[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Automatic allocation of swap



Osamu Aoki wrote:

On Mon, Aug 02, 2004 at 09:21:05AM +0800, John Summerfield wrote:
Osamu Aoki wrote:
On Sun, Aug 01, 2004 at 09:04:50PM +0800, John Summerfield wrote:
By spreading data over multiple partitions with great gobs of free space between small (after install) amounts of data, you're forcing longer seeks.

Yes.  We all know this.  Let's not argue over belief.

If you want us to change something, few ways to do this.

1. clear facts (bench mark result of different configurations)
2. reference some authoritative documents.

and patch always help.  Really, I did not say putting swap at the edge
of disk is a good thing.  I was annoyed by calling this "stupid".

Please do not call action of DD stupid even if they are.


1. You suppose I'm capable of creating a patch. In fact, my C skills almost extend far enough that I can read some code.

2. Belief? It seems so obvious to me, it takes me longer to walk across two streets than it does one.

In the 70s I was a systems programmer on working on IBM mainframes running OS/VS . One of the tasks we took very seriously was the minimising of head movement on our disk drives. So far as I know, the laws of physics that describe their behaviour have not been repealed.

3. I've reread what I said. You misquoted me. I described the RULE OF THUMB as stupid. I don't believe it originated with the DDs.

However, as a ROT, blind adhearance to ROTs is stupid.

Understand the ROT and you will understand its limitations. I rather think that ROT was intended to describe an upper limit. Later interpretations, like intermpretations of Murphy's Law deviate from the truth.


Here is my desktop pc. It's performance sucks bunnies through capilliary tubes:
summer@Dolphin:~/d-i$ /sbin/swapon -s
Filename Type Size Used Priority
/var/swapfile                           file            524280  455604  -1
/var/swapfile2                          file            524280  69296   -2
summer@Dolphin:~/d-i$ df -lh
Filesystem            Size  Used Avail Use% Mounted on
/dev/hda3              29G   27G  639M  98% /
                                         ^
                                         |
                                     no wonder, sigh.

tmpfs                 189M     0  189M   0% /dev/shm
/dev/hda1              23M   21M  1.4M  94% /boot
summer@Dolphin:~/d-i$
...
That's almost certainly bad. You probably don't spend much time reading program files and documentation in comparison with the time reading and writing variable data.

I have been reading and editing documentation a lot.  Please install
debian-reference-en package from Sarge :-)  Then you know what.

http://www.debian.org/doc/manuals/debian-reference/reference.en.html

Oh, please read install guide too.  That has good amount of partitioning
etc. too.
http://www.debian.org/releases/stable/installmanual

Well, I was really referring to your (plural - many DDs) computers.

You're a software developer writing in C? Would you say, less than a second to read gcc then a minute or so for a compile? Sure, it depends, but the ratios will be somewhat like those.

I usually do not write C nor compile it much.  (With my old 386, I used
to compile my kernel or patched-X for a day, though.)

Really, Linux or Windows, it is bad idea to fill actively used disk up
to 98%.  (For your case / partition.  /boot may be OK since it is
practically read-only.)  For Linux, 90-95%, for windows 60-70% is my
common sense usage.
I agree with that. Breaking the disk into lots of partitions makes it more likely that one will fill andcause problems. Recently I installed a server running Sarge and I find / is overfilled to the point I have problems installing packages:
ns:~# df -t ext3 -m
Filesystem           1M-blocks      Used Available Use% Mounted on
/dev2/root2                186       176         1 100% /
/dev/sda1                   32        18        13  58% /boot
/dev/sda3                 3024       692      2179  25% /usr
/dev/sda4                 5378      4714       391  93% /var
/dev/sdb1                17241     12973      3394  80% /home/users
/dev/hda2                37572     28382      7664  79% /mnt/rhl
/dev/hdb2 76905 47356 27987 63% /mnt/rhl/var/ftp/pub/linux
ns:~#

Cheers,

Osamu

PS: FYI
  None of these habits should be carried over to Linux and ext2. Linux
  native file systems do not need defragmentation under normal use and this
  includes any condition with at least 5% of free space on a disk. There is
  a defragmentation tool for ext2 called defrag, but users are cautioned
against casual use. ..... From: Linux Partitioning mini-FAQ http://pw1.netcom.com/~kmself/Linux/FAQs/partition.html
I don't know why you're telling me Linux is better than Windows. I've never been a regular Windows user: I used to us OS/2, and before that MSDOS & DRDOS. (concurrent) CPM/M-86, CP/M, BewdOs80. Not Windows though.


  He advocates 3x RAM for swap.
You should read that in context. He also makes it clear that's for everyone. eg VA Linux uses 1x.

Besides that, the implied ording of partitions is wrong. Two or three swap partitions on one disk makes no sense at all: if Karsten thinks otherwise, he needs to justify it.



If you think my arguments are wrong, explain why they're wrong.

--

Cheers
John

-- spambait
1aaaaaaa@computerdatasafe.com.au  Z1aaaaaaa@computerdatasafe.com.au
Tourist pics http://portgeographe.environmentaldisasters.cds.merseine.nu/



Reply to: