uio.no/lyx/

Andrew

-- 
http://counter.li.org/cgi-bin/runscript/display-person.cgi?user=45690


Re: Sound Apps only work as root??

1999-09-20 Thread Andrew Hately
John Foster wrote:
> 
> Sorry folks I mean the apps do not work. Sound is fine. The applications
> do not work either from xterm or from Ice-wm running Gnome. They are on
> the menus but nothing happens when I click on them.

Did you make /dev/dsp* /dev/sequencer* /dev/mixer* /dev/midi* etc all usable
by the right groups and users are you last rebuilt the kernel?

Andrew


metafont

1999-09-20 Thread Andrew Hately
Is there a debian metafont distribution?
Andrew


Re: dselect error

1999-09-20 Thread Andrew Hately
Andrew Hately wrote:
> 
> Kenneth Litko <[EMAIL PROTECTED]> wrote:
> >
> > Ok, I keep getting this error from dselect
> > and it is driving me nuts.  Here's
> > the error:
> >
> > > internal error - no filename at -e line 12,  chunk 13
> > >
> > > installation script returned error status 1.
> > > Press RETURN to continue.
> 
> Me too, except I get it at chunk 28.

More:

Its the file /usr/lib/dpkg/methods/disk/install that is breaking about line
12 of some inlined perl.
The problem seems to be with a file /var/lib/dpkg/predep-package (something
like that, I left my notes at home).
If you simply find the .deb mentioned in predep-package, install it manually
with 
# dpkg -i blah_blah_blah.1_2.3.deb
then delete /var/lib/dpkg/predep-package.
Dselect's Install method should work after that.
I had this stall for a few different packages; exact names on a piece of
paper at home, I'm afraid.
Full report tomorrow.
Andrew


dselect error

1999-09-18 Thread Andrew Hately
Kenneth Litko <[EMAIL PROTECTED]> wrote:
>
> Ok, I keep getting this error from dselect
> and it is driving me nuts.  Here's
> the error:
> 
> > internal error - no filename at -e line 12,  chunk 13
> >
> > installation script returned error status 1.
> > Press RETURN to continue.


Me too, except I get it at chunk 28.

I've got a heterogenious network at home with a sparc station 2 and some
x86 machines. (40 bogomips on the sparc, but it feels like a lot fewer.)
I bought a couple of slink distributions from tree.uk.com and merged
them on my file server and am attempting to install from there. The
200MB disk X-server-only sparc installed OK, and my 200MB disk 486
router-only installed OK. When I try to install on my 'main' machine I
get the above failure message. Installing from the CDs seems to work
better though the failure occurs in an as yet unpredictable way. On my
fourth attempt to get it going I reinstalled everything, selected a
minimal initial configuration, and then began selecting packages one by
one and installing them. All went well till I selected apache which
depends on a slew of other stuff - I suspect one of the .deb's is
corrupt - but as soon as the install method tries to install the twenty
or so files, it falls over with the above.
I've looked at the dselect and dpkg documentation and am curious to look
inside /var/lib/dpkg/methods to see if I can turn on some kind of
debugging flag to get some slightly more useful output from the "mounted
install" method. However I don't know any perl (yet) so I might not be
able to achieve much.
Anyone else seen this?
Anoyone got a clue whats going wrong?
Anyone recommend a "learning perl" book for a busy software engineer
with a strong regard for Ada and a very low opinion of c, c++, java, apl
and other languages which accept bursts of line noise as syntactically
valid.
thanks in advance
Andrew