Swap question

Swap question

Post by Dean Jan » Thu, 07 Nov 1996 04:00:00



I have a (hopefully quick) question for the SCO gurus:  I have a system
with the following:

# memsize
83488768

# swap -l
path               dev  swaplo blocks   free
/dev/swap          1,41      0 245760 245760

2:35pm  up  2:27, 27 mins,  36 users,  load average: 6.88, 6.62, 6.86

What seems strange to me, is no matter how high the load goes (it climbs
to 10 or so) I never use any swap space.  Is swap -l telling the truth?  
sar -r tells me the same, swap is never used.  I'm new to SCO, but not
Unix, my gut feeling is something isn't right here.  I expect some swapping
with that number of users and load, even under lighter usage you'd expect
some swap space usage.  Am I just not understanding SCO memory managment?
Am I getting lucky and having the working sets of the processes so tiny they
all fit in mem?

Thanks for any insight.
--
---------
  ////
 |  00   -Dean Jansa

  \ ~/    University of Wisconsin-Madison, Computer Sciences Systems Lab
---------

 
 
 

Swap question

Post by Jerry Heym » Thu, 07 Nov 1996 04:00:00


: I have a (hopefully quick) question for the SCO gurus:  I have a system
: with the following:

: # memsize
: 83488768

: # swap -l
: path               dev  swaplo blocks   free
: /dev/swap          1,41      0 245760 245760

: 2:35pm  up  2:27, 27 mins,  36 users,  load average: 6.88, 6.62, 6.86

: What seems strange to me, is no matter how high the load goes (it climbs
: to 10 or so) I never use any swap space.  Is swap -l telling the truth?  
: sar -r tells me the same, swap is never used.  I'm new to SCO, but not
: Unix, my gut feeling is something isn't right here.  I expect some swapping
: with that number of users and load, even under lighter usage you'd expect
: some swap space usage.  Am I just not understanding SCO memory managment?
: Am I getting lucky and having the working sets of the processes so tiny they
: all fit in mem?

Possibly.  A good tool to have/use is u386mon.  This will allow you to look
at all sorts of kernel information as the system is running.  Things like
disk i/o, memory (vs. swap) used, process running/in run queue, etc.

Depending on the version of SCO you're using (uname -X will give it to you),
you'll need the appropriate binary.  I know that u386mon for SCO 5 is available
via ftp from SCO (look for the Skunkworks directories from the WWW at
www.sco.com)

: Thanks for any insight.
: --
: ---------
:   ////
:  |  00   -Dean Jansa

:   \ ~/    University of Wisconsin-Madison, Computer Sciences Systems Lab
: ---------

jerry
--

Combat System Simulation                   | http://www.arlut.utexas.edu/~jerry
Applied Research Laboratories of UT Austin | "Software is the difference
P.O. Box 8029, Austin, TX 78713-8029       |     between hardware and reality"

 
 
 

Swap question

Post by Bela Lubki » Fri, 08 Nov 1996 04:00:00



> I have a (hopefully quick) question for the SCO gurus:  I have a system
> with the following:

> # memsize
> 83488768

> # swap -l
> path               dev  swaplo blocks   free
> /dev/swap          1,41      0 245760 245760

> 2:35pm  up  2:27, 27 mins,  36 users,  load average: 6.88, 6.62, 6.86

> What seems strange to me, is no matter how high the load goes (it climbs
> to 10 or so) I never use any swap space.  Is swap -l telling the truth?  
> sar -r tells me the same, swap is never used.  I'm new to SCO, but not
> Unix, my gut feeling is something isn't right here.  I expect some swapping
> with that number of users and load, even under lighter usage you'd expect
> some swap space usage.  Am I just not understanding SCO memory managment?
> Am I getting lucky and having the working sets of the processes so tiny they
> all fit in mem?

You have 80MB of RAM.  It's quite possible -- likely, given your
findings -- that their combined processes are in fact consuming less
than 80MB of RAM.

Examine `ps -elf` output.  Get tls518 from ftp.sco.com:/TLS, which
contains `memhog`, a utility that will show top memory consumers.

The machine I'm posting on has 22 users logged in, 128MB of RAM, 73MB
free RAM, and no swap space in use.

Your load average may be unrealistically bloated, there's a bug that
causes NFS server processes to record a constant load of X.0, when in
fact they aren't causing any processing at all.  See:

  http://www.sco.com/cgi-bin/ssl_reference?430045

- Show quoted text -

Quote:>Bela<

 
 
 

Swap question

Post by Nick Bree » Wed, 13 Nov 1996 04:00:00


: I have a (hopefully quick) question for the SCO gurus:  I have a system
: with the following:
: # memsize
: 83488768
: # swap -l
: path               dev  swaplo blocks   free
: /dev/swap          1,41      0 245760 245760
: 2:35pm  up  2:27, 27 mins,  36 users,  load average: 6.88, 6.62, 6.86

: What seems strange to me, is no matter how high the load goes (it climbs
: to 10 or so) I never use any swap space.  Is swap -l telling the truth?  

It depends a great deal on your applications.  Our college's student
UNIX server (runs primarily text-based WordPerfect, email, newsgroup
reading, and the Library card-catalog) had only 80 megs until
a couple of months ago, and handled 70-90 users quite handily.  This
year we've added a new Library system and our user count keeps going
up, so we've gone up to 128 megs.

Our maxim with a major server like that has always been that when it
swaps, we add more RAM.

: sar -r tells me the same, swap is never used.  I'm new to SCO, but not
: Unix, my gut feeling is something isn't right here.  I expect some swapping
: with that number of users and load, even under lighter usage you'd expect
: some swap space usage.  Am I just not understanding SCO memory managment?
: Am I getting lucky and having the working sets of the processes so tiny they
: all fit in mem?

I think this is it.

In the PC world, we get used to thinking of A P150 w/32 megs
being how much one person needs.  With a "Real OS", that could
run 10-15 people.  :-)  (Of course, these 10-15 people have
to be people who are satisfied with a text-based interface, who
are becoming increasingly difficult to find....)

Nick
--
---------------------------------------------------------------------

Systems Analyst & Programmer                           (712) 722-6299
Dordt College, Sioux Center, IA
---------------------------------------------------------------------

 
 
 

Swap question

Post by rnar.. » Fri, 15 Nov 1996 04:00:00


hello

our sco system has a wantek qic-02 tape drive that is not functioning
properly.  the os system is sco unix 4.2 v 3.2.  the type of error
messages include 'tape read error', 'io error', etc.

the tape will not read or write, but when addressed, it will try, try, and
then hang.

it will respond to most commands, including tape reset.  we have cleaned the
heads, reseated the controller, reconnected the cables...

any ideas?  thanks for your help.


 
 
 

Swap question

Post by Jeff Lieberma » Mon, 18 Nov 1996 04:00:00


: our sco system has a wantek qic-02 tape drive that is not functioning
: properly.  the os system is sco unix 4.2 v 3.2.  the type of error
: messages include 'tape read error', 'io error', etc.
: the tape will not read or write, but when addressed, it will try, try, and
: then hang.

If you are getting io error or read error, it's trying to read
and not doing a very good job of it.  My guess is that if you
can hear the tape move (drive attempting to read or write) then
the head is worn beyond redemption.

An important clue would be if this happened suddenly or gradually.
If suddenly, look for connector problems, cable problems, filth,
mechanical problems, and other physical damage.  If it happened
gradually, look for a worn head (the slightest gouge and it's gone),
loose head mounting screws, slipping belt drive, more filth, and
wrinkled tapes.

[x] Email to author  [ ] To mailing list  [x] Posted to newsgroup
--
# Jeff Liebermann  Liebermann Design  150 Felker St #D  Santa Cruz  CA  95060

# 408.699.0483 digital_pager    73557,2074  cis [don't]

 
 
 

Swap question

Post by Richard T. Naru » Mon, 18 Nov 1996 04:00:00


The tape problem seems to have happened over time...however, I have just recently
walked into this problem, and am attempting to assist a friend.  Since the original
post, I have acquired a used loaner replacement drive...unfortunately it responds in
exactly the same fashion...drive spins, io error on read/write.  All cables have
been re-checked/replaced...any other guesses?

...rtn

 
 
 

Swap question

Post by Richard T. Naru » Mon, 18 Nov 1996 04:00:00


hello

i am reposting since i was sleeping when i posted the original, and
inadvertantly attached to a thread, rather than created a new post...

---
our sco system has a wantek qic-02 tape drive that is not functioning
properly.  the os system is sco unix 4.2 v 3.2.  the type of error
messages include 'tape read error', 'io error', etc.

the tape will not read or write, but when addressed, it will try, try, and
then hang.

it will respond to most commands, including tape reset.  we have cleaned the
heads, reseated the controller, reconnected the cables...

any ideas?  thanks for your help.


---

Jeff Liebermann replies...

If you are getting io error or read error, it's trying to read
and not doing a very good job of it.  My guess is that if you
can hear the tape move (drive attempting to read or write) then
the head is worn beyond redemption.

An important clue would be if this happened suddenly or gradually.
If suddenly, look for connector problems, cable problems, filth,
mechanical problems, and other physical damage.  If it happened
gradually, look for a worn head (the slightest gouge and it's gone),
loose head mounting screws, slipping belt drive, more filth, and
wrinkled tapes.

[x] Email to author  [ ] To mailing list  [x] Posted to newsgroup
--
# Jeff Liebermann  Liebermann Design  150 Felker St #D  Santa Cruz  CA  95060

# 408.699.0483 digital_pager    73557,2074  cis [don't]

---

rnarum replies...

The tape problem seems to have happened over time...however, I have just
recently walked into this problem, and am attempting to assist a friend.  
Since the original post, I have acquired a used loaner replacement
drive...unfortunately it responds in exactly the same fashion...drive spins,
io error on read/write.  All cables have been re-checked/replaced...any other
guesses?

 
 
 

Swap question

Post by Bill Vermilli » Wed, 20 Nov 1996 04:00:00




Quote:>The tape problem seems to have happened over time...however, I have just recently
>walked into this problem, and am attempting to assist a friend.  Since the original
>post, I have acquired a used loaner replacement drive...unfortunately it responds in
>exactly the same fashion...drive spins, io error on read/write.  All cables have
>been re-checked/replaced...any other guesses?

I've seen this many times.   It's the controller card.  You are
going to have to hunt around to find one - they don't make them
any more.

Last one I had a problem with I had to replace with a SCSI
drive and controller.

--

 
 
 

Swap question

Post by Jim Kot » Fri, 22 Nov 1996 04:00:00


Richard,

If it is the controller causing the problem with the QIC-02 tape drive, and
you're in the US, there's a third-party computer maintenance firm that I
signed up with at the last company I worked. They have field technicians in
some 40 or 45 states and had GREAT service. Their name is Scantron-FPC, phone
number 1-800-228-3628.

--
Jim Kotan
                ----- Disclaimer -----
All comments, statements, recommendations and/or opinions are mine
alone. Nobody else's. I am not an expert at anything. Except maybe at
doing or saying stupid things.

 
 
 

1. Newbie: swap question

Hello all,

I have just recently dived into the world of Linux - RedHat6.1

In the RH install docs, it said that I should make my swap "16MB or the
size of my RAM, whichever is more".  In my case that happens to be 160MB
... so I did it.  Then I read the Linux FAQ.  There is a question "How can
I have more than 128MB of swap?"  I didn't even know that 128 is the limit!

So my question(s) is(are):
Is there any "optimal" way to cut up the swap spaces, i.e.: make two of the
same size, one 128 and the other the remainder?
Is there a reason why you would ever want to have more swap?

Thanx, SiKing

2. nfs permission denied

3. newbie swap question

4. modem problem

5. Byte swapping question (Intel v. Next)

6. 'update' eats all my CPU time!!

7. swap question

8. Why can't I uudecode?

9. CPU Swapping Question

10. SWAP Question

11. swap question

12. NetBSD Swapping question

13. Kernel Hot-swap question.