NetNews Usenet Archive 1992 #19/NN_1992_19.iso/spool/comp/sys/sgi/13109

Internet Message Format | 1992-08-31 | 4.4 KB

Path: sparky!uunet!olivea!hal.com!decwrl!mips!mips!odin!sgihub!zola!zuni!anchor!olson
From: [e-mail] (Dave Olson)
Newsgroups: comp.sys.sgi
Subject: Re: How to build a System-Disk ??
Message-ID: <[e-mail]>
Date: 31 Aug 92 22:07:19 GMT
References: <vC+vr*[e-mail]> <[e-mail]> <[e-mail]> <[e-mail]> <[e-mail]> <[e-mail]>
Sender: [e-mail] (Net News)
Organization: Silicon Graphics, Inc.  Mountain View, CA
Lines: 84

In <[e-mail]> [e-mail] (James Gardiner [Hunter]) writes:
| I vote 1 for the Sun partition program over SGI however.

I'd be interested (seriously) in what about it you found easier
to use.  I'm certainly open to suggestions for making fx easier,
particularly in the area of repartitioning.

| >Yes, but swap size is simple.  That has nothing to do with deciding
| >that you want 4 seperate filesystems (which is likely to produce
| >worse performance, by the way), *and* that you wanted to use different

| Why would it produce worse disk performance please?

Because with more filesystems, you are likely to be seeking over
a larger section of the disk.  Of course, if all of them are 100%
full all the time, it is pretty much of a wash (except that the
kernel caches for the directory lookup may fill up and be
replaced more often).

| Anyway, I have seperate partitions because I have users which have no
| consideration on disk space.  I do not quotas personally so I put them
| on there own partition and if they fill it up its there problem.

Then you are pretty much stuck with your solution.

| >partition numbers than the standard partition numbers for the root
| >and user filesystems.

| Well if something told me it HAD standard partition numbers for / and /usr
| then this wouldn't have been an issue anyway.

But of *course* there are standards; surely you don't think we just
use a random number generator!  I can understand that you might not
have known what they were, but I would think that *anyone* changing
partition tables would have assumed that the current values are the
defaults, wouldn't they?

| >| Falling back on defaults does not sound very professional to me.
| >| For example, I pumped up the swap 20-30 meg over default.  Today I wish
| >| I pushed it up another 20-30 over that.
|
| >Again, so what?  Changing swap size is easy under 4.0.1, and has
| >nothing to do with your complaints about partitions.  (You may have
| >problems changing your swap partition size, but you certainly
| >never mentioned it.)

| We are talking 2 different issues here.
| To change your swap size you have to edit the partiton table.
| To do so you have to take so much of that partition and put onto this etc.
| Its not like.  I want x swap where remander y will be /usr.

You are wrong.  This is *exactly* what the repartition menu in fx
does!  It increases swap at the expense of the /usr partition, and
does all the calculations itself.  Did you not use the repartition
menu in fx?  I thought I had made the help and prompts extremely
clear on this; obviously I haven't.  What could have made it
clearer?

| They are annoyances yes but in my opinion, caused by lack of info in
| the manuals.  Annoyances that just should not have happened or at best
| be easier to solve.
| Remember here.  I was comming at this from a TOTALLY ignorate possition.
| I had never used an SGI for more then general playing with demos before
| in my life.

Most people haven't.  I'd be interested, as I think I said, in explict
feedback on what could be improved.  In fact, I think some of the manuals
have a card bound into them asking for such feedback or comments.

| First time users never react like manual writers think they should.

That is a given.  We *try* to guess, but everybody does things just
a bit differently, and looks for information in a different way. We
just have to do the best we can, and adjust the documentation
and programs when we find we guessed wrong for large numbers of people.

It is much harder to beta documentation than software; simply because
people either won't read it, or don't read it in the same way as somebody
in critical (and fast) need of information...

--
Let no one tell me that silence gives consent,  |   Dave Olson
because whoever is silent dissents.             |   Silicon Graphics, Inc.
    Maria Isabel Barreno                        |   [e-mail]