Re: separate filesystem for a program on OpenServer 5.0.7

• Previous message: Jean-Pierre Radley: "Re: lp error message Reply-To: scomsc@xenitec.on.ca"
• In reply to: J. Graue: "Re: separate filesystem for a program on OpenServer 5.0.7"
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

Date: 10 Jun 2005 19:32:51 -0400

----- Original Message -----
From: "J. Graue" <info@nospam.com>
Newsgroups: comp.unix.sco.misc
To: <distro@jpr.com>
Sent: Friday, June 10, 2005 5:16 PM
Subject: Re: separate filesystem for a program on OpenServer 5.0.7

>
> "Tony Lawrence" <foo@pcunix.com> wrote in message
> news:SfadnapqPqKxcDTfRVn-rg@taconic.net...
>> J. Graue wrote:
>> > Hi, John:
>> >
>> > "John DuBois" <spcecdt@armory.com> wrote in message
>> > news:11agm6u9584t89f@corp.supernews.com...
>> >
>> >>In article <11aelsuslg404bd@corp.supernews.com>,
>> >>J. Graue <info@nospam.com> wrote:
>> >>
>> >>>Hi, all:
>> >>>
>> >>>I have a new system onto which I'd like to load OpenServer 5.0.7. It
>> >
>> > would
>> >
>> >>>be nice to use separate filesystems for the operating system, and for
>> >>>a
>> >
>> > mail
>> >
>> >>>server program that I would like to install on the rest of the drive.
>> >>>
>> >>>When I install 5.0.7, I can limit, I'm sure, the partition into which
> the
>> >
>> > OS
>> >
>> >>>(operating system, here, not OpenServer) resides. Then, I can create
>> >>>another filesystem into which the program and its data will reside, if
>> >
>> > only
>> >
>> >>>I knew how! From my reading of the man pages, I get this:
>> >>>
>> >>>I use fdisk to create a second partition from the unused portion of
>> >>>the
>> >
>> > disk
>> >
>> >>>(IDE, by the way). I then use divvy to create a division? Is this
>> >>>so?
>> >
>> > How
>> >
>> >>>do I create the filesystem in which the program and data will reside?
>> >>
>> >>Your best plan is to create a single partition. During install, select
>> >
>> > the
>> >
>> >>option to do an interactive divvy, and simply create an extra
>> >>filesystem
>> >
>> > on
>> >
>> >>an unused division. You could also leave some unused space in the
>> >
>> > partition,
>> >
>> >>in which case, yes, later you'd use divvy to allocate it to a
>> >>filesystem
>> >
>> > and
>> >
>> >>create the filesystem.
>> >
>> >
>> > In fact, it turned out to be my ONLY plan, since nothing else worked!
>> > I
>> > wasn't able to keep any part of the drive 'unused.' If I tried, things
>> > would go "whacky," like kernel panics, etc. I feel pretty confident
> that
>> > there's nothing wrong with the drive, but, well, I don't know.
> Personally,
>> > I'd guess that it's an error-between-chair-and-keyboard, but I don't
> know...
>> > thanks for your advice.
>> >
>> > Best regards,
>> >
>> > Jim
>> >
>> >
>>
>>
>> You COULD have left it as it was with the second partition. You also
>
> Yeah, but...
>
>> COULD have selected "Interactive Fdisk/Divvy" as an installation option
>> and laid out whatever you wanted before installation.
>
> Yeah, but, maybe because I didn't know how to use divvy well, I couldn't
> figure out how to name the divisions in that partition, etc. I think if I
> had the two partitions in front of me, again, I'd probably be able to
> muddle
> through. The next time I'm doing an OpenServer installation, I might get
> a
> chance to try this. At the time, though, after I read your article, it
> didn't make much sense to have a second partition on the one drive. It's
> just that the documentation for the mail server software that I was about
> to
> install mentioned that it would be best to keep it on a separate
> partition.
> Since it's SCO OfficeServer, I thought they'd know to what they were
> referring, but what they probably meant, as I go back to read it, again,
> is
> that I could get away with just using a separate filesystem, not a
> separate
> partition.
>
>> You CAN'T go back
>> into divvy after an install and change end points downward.
>
> Ah! Now that I've tried it several times and proven that, in fact, one
> cannot use divvy after an install and change end-points downward, I feel
> so

Or upwards, usefully.
Enlarging a divvy section probably doesn't break the filesystem that was on
it, but the filesystem does not grow either.
The space between the end of the filsystem and the new endpoint would become
unuseable.
Before that it was useable space because it could have been the beginning of
the next divvy section and a filesystem could be put on it.

Personally I think divvy has always been a source of unnecessary confusion.
Divvy does too many things at once in the name of automation but I think
it's just counterproductive a lot of the time because it obfuscates the
basic facts and causes a new user to think that they have to do things the
way divvy does it. You need to use it, but also you need to know which parts
of it's actions you can safely disregard or override.

Divvy does 3 main things, none of which have to be done the way it does
them. Indeed, 2, well, 1.5 of the 3 don't necessarily have to be done at all
ever in any way.
Further, some of what it does, even if it's allowed to do it, can be ignored
after the fact.

1) It creates a divvy table which portions out the fdisk partition into up
to 7 sections.
the divvy table functions much like the fdisk table does for the whole disk.

2) It creates various forms of formatting within the sections it created.
That is, it makes filesystems in some, makes others into swap, makes others
into reserved system space.

3)
  3a - it creates /dev/xxx nodes for all the sections
  3b - it makes "names" for the sections and makes more /dev nodes to the
same sections but named after the "names"

There ends up being several ways to refer to some of the exact same features
on the disk.
Some features /dev nodes are named after the filesystem mount point or what
the space is used for. (/dev/boot, /dev/swap, /dev/u)
Some are named after some represetation(s) of it's address. (/dev/dsk/0sd,
/dev/hd0d, (which are the same thing))
And some may have both.

Then, even when you think you have a handle on all that there are
inconsistencies to throw you for a loop.
So you figure out that instead of /dev/dsk/0sc there is /dev/u, "aha, I get
it, the mount point is /u so they name it /dev/u", kinda neat, kinda
sensible in a way, except not, but whatever, accept it an move on.
So, /dev/boot ... bzzt, the mount point is /stand and there is no
/dev/stand. /dev/root is excusable since it's a special case, the mount
point is / and you can't very well name it /dev//
But thats exactly one of the reasons why it's a silly idea naming device
nodes after the filesystem anyways in my not exactly humble opinion.

You don't have to feel bad about being confused by divvy, or by the parent
mkdev hd process which has similar traits though not as bad.
It's full of unnecessarily unclear elements. You can have things arranged
pretty much any way you want them, but you'd never know it from following
the official common directions.

Brian K. White -- brian@aljex.com -- http://www.aljex.com/bkw/
+++++[>+++[>+++++>+++++++<<-]<-]>>+.>.+++++.+++++++.-.[>+<---]>++.
filePro BBx Linux SCO Prosper/FACTS AutoCAD #callahans Satriani

• Previous message: Jean-Pierre Radley: "Re: lp error message Reply-To: scomsc@xenitec.on.ca"
• In reply to: J. Graue: "Re: separate filesystem for a program on OpenServer 5.0.7"
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]