Is disk-overflow-partition viable?

unruh · External Since: Nov 14, 2011 Posts: 98

On 2012-04-16, wrote:
> In article , floyd.RemoveThis@apaflo.com (Floyd L. Davidson) wrote:
>
>> Richard Kettlewell wrote:
>> > writes:
>> >> My main data partition is near full.
>========
> Ok, this is what I've found out and what I plan to do:-
> The <transparent overflow> is apparently called LVM;
> but it seems that few users know about it or use it, and
> I'm not happy that I started, but didn't finish using it.
>
> Whereas the "buy yourself a new one" is a simplistic
> suggestion, similar to the idea that you can always buy
> some snake-oil solution for your health, instead of managing
> your life-style, I'll go with Floyd Davidson's suggestion:
> get a BIG one and keep the old disks as backup; and migrate
> incrementally.

What are you talking about? If you run out of disk space, buying more
disk space is NOT snake oil. It is a solution to your ptoblems. Just as,
if your car runs out of gas, buying more gas is a solution, not snake
oil. Now you may also want to conserve gas by driving less and by
driving more carefully, which is fine.
And exactly how is "buy yourself a new one" different from "gget a BIG
one"?
And just how is "migrate incrementally" a solution, except one that is
almost certain to give you headaches when you forget just what you have
already migratied, what is new and what is old and where it is stored.
Just use rsync and migrate it all. Save the old one for a while and then
when you find that you moved everything, then erase it and use that
disk for backup (surely you have a backup already!!-- if not who are you
to lecture others on lifestyle?)

>
> This `fdisk -l` extract gives a glimse into the complexity
> of the situation: ...
> /dev/hdd38 3167 3239 586341 83 Linux
> /dev/hdd39 3240 3385 1172713+ 83 Linux
> /dev/hdd40 3386 3401 128488+ 4c Unknown
> /dev/hdd41 3402 3417 128488+ 4c Unknown
> ...

So? Yes, it is time that you consolidated and stopped carving up your
disks into teeny weeny partitions.( exactly why did you do that in the
first place?)

>
> I've just refound a vitally important letter dating from 1999
> on hdc37, relating to the incompetent and corrupt confiscation
> of my rental property, here in 'new-South-Africa'.

Good. And?

>
> So "Buy yourself a new one" doesn't solve historic problems.
> The reason why I've got plenty of type "4c" partitions, is
> that linux text editors are simply not good enough.
> So if I've got heavy-cognitive-load editing to do, I
> use ETHOberon. LEO: the linux based version, which I
> can now use, only came out after I had already needed to
> create one-partition-per-topic for Native-oberon [ie.
> running natively on x86].

When was that? 1985?
How in the world did "one partition per topic" ever work?
And what has that to do with Linux text editors? I have used Linux text
editors since 1990 and never felt called upon to do anything like "one
partition per topic"?

>
> I've been using linux/`wily` a lot lately. Which is based on
> Plan9's acme; which got it's ideas from ETHOberon, which
> apparently got ideas from <Parc ?>. But for complex:
> multi-copies-on-screen, cutNpaste, SearchNreplace,
> recolourTextStretches ..etc. without having to bob your
> head up-and-down between screen and keybrd, the
> ETHOberon family is needed.

No idea what yo uare talking about.

>
> A further important reason why "just buy yourself a new
> one" can be disasterous [you can't buy a new brain to
> expand your decades of aquired knowledge], is that
> countless files "know about" eg.
> /mnt/p11/Legal/CivilProc/OnThePapers
> which is a file on partition11 of the 80GB disk.
> And its name, recorded in other files can't change from
> "/mnt/p11/...".

It is called soft links. And that is another reason why your previous
procedure had some difficulties, shall we say.

>
> So I foresee a lot of problems.
> But if it works, I'll buy a second one and do regular
> backups as Davidson suggests
>
> The other thing that confuses me:
> I bought a Win7 netbook, because I mistakenly though
> that linux couldn't handle my fixed-wireless-phone;
> and `fdisk -l` shows sda1,2,3,4.
> So it seems that the new PC's don't use IDE ?

No. Linux switched so that all of the disks are now treated as sd disks
(esp the sata disks-- it may be that ata disks still get called hda)

>
> So, can you buy a disk that runs on the old 'IDE PC'
> and will be usable on a new PC?

If the old pc uses ATA and the new one only uses SATA, then no.

>
>
>== TIA.
>

Robert Riches · External Since: Sep 10, 2010 Posts: 42

On 2012-04-15, joop g wrote:
> Kenny McCormack wrote:
>
>> In article , unruh
>> wrote: ...
>>>No idea why the above is "error prone"
>>>PS. do not erase the stuff that was on the /old/data/partition before
>>>until you are sure everything works.
>>
>> I think you've just answered your own question...
>>
>> (In case this was too cryptic for you, chew on this: Whenever you copy
>> data from one place to the other, there's always the risk that something
>> goes wrong in the copy - and it is not always easy to determine if this
>> has happened or not)
>>
>
> Well, as far I know, errors in copying are quite rare. And if you want to be
> quite sure, you can also do a diff -r to check them...

That won't catch differences in metadata: file ownership,
permissions, and so forth. Doing

tar -cf - -C old-dir | tar -df - -C new-dir

can catch those differences, too--in case you're paranoid.

--
Robert Riches
spamtrap42 RemoveThis @jacob21819.net
(Yes, that is one of my email addresses.)

no.top.post · External Since: Oct 20, 2009 Posts: 71

In article , unruh wrote:

> On 2012-04-16, wrote:
> > In article , floyd.TakeThisOut@apaflo.com (Floyd L. Davidson) wrote:
> >========
> > Ok, this is what I've found out and what I plan to do:-
> > The <transparent overflow> is apparently called LVM;
> > but it seems that few users know about it or use it, and
> > I'm not happy that I started, but didn't finish using it.
> >
> > Whereas the "buy yourself a new one" is a simplistic
> > suggestion, similar to the idea that you can always buy
> > some snake-oil solution for your health, instead of managing
> > your life-style, I'll go with Floyd Davidson's suggestion:
> > get a BIG one and keep the old disks as backup; and migrate
> > incrementally.
>
> What are you talking about? If you run out of disk space, buying more
> disk space is NOT snake oil. It is a solution to your ptoblems. Just as,
> if your car runs out of gas, buying more gas is a solution, not snake
> oil. Now you may also want to conserve gas by driving less and by
> driving more carefully, which is fine.
> And exactly how is "buy yourself a new one" different from "get a BIG
> one"?

From my `cfdisk` extract you can see that I've often got up to
50 partitions on a IDE. The total IDE is probably less than 25%
filled. So the idea [completely different from your's car's SINGLE
gas tank] of <virtually expanding any of the partitions> seemed
attractive.

The one that has filled up is a hdx11.
I expect not to have a good reason to create many small
partitions in future; but the old data must live.

BTW Ionce found an interesting script which called fdisk
iteratively, and output to help fix hdx26's partition table
which of course made ALL other > 26 unreachable!

> And just how is "migrate incrementally" a solution, except one that is
> almost certain to give you headaches when you forget just what you have
> already migratied, what is new and what is old and where it is stored.
Sure, it's too error prone.

> Just use rsync and migrate it all. Save the old one for a while and then
> when you find that you moved everything, then erase it and use that
> disk for backup.
No, I'd just keep them as the 2012 status.
What you can do [because you don't want to destroy your
familiar classification arrangement [ontology], is to
appropriately increase the space for classifications that have
shown they needed more than you previously anticipated.

> So? Yes, it is time that you consolidated and stopped carving up your
> disks into teeny weeny partitions.( exactly why did you do that in the
> first place?)
You don't want to know

> >
> > I've just refound a vitally important letter dating from 1999
> > on hdc37, relating to the incompetent and corrupt confiscation
> > of my rental property, here in 'new-South-Africa'.
>
> Good. And?
So, keeping access to the old texts is most important.
And part of that is not destroying the old classification
knowledge. The most valuable component is your
knowlwdge and you can't re-BUY that.

> > A further important reason why "just buy yourself a new
> > one" can be disasterous [you can't buy a new brain to
> > expand your decades of aquired knowledge], is that
> > countless files "know about" eg.
> > /mnt/p11/Legal/CivilProc/OnThePapers
> > which is a file on partition11 of the 80GB disk.
> > And its name, recorded in other files can't change from
> > "/mnt/p11/...".
>
> It is called soft links. And that is another reason why your previous
> procedure had some difficulties, shall we say.
Yes `ln -s` can be nice, but one reason why I've not used that
is that I've used the IDEs like floppies: exposed on their cables,
and I transfer some when I move between locations.

> > The other thing that confuses me:
> > I bought a Win7 netbook, because I mistakenly though
> > that linux couldn't handle my fixed-wireless-phone;
> > and `fdisk -l` shows sda1,2,3,4.
> > So it seems that the new PC's don't use IDE ?
>
> No. Linux switched so that all of the disks are now treated as sd disks
> (esp the sata disks-- it may be that ata disks still get called hda)
Well I've just tested DebEtch on a Win7 netbook and `fdisk -l`
shows all sdx, but grub only sees/wants hdx.
+++++++++++++++++++

A related topic what should be simple: lilo/grub; because it's only
got 3 concepts;
1. the device that BIOS jumps to initially [AFAIK, in old times it
then jumped to the first bootable partition, of 4 or less]
2. the partition which is "/"
3. the path/fileName of krnl & initrd
yet the docos are thousands of lines of confusion.
lilo even wants to pretend it needs Latex documentation
[for artistic effect?]. What is wrong with these clowns ?
Have they never head of 'hello world', where you start with
the minimal demo, and only then incrementally refine it!

unruh · External Since: Nov 14, 2011 Posts: 98

On 2012-04-17, wrote:
> In article , unruh wrote:
>
>> On 2012-04-16, wrote:
>> > In article , floyd.DeleteThis@apaflo.com (Floyd L. Davidson) wrote:
>> >========
>> > Ok, this is what I've found out and what I plan to do:-
>> > The <transparent overflow> is apparently called LVM;
>> > but it seems that few users know about it or use it, and
>> > I'm not happy that I started, but didn't finish using it.
>> >
>> > Whereas the "buy yourself a new one" is a simplistic
>> > suggestion, similar to the idea that you can always buy
>> > some snake-oil solution for your health, instead of managing
>> > your life-style, I'll go with Floyd Davidson's suggestion:
>> > get a BIG one and keep the old disks as backup; and migrate
>> > incrementally.
>>
>> What are you talking about? If you run out of disk space, buying more
>> disk space is NOT snake oil. It is a solution to your ptoblems. Just as,
>> if your car runs out of gas, buying more gas is a solution, not snake
>> oil. Now you may also want to conserve gas by driving less and by
>> driving more carefully, which is fine.
>> And exactly how is "buy yourself a new one" different from "get a BIG
>> one"?
>
> From my `cfdisk` extract you can see that I've often got up to
> 50 partitions on a IDE. The total IDE is probably less than 25%
> filled. So the idea [completely different from your's car's SINGLE
> gas tank] of <virtually expanding any of the partitions> seemed
> attractive.

You can of course do what you want. Why anyone would put 50 partitions
on a disk is completely beyond me. As you discovered you will always run
out of room on one of the partitions when all the others are almost
empty.

>
> The one that has filled up is a hdx11.
> I expect not to have a good reason to create many small
> partitions in future; but the old data must live.

So copy it all into separate directories on a single new partition.

>
> BTW Ionce found an interesting script which called fdisk
> iteratively, and output to help fix hdx26's partition table
> which of course made ALL other > 26 unreachable!

Wasn't that fun. You presumably also enjoy apeing Monty Python's
Ministry of funny walks.

>
>> And just how is "migrate incrementally" a solution, except one that is
>> almost certain to give you headaches when you forget just what you have
>> already migratied, what is new and what is old and where it is stored.
> Sure, it's too error prone.
>
>> Just use rsync and migrate it all. Save the old one for a while and then
>> when you find that you moved everything, then erase it and use that
>> disk for backup.
> No, I'd just keep them as the 2012 status.
> What you can do [because you don't want to destroy your
> familiar classification arrangement [ontology], is to
> appropriately increase the space for classifications that have
> shown they needed more than you previously anticipated.
>
>> So? Yes, it is time that you consolidated and stopped carving up your
>> disks into teeny weeny partitions.( exactly why did you do that in the
>> first place?)
> You don't want to know
>
>> >
>> > I've just refound a vitally important letter dating from 1999
>> > on hdc37, relating to the incompetent and corrupt confiscation
>> > of my rental property, here in 'new-South-Africa'.
>>
>> Good. And?
> So, keeping access to the old texts is most important.
> And part of that is not destroying the old classification
> knowledge. The most valuable component is your
> knowlwdge and you can't re-BUY that.

I have never advocated getting rid of the data. Just organising it more
rationally.

>
>> > A further important reason why "just buy yourself a new
>> > one" can be disasterous [you can't buy a new brain to
>> > expand your decades of aquired knowledge], is that
>> > countless files "know about" eg.
>> > /mnt/p11/Legal/CivilProc/OnThePapers
>> > which is a file on partition11 of the 80GB disk.
>> > And its name, recorded in other files can't change from
>> > "/mnt/p11/...".
>>
>> It is called soft links. And that is another reason why your previous
>> procedure had some difficulties, shall we say.
> Yes `ln -s` can be nice, but one reason why I've not used that
> is that I've used the IDEs like floppies: exposed on their cables,
> and I transfer some when I move between locations.
>
>> > The other thing that confuses me:
>> > I bought a Win7 netbook, because I mistakenly though
>> > that linux couldn't handle my fixed-wireless-phone;
>> > and `fdisk -l` shows sda1,2,3,4.
>> > So it seems that the new PC's don't use IDE ?
>>
>> No. Linux switched so that all of the disks are now treated as sd disks
>> (esp the sata disks-- it may be that ata disks still get called hda)
> Well I've just tested DebEtch on a Win7 netbook and `fdisk -l`
> shows all sdx, but grub only sees/wants hdx.
> +++++++++++++++++++
>
> A related topic what should be simple: lilo/grub; because it's only
> got 3 concepts;
> 1. the device that BIOS jumps to initially [AFAIK, in old times it
> then jumped to the first bootable partition, of 4 or less]

Nope that was Windows, not Linux. Linux cold always (within the
limitations of the bios) boot from anywhere on the disk
(The problems with the bios was that in order to load the b ootloader or
the kernel, it needed to use the raw sector reading ability of bios,
since no kernel existed at that point in the boot cycle to get at the
full disk)

> 2. the partition which is "/"

Nope. /boot That could be on / or on another partition altogether.

> 3. the path/fileName of krnl & initrd

/boot/vmlinuz or /boot/initrd if that is what you want to use. You do
not need an initrd, but then must make sure that all modules needed for
booting are compiled into the kernel, and not as modules.

> yet the docos are thousands of lines of confusion.
> lilo even wants to pretend it needs Latex documentation

???? No idea what youare talking about. lilo does not need anything of
the sort. compiling the lilo package may want it to latex the docs, but
that is a different issue.

> [for artistic effect?]. What is wrong with these clowns ?
> Have they never head of 'hello world', where you start with
> the minimal demo, and only then incrementally refine it!

???

Somehow given your procedures, I doubt that you should be the one to give
advice to others.
>
>
>

no.top.post · External Since: Oct 20, 2009 Posts: 71

In article , unruh wrote:

> On 2012-04-17, wrote:
> > In article , unruh wrote:
> >
> You can of course do what you want. Why anyone would put 50 partitions
> on a disk is completely beyond me.
I don't believe it IS beyond you.
It's just that you're comfortable not thinking beyond your experiences.
Which could be good for a stable society.

> As you discovered you will always run
> out of room on one of the partitions when all the others are almost
> empty.
You don't need to discover.
Logic tells you that the Micro$loth way of having ONE BIG C:
means eliminating the <fragmented pieces problem>.
And back to your analogy of the fuel tank: you could put all the
liquids in your household in one container.

J G Miller · External Since: Dec 08, 2004 Posts: 360

On Friday, April 20th, 2012, at 06:50:44h +0000, Chris Glur suggested:

> And back to your analogy of the fuel tank: you could put all the
> liquids in your household in one container.

So it is not a good idea to keep all your partitions on the same disk.

Every computer should have two.