[comp.sys.dec] disk partitions - need advice

mlee@attctc.Dallas.TX.US (Michael Lee) (01/19/90)

We are running Berkeley 4.3 on a VAX 8650.

We have one ra60 disk drive,
	six ra81 disk drives,
and	two ra82 disk drives.

Most of the partitions are about 93 Megabytes.
All the others are smaller.

I am making plans to change the file system partitions to something
closer to the size of one tape each (160 Megabytes @ 6250 bpi).

An ra81 has a capacity of about 450 Megabytes (raw) so dividing that by 3
gives 150 Megabytes per partition.

An ra82 has a capacity of about 620 Megabytes (raw) and I plan to put the
kernel and swap space on them (about 20 Megabytes) which leaves 600
Megabytes.  Dividing that by 4 gives 150 Megabytes per partition.

The ra60 has about 205 Megabytes and I plan to split that into smaller
pieces and put a spare kernel and other small file systems on it.

My questions are: 

I have read that:

1. One should break up the partitions into as many different partitions
   as possible.

2. It is only important to break up /, /usr, and /tmp onto different
   partitions.

These two guidelines seem contradictory.

Also, the partitions provided by Berkeley and Ultrix are slightly
different.  For an ra82, the total space under Berkeley is 620 Megabytes
and under Ultrix is 622 Megabytes.  I have never seen any explanation of
the partition sizes except for swap, /, and /tmp.

***
***
Also: When I run newfs or mkfs, I get a warning: unallocated blocks.
Should the partitions be selected to minimize the number of wasted blocks?

Should a new partition always start on a new cylinder or track or
something?  If so, how can I make sure it does?
***

I started thinking about this because the system I inherited has 38
file systems and the largest one is 93 Megabytes which is about 60%
of one tape at 6250 BPI.  The new larger file systems will fill each tape
to about 80 to 90%.  This will be closer to optimal for dumps and will
minimize tape handling for restores.

Maybe I am be missing something.  What?

Please e-mail and I will post a summary.

Thanks,

Michael Lee
uunet!cs.utexas.edu!necssd!necbsd!mlee
214-518-4608