-
Posts
409 -
Joined
-
Last visited
Content Type
Profiles
Forums
Events
Posts posted by banjo
-
-
I am trying to create a new boot disk which is a
copy of my current, smaller disk. I have not cloned
the disk because the new disk is larger and I have
restructured the partitions to make the new disk a
bit more friendly to the way I want to use the system.
So I am constructing the new disk manually.
Here is the partition info for the new disk:
part1 is /
part5 is swap
part6 is /home
part7 is /usr
Here is what I did:
I installed the new drive as the primary IDE slave
along with the primary IDE master in order to do the
copies, etc. so the new disk showed up as /dev/hdb.
I booted knoppix.
I partitioned the new disk (cfdisk), made part1 bootable,
made file systems, and copied all of the files.
Everything looks to be in good order.
I then attempted to install lilo on the new disk.
> umount /dev/hda (just to be safe)
> mkdir /mnt/newroot (shows up in the knoppix RAM disk /mnt)
> mount -w /dev/hdb1 /mnt/newroot
> mount -w /dev/hdb6 /mnt/newroot/home
> mount -w /dev/hdb7 /mnt/newroot/usr
> cd /mnt/newroot
> sbin/lilo -v -v -r /mnt/newroot
Everything seemed to go well, and I was informed that
the Master Boot Record had been written. I then
opened the box and attached the new hard drive as the
only drive in the system and as primary IDE master,
jumpers set appropriately.
I checked in CMOS setup and the new disk was recognized
properly.
When I continued the boot, the computer went through the
POST and then froze with a black screen and a blinking
cursor in the upper left corner.
No boot. No errors. No messages. No nothin'.
Nada. Zilch. Black screen of death.
So I googled.
The explanation for the problem that I have found in
the docs is that either no lilo is installed at all
in the MBR or the partition is not "active".
I must be missing something important in how
to use lilo. Here are my questions:
What is an "active" partition?
Is that different from a "boot" partition?
How do I specify the active partition in cfdisk?
I don't see a flag or menu for "active" in cfdisk.
The lilo.conf on the new disk refers to /dev/hda
as the boot disk because that is where the disk will
reside when I reboot. Does lilo.conf have to refer
to /dev/hdb at install time since that is where the
disk sits when I install lilo?
Should I install the new disk as primary IDE master
and *then* boot knoppix and run lilo?
Does it matter?
What does mkswap do? If I have already defined
partition 5 as a swap partition, is that enough?
I don't remember running mkswap.
Many thanks to the fine folks on this board for
all the help. My hope is to know enough one day
to answer these kinds of questions for others.
Linux rocks! (if I can just get it to boot....)
Banjo
(_)=='=~
-
Thanks for the info.
My MOBO is an Intel D845PESV which I bought in May of last year.
I will see if I can find out whether the old cylinder limit
is fixed in it. I want to be sure that I don't run into any problems
in the root directory, if you get my drift.
Mandrake set me up with a root partition of 5.8 Gig, and I have now
used about 3.6 Gig. Last night I did some du's and found out
that 3.3 Gig are used by /usr and about 0.3 Gig are used for
the rest of it.
So I thought that I might make a separate partition for
/usr and set it up as it is now for the rest of it
(with larger partitions, of course). That avoids the mystical
8 Gig limit on root and gets me all the space I need for /usr.
And yes, I plan on putting most of the partitions into a large
extended partition to avoid that yet-another-ridiculous-limit
of four partitions on a disk problem.
Funny story: Way back in the early '80's a friend of mine
was at a football game when a scrolling advertisement came
across the scoreboard. The bank ATM's were brand new then,
and the ad was something along the lines of:
"Baybank ATM's. Now serving 255 locations!"
My friend said that he started laughing so hard he
fell out of his seat. The folks around him could not
understand what was so funny.
We figured that some poor software weenie lost his
job that day............ or got promoted to Management.
Banjo
(_)=='=~
-
The issue with the "small" 8 Gig root is that my son
wants to install some fairly greedy game programs
later. I have heard numbers as large as 3 Gig for a
program.... :o (remember when a 10 Meg drive
was huge?). Two or three of those things and my
"small" 8 Gig root partition is looking full.
So, the issue is not that I cannot get it working without
a huge root partition. The issue is that the root partition
may become huge later, and I don't want to run out of
space.
The reason I flagged /usr for a link to a larger space
is that I was under the impression that most of the
user apps are installed there. Perhaps I am wrong.
I still have some more studying to do about how the
file system is laid out. What I want to do is to link
the directories which are going to grow huge out to
a very large space.
Maybe I need to step back again and go back to
basics.
Does anybody know how I can find out if the
1024 cylinder limit on the boot sector is still an
issue?
I continue to ponder.
Banjo
(_)=='=~
-
Well, I am certainly learning a lot about how Linux
hangs together. This board is a great resource.
I have downloaded a copy of Linux-Filesystem-Hierarchy.pdf
and I am now studying it. When I figure this out and get
it done I will post my final answer on the board.
(Could be a while.... I am a slow reader... and my finger
gets tired). :D
One of the great things about Linux and Unix is
that there are many ways to get the job done. We
are limited only by our creativity.
Banjo
(_)=='=~
-
That makes sense. But I wasn't thinking of mounting it
on a different disk. I was thinking of just creating
a symbolic link of /usr to /home/usr. That should
leave the original /usr intact but unused. If I copied
all of /usr over to /home/usr before creating the link
the system should not see any difference on reboot.
Since all of my partitions will be on a single hdd,
if the hdd dies I am toast anyway.
I am assuming that the OS would not care that the
/usr was actually a link rather than a real directory.
Sometimes the low level stuff can treat a link in
a slightly different fashion.
Banjo
(_)=='=~
-
OK. Here is another crazy idea I had.
We used to link /tmp to /home/tmp when root
was too small for the huge temp files we were
creating. It worked great.
Suppose that I did a smallish (say, 8 Gig) root
partition as the first primary partition and then,
once I get it going I could link /usr and /var to
/home/var to get more space (copying current
contents over, of course).
Has anybody ever tried to skirt the size issue
doing something like that?
Banjo
(_)=='=~
-
Thanks for the feedback.
It was making me nervous to put root and /home on the same
partition, but I wasn't sure why. I suppose that if it ever became
necessary to wipe the / partition it could get ugly.
I did not want to re-install on the new disk in order to avoid
having to reconfigure all the apps. But that may be the
quickest thing to do after all.
I will go do some more research before jumping in.
Thanks again.
Banjo
(_)=='=~
-
I currently have a disk situation which I am trying to
solve. The problem is that my 40Gig disk is making a
noise when it spins up, so I have procured a new
disk that is 120G to replace it. So far, so good.
Mandrake partitioned my 40G disk at install time
in the following manner:
Size Used Avail Use% Mounted on
part1 /dev/hda1 5.8G 3.6G 2.0G 65% /
part2 extended partition
part5 /dev/hda5 swap 494 Meg
part6 /dev/hda6 31G 1.9G 29G 7% /home
This is on a home system used primarily for the
usual email and browsing functions.
So this works fine for me, except that the root
partition is getting a bit full, and my son wants
to install some large games later.
What I want to do is to partition my new disk
in a more friendly fashion, giving more room to
root for these large installs. Then I will copy the
contents of the old disk to the new one and make it
bootable. I have studied lots of tutorials on how to do
this, but I still have some questions for which
I have not found answers. So I thought that I would
post them on this board.
I am thinking of partitioning the new disk in
the following manner:
part1 /dev/hda1 500Meg /boot
part2 extended partition
part5 /dev/hda5 swap 500 Meg
part6 /dev/hda6 119G /
By putting all of root (including /home) on
one partition, I avoid the unanswerable question,
"How big do I make root to avoid running out of space
and yet avoid wasting space I could have used on /home?"
Basically, most of the disk would be a single partition
similiar to (oh GAG!) fnWindows. The root directory can
then use as much space as it needs.
The reason for the small /boot partition is that I read
somewhere that the bootable partition cannot be larger
than about 8Gig to make it all adressable by the BIOS.
That might not be big enough for the memory hogging games.
Hence, it is just /boot, not root.
Here are my questions:
1). Is it still true that the bootable partition
must be less than 8 Gig? I have a new Intel 2 GHz motherboard
procured last May. I understand that the size limit
is a BIOS limitation. Can I simply partition the
disk as the old one is now and just make a large,
say 40G, root partition? If I do it that way, I should
not have to change the config files.
2). Is there an overpowering reason for having /home
on its own partition, separate from / ?
I see nothing but negatives from having to guess how
much space to give the root partition instead of simply
letting it have whatever it needs of the disk.
I do not do system backups of partitions, being of the mindset
that it is quicker and easier to reinstall the system
on a new hard disk than it is to do hours and hours of
system backups over the life of the disk just to save a
couple of hours to reinstall. User files are backed
up separately. Am I messing up if I put root and /home
on the same partition?
3). When I partition the new disk, what do I name the
partitions?
I will partition the new disk installed as /dev/hdb
and then copy the directories from /dev/hda to /dev/hdb,
then install the new disk as /dev/hda, run lilo and
reboot.
/proc/partitions shows the partitions as part1, part2, part5, and part6.
HardDrake shows them as devices hda1, hda5, and hda6.
Do the partition names matter?
I assume that the new disk partitions will be mapped to /dev/hdaX
by the BIOS regardless of partition names when I install the
new disk as IDE0 master? Is this correct?
Thanks in advance for your help. If I get this to
work I will publish what I did on the board.
Banjo
(_)=='=~
-
Thanks for the help.
This will probably never happen again; my Mandy 9.1
is pretty stable. But it is nice to know what to look
for anyway.
Banjo
(_)=='=~
-
I had a strange thing happened on my Mandy 9.1
last night.
I was on line using KPPP on KDE when I closed down
Mozilla, and the launcher bar at the bottom of
the screen went away with it.
Everything was running fine, except that I no
longer had the task bar with the icons in it so
I could no longer de-iconify KPPP to hang up
and log out. I did a ps -ax in the CLI, and
the launcher processes seemed to
be running, and KPPP was still connected and
working. I just wanted to display the KPPP
dialog box so I could hang up.
I tried to kill the KPPP process, to
see if that would hang it up, but it did not
work. So, I logged out, and it hung up.
Problem fixed, but I am curious whether there is
a way from the CLI to cause the launcher to redisplay its
task bar or to force iconified programs to display
their dialog boxes.
Is there a way to tell the KDE WM to do this?
What is the name of the KDE WM process?
This is not a big deal because it was easy to
get around it, but I am curious about how the
desktop hangs together (this KDE is not your
grandfather's WM).
Thanks in advance
Banjo
(_)=='=~
P.S. Linux rocks
-
Last May I built a PC from parts and put Mandrake
9.1 on a virgin disk. This computer has never seen
fnWindows. It runs great.
The Linux computer sits next to a Dell with
fnWin98se on it. We get the blue screen of death
several times every day on the Dell. The Linux
computer has not crashed even once yet since
I built it.
I have had to crack the Dell box several times
to wiggle wires and get it working again. The
hardware in it is crap. The single hard disk is
not mounted in a bay, but bolted to the front
panel, and is connected with a tiny, non-standard
IDE cable that barely reaches it. The floppy drive is
a non-standard config with no facia and no button,
which are integrated into the cheap plastic front
panel. The floppy is also connected with a tiny,
non-standard cable. The mother board and power
supply have been rewired by Dell so that you cannot
replace the mother board with a standard board
without also changing out the power supply or you
will fry the board immediately upon power up.
I am done with fnDell and done with fnWindows.
Linux rocks.
Banjo
(_)=='=~
-
I feel better already.
Mandrake rocks!
Banjo
(_)=='=~
:D
-
But with RedHat pulling their desktop product off
the market, what does that do to Mandrake?
Isn't Mandrake based on RedHat?
Banj(_)=='=~
-
Sorry to hear that you are disappointed.
I paid just over 80 bucks for the whole shootin' match.
And that includes all the apps that I use, some of
which are way better than the fnWindows apps that
I was using on the fnWindows box.
I have run into some bugs, but so far there have been no
showstoppers.
I think that Mandrake was worth every cent that
I paid for it. This does not seem like overcharging to
me when compared to the hundred$ of $$$ that I
have spent on fnMicrocrap that crashes all the time
and spams me with my own equipment.
Just my $0.02
Banjo
(_)=='=~
-
Thanks for the pointer.
I think that the internal workings of this
are starting to gel in my mind (a scary
thought).
Banjo
(_)=='=~
-
So, then all of this is owned by root?
And then you go in and mess around with
the source code as root and rebuild?
That makes me nervous. I don't like to play
around as root. It appears that this rpm system
is designed mostly for grabbing the source and
building a package as is, and not so much for folks
who want to make changes.
Am I still missing something?
(sorry to be such a pain....)
Banjo
(_)=='=~
-
So rebuilding the package extracts the source
from the tar.gz down in /usr/src? I guess I assumed
that installing it would do that.
Ah, me. So much to learn.... so little time.
Thanks for the info. I am not at my Linux box
right now, so I will have to try this when I get
home.
Banjo
(_)=='=~
-
I have a real nOObie question this time.
I downloaded and installed a source rpm
called abcm2ps-3.7.11-1.src.rpm
The install went OK as far as I can
tell. But now I cannot find the source
anywhere.
Where did the source code go?
I suppose that this is in a tutorial
somewhere, but I can't find that either.
If I want to modify the source before
building it, where do I look for it to do
that? Do I have to get a tar.gz?
I know that I must be missing something
obvious, but I am having a brain cramp
or something.
A pointer to a tutorial is fine if there
is one that covers this subject.
Thanks in advance.
Banjo
(_)=='=~
-
It is all about the money.
I dumped fnWindoze for Linux last May.
Since then, my Linux box has never crashed.
The apps are mostly better than M$ apps in MHO.
The viruses are no longer a worry for me.
The documentation about the way the apps
and the OS really work is far better than M$
docs, which cost big $$$$ from M$. I can fix
it myself without re-installing the OS.
... and for all of this I didn't have to give up
a single dime to M$.
... and I will upgrade my OS when I am ready....
not when M$ decides that I owe them some more $$$
... and I don't have to report my hardware serial numbers
to M$ and beg them to allow me to run the OS
that I paid for....... yes... I paid for the Power Pack.....
... and it's worth every dime.
Need I go on?
The corporate world is starting to feel the pain,
so they are trying more dirty tricks to keep
their stranglehold on American greenback.
I am sooo glad that I made the switch.
(sorry for the rant)
Banjo
(_)=='=~
-
Well, I finally got my attachment from the email.
Suspecting that something was wrong with the
headers in the original, I edited the other message
(the one that worked), deleting the base64 code that
was there and replacing it with the base64 code from
the errant message. So, if all is well, I should have
the .pdf data stored in the other message using
the other file name etc.
Then, I used munpack to extract the new data.
It worked fine, so I changed the file name back to
a .pdf file and I can read it using Adobe.
:D
I did not figure out exactly what is wrong with the
MIME headers in the original message. I guess I
won't because I have no idea how to start debugging
something like that.
But I have my file. Do I get to stop being a
NOObie now? :lol:
Banjo
(_)=='=~
-
Just in case anybody else is interested in this
I thought I would post my progress trying to
unpack an errant email with an invisible attachment.
I captured the email in a .txt file so that I can
mess with it. After a few hours of research I
found mpack-1.5-9.i686.rpm and installed it.
This installs mpack and munpack, which are
tools to pack and unpack MIME parts.
These are old CLI tools from the dark ages
of email.
I ran munpack against my text file like this:
~/apps/mpack>munpack raumail.txt Did not find anything to unpack from raumail.txt
This is consistent with what my readers are
telling me................ that the attachment is not there.
But when I open the text file and scroll down
to the section where the attachment resides,
the MIME part is there along with all the base64
codes and 2.2 Meg of junk.
To test a bit further, I sent myself a different
email with an attachment of a zip file. I converted
that email into a text file and ran munpack against
it. It extracted the .zip file just fine.
So, now I am off to compare the two email files
to see why the Mac-generated file thinks that
it has no attachment. They both look very similar
in content.
I am learning a lot about the nitty-gritty of
the innards of email. Pretty scary.
Banjo
(_)=='=~
-
I recently received an email from a Mac computer.
The email has a .pdf file embedded in it in MIME
format, and Kmail will not display it as an attachment.
If I view the email in plain text format, I can see the
MIME section, so I know that it is there. It starts
out like this:
--MS_Mac_OE_3147240150_3640249_MIME_Part-- --MS_Mac_OE_3147240150_3640249_MIME_Part Content-type: application/pdf; name="Rau_Fall_2003.pdf"; x-mac-creator="4341524F"; x-mac-type="50444620" Content-disposition: attachment Content-transfer-encoding: base64 JVBERi0xLjQNJeLjz9MNCjM1NCAwIG9iag08PCANL0xpbmVhcml6ZWQgMSANL08gMzU2IA0v SCBbIDIyMDcgMzM0IF0gDS9MIDE3MTIyODcgDS9FIDM0MTQzNSANL04gNSANL1QgMTcwNTA4 ..... etc...............
After the header I can see the MIME encoded
data.
The other parts of the email are MIME types that
are text/plain and text/html, and Kmail has no
trouble decoding and displaying them.
Does anybody here know if I can use some other
tool to decode this document? I am not familiar
enough with the innards of email and MIME
to figure it out.
Thanks in advance
Banjo
(_)=='=~
BTW, I found out quite by accident (I went for
shift key and bumped the 'v' key...... don't ask)
that Kmail will show the actual text of a message
by selecting the message in the message list
and typing the 'v' key.
Very kewl
-
I learn the coolest things on this board.
Thanks for the tip about /etc/ppp/ip-up.local.
Does it run as root? My computer will not update
the clock unless I am root. I will give it a try
when I am at home (I am forced to use fnWindoze
here at work )
I did have another question about the time servers
though. I went through the list of stratum 2 NTP servers
at the NIST web sites, but every one that I tried to
connect to refused my connection. I am so confused.
I thought that they are supposed to be more
accessible than the stratum 1 servers ???
So, I have temporarily gone back to time.nist.gov,
which appears to accept my query with no problems.
Banjo
(_)=='=~
-
Well, I did install it......... and then I didn't know
where it was...
I did eventually find the links to the directory right
there in the KDE menu under Documentation
So there you go.
Re. the differences in file structures...... being an
old Unix hack I do tend to poke around and cat out
all manner of things just to see what is in 'em.
One of the best places to look is in the /proc directory,
which did not exist on the Unix systems that I first
learned on. A lot has changed since 1976.........
... oops............. did I say that out loud?....
Well, this old dog is learning new tricks every day.
I am soooo glad that I am bailing out of fnWindoze.
Banjo
(_)=='=~
Problem with lilo on a new disk
in Hardware
Posted
Thanks for the info.
I think that my next move will be to try lilo again
with the new disk as /dev/hda. I am suspicious that
my last attempt, where the new disk was /dev/hdb,
somehow wrote the new MBR into the void (or,
horrors, onto the other disk). But at the point where I
ran lilo, /dev/hda was not mounted at all. So, I don't
know where it went.
I have already edited /etc/fstab on the new disk,
so the mounts should happen if I can get lilo to
work.
Being an old Unix hack, my fingers know vi. :D
I don't even have to think about it.
Heck, I have done whole projects using ed, which
begat ex, which begat vi, which begat vim, which
begat gvim, which..................... well you get my point.
BTW, do you know if "active" is the same as
"boot"? My part1 is marked as "boot", but I could
not find "active" anywhere in cfdisk.
Thanks again for the help.
Banjo
(_)=='=~