[comp.unix.ultrix] Partitioning a disk

rr@csuna.cs.uh.edu (Ravindran Ramachandran) (10/20/89)

   First, let me apologize for the dual posting of my previous
article. Something went wrong, and the News program refuses to
let my cancel the article, saying that I'm not the owner.

   Now, for my query. I want to partition my RZ55, which right
now has the default 'a' and 'g' partitions as '/' and '/usr'.
Now, I want to create a new partition '/users' as the 'f'
partition. The SUN computers have a program called format that
lets you choose the values in terms of head, track and sector.
I wish that (or perhaps am ignorant that) such a program exists
for the DS3100s. BTW, why are they called pmax? My real couple
of questions follow:

(1) Is there any method in the madness of the values for the
    default partition?

(2) Is there any minimum value (or any minimum value of which
    it should be a product) for the size of a partition?

    Basically, from where, and of what sizes can my partition
    be so that I make the best use of the disk layout?

Thanks everybody for the ton of information that I usually get
back; helps me see things from various perspectives,

  Regards,
    --Ravi-

alan@shodha.dec.com ( Alan's Home for Wayward Notes File.) (10/20/89)

In article <15365@uhnix1.uh.edu>, rr@csuna.cs.uh.edu (Ravindran Ramachandran) writes:
> 
>    Now, for my query. I want to partition my RZ55, which right
> now has the default 'a' and 'g' partitions as '/' and '/usr'.
> Now, I want to create a new partition '/users' as the 'f'
> partition. 

	Since the A, B and G partitions use the whole disk
	your new F partition will have to overlap some
	existing partition.

> The SUN computers have a program called format that
> lets you choose the values in terms of head, track and sector.
> I wish that (or perhaps am ignorant that) such a program exists
> for the DS3100s. 

	The program you want is chpt(8).  If you want to construct
	a new partition, you give it a beginning sector and length.
	The man page should have enough information for what you
	want.

	On the RZ55 the G partition is already setup to be divided
	into D, E and F partitons of equal length.  If your /usr
	uses enough less than two thirds of the current G partition 
	then you can use the current F and construct a new partiton
	of the bottom two-thirds to be /usr.

	To change the partition table of the system disk:

		0.  Backup /usr.  You might also want to verify
		    that the tape is good by restoring random files
		    from it.

		1.  Shutdown the system down.

		2.  Boot the standalone system and choose the System
		    Management option (#3).

		3.  Use chpt to construct the new partition tables.

		4.  Boot the single user system and newfs to build
		    the new file systems.  You may need to edit 
		    /etc/disktab to reflect the new sizes.

		5.  Restore /usr.

		6.  Edit /etc/fstab to reflect the new file system
		    layout.

> BTW, why are they called pmax? 

	PMAX was the internal name used during development.  The
	PVAX (VAXstation 3100) was being developed at the same
	time, so it may have been the source of the name.  In
	general the intenal name shouldn't be used, but sometimes
	habits are hard to break.

> My real couple of questions follow:
> 
> (1) Is there any method in the madness of the values for the
>     default partition?

	I sometimes wonder, but yes there is.  The A partition
	on the RISC systems is 16 MB raw space.  The B partition
	(normally used for page/swap) is 64 MB and the rest of
	the disk (RZ23 and RZ55) is the G partition.  I would
	suspect that partition tables for DSA disks on RISC
	systems will be similar.

	On VAX systems the A partition is about 7 MB and the
	B partitions varies depending on the size of the disk.
	On the RA81 and smaller disks the B parititon is about
	16 MB.  On the RA82 and RF71 it's about 32 MB.  On the
	RA90 it's about 62 MB.

	The rest of the disk seems to be divided up according
	one of three schemes depending disk size.  On "small"
	disks (RD53, RD54, RZ55, etc) the remainder of the disk 
	is the G parititon.  On the older DSA disks (RA60, RA80
	and RA81) there is small G parition that once upon a
	time was big enough for /usr and the rest of the disk
	was H.  On the new DSA disks (RA70, RA82 and RA90) the 
	G partition is "large" and the H partition is the rest
	of the disk.  "Large" seems to vary depending on the size
	of the disk, but for the RA90 and RA82 the ratio of G
	to H partitions is about 3/4ths : 1/4th.

	The RF30 and RF71 look like they use a scheme similar
	to the older DSA disks.
	
> 
> (2) Is there any minimum value (or any minimum value of which
>     it should be a product) for the size of a partition?
>
	For DSA and SCSI disks they don't seem to care that
	a partition starts any special place.  The minimum
	size depends on usage.  You should always try to 
	maintain at least 10% free space on a disk that will 
	be written to.  If a disk is going to be read-only or
	read-mostly where write performance is less important
	you can get away with less free space.

	If you can predict how much space will be used by a
	file system then you can set the size accordingly.
	File system overhead (inodes mostly) will vary according
	to the number of cylinder groups on the file system.  The
	RF71 for example uses about 7% of the disk for inodes
	due to the large number of cylinders (and therefore cylinder
	groups).

>     Basically, from where, and of what sizes can my partition
>     be so that I make the best use of the disk layout?
>
	Having said all that, I would guess that you will have
	a difficult time repartitioning G to still have room for
	/usr and leave enough for a useful /users file system.
	If it's an option you might be better off getting an RZ23
	for the /users file system.

> Thanks everybody for the ton of information that I usually get
> back; helps me see things from various perspectives,
>
	I hope this ton of infromation will be useful.
 
>   Regards,
>     --Ravi-

-- 
Alan Rollow				alan@nabeth.enet.dec.com

chris@mimsy.umd.edu (Chris Torek) (10/20/89)

In article <453@shodha.dec.com> alan@shodha.dec.com writes:
>PMAX was the internal name used during development.  The
>PVAX (VAXstation 3100) was being developed at the same
>time, so it may have been the source of the name.  In
>general the intenal name shouldn't be used, but sometimes
>habits are hard to break.

Unfortunately, in general the name `PMAX' is the only way to convince
DEC salesdroids that you mean `*DEC*station 3100' and not `*VAX*station
3100'.

Giving two similarly-packaged machines the same number was a big mistake.
-- 
`They were supposed to be green.'
In-Real-Life: Chris Torek, Univ of MD Comp Sci Dept (+1 301 454 7163)
Domain:	chris@cs.umd.edu	Path:	uunet!mimsy!chris