home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.sys.hp
- Path: sparky!uunet!munnari.oz.au!spool.mu.edu!umn.edu!lynx!nmsu.edu!opus!jthomas
- From: jthomas@nmsu.edu (James Thomas)
- Subject: Re: Disk partitioning/layout (Was: New 877 Installation)
- In-Reply-To: jthomas@nmsu.edu's message of Mon, 28 Dec 1992 23:14:10 GMT
- Message-ID: <JTHOMAS.92Dec29171108@navajo.nmsu.edu>
- Sender: usenet@nmsu.edu
- Organization: NMSU Computer Science
- References: <JTHOMAS.92Dec24140611@navajo.nmsu.edu>
- <JTHOMAS.92Dec28161410@navajo.nmsu.edu>
- Date: Wed, 30 Dec 1992 00:11:08 GMT
- Lines: 62
-
- In article <JTHOMAS.92Dec28161410@navajo.nmsu.edu> I write yet again :-)
-
- Jim> a) I would like to be able to define the partition sizes. For
- Jim> example, a stock 48M swap partition may have been OK in the "old"
- Jim> days, but systems have more memory now. The 877 we got was ordered
- Jim> with 128M of memory (gee, somewhere between 3200 and 6400 times the
- Jim> memory in the 1620 I started on ... :-) . It was set up with a 1106M
- oops, too much indecision -----------> 105M
-
- Jim> primary swap space - not enough for a full dump, for example. Suppose
- Jim> I want 145M of primary swap? The choices are 105M, 126M, and 958M.
- Jim> Not very appropriate. What do I do when HP tells me they can't
- Jim> support the system because I can't give them a complete system dump ??
- Jim> :-) :-)
-
- For the C2474S disk, we are allowed 24M and 48M for sections 0 and 1 (or
- vice versa for 15 and 14), followed by 126M, 29M, 105M, and 958M for
- sections 10, 3, 4, and 5. All of the overlapping section options except 7
- include section 5 plus one or more adjacent sections towards the "front" of
- the disk. It is not possible to make a reasonable sized swap area plus a
- single root partition (e.g., S6=2M + S15=200M [swap] + S13=the rest [root])
- or a reasonable root partition plus a reasonable swap area plus a user
- partition (e.g., S6=2M + S0=200M [root] + S1=200M [swap] + S11=the rest
- [user]). If we (the "royal we") insist on making "unix"(tm) the only
- choice, we're going to have to do better than this.
-
- Jim> b) Back 20 years ago, I could tell TOPS-10 where I wanted its
- Jim> equivalent to the inode table. That is still missing in unix. If I
- Jim> set a disk up with only section 2, I would like to be able to say that
- Jim> the inode table should be put smack in the middle (if I don't want a
- Jim> single file, contiguous though unix also can't particulary do that
- ooops again -> particularly
- Jim> anyway). That way seeks are minimized when the relatively common
- Jim> requirement to find some inode information happens.
-
- Picture for the 1.4G disk being used as an example:
-
- section size use configuration
-
- 0 200M root inodes at high end
- 1 200M swap swap
- 11 890M user inodes at low end
-
- or:
- section size use configuration
-
- 0 400M root inodes at high end
- 1 200M swap swap
- 11 690M user inodes at low end
-
- or:
- section size use configuration
-
- 12 1290M user inodes in middle
-
- . That way the inode location can be optimized (maybe even automatically
- by sam :-) .
-
- As it is, I'm stuck with using a 48M swap in section 15 and root in section
- 13 and no other options :-(
-
- Jim jthomas@wsmr-emh82.army.mil
-