SUMMARY: Q: swap space

From: Tuna Tugcu (
Date: Tue Feb 24 1998 - 09:33:18 CST

   Hi SMs,
   Sorry for another late summary; well, may be it is not even worth for
a summary since it turned out to be unrelated to Solaris. The error was
due to an error in the tool and some mistakes in my simulation model. I
applied a patch supplied by mil3 and also corrected the errors in my
model to solve the problem.

   Tuna Tugcu

Special thanks go to:
SW <>
Joseph S D Yao <> (Mr Rene Occelli)
Matt Reynolds <> (Brad Young )
Casper Dik <casper@holland.Sun.COM>

Here is the original question:
   Dear SMs,
   I have made some changes on vfstab to increase my swap space
2.5). Now I'm having some memory problems with a simulation tool.
Before accusing the software for improper use of memory I want to make
sure that I haven't done something wrong.
   Below you can find the related lines from vfstab (two disks
/dev/dsk/c0t0d0s1 - - swap - no -
swap - /tmp tmpfs - yes -
/dev/dsk/c0t2d0s1 - - swap - no -
/dev/dsk/c0t0d0s4 - - swap - no -

   Here is the output of "swap -l":
swapfile dev swaplo blocks free
/dev/dsk/c0t0d0s1 32,1 16 262944 252016
/dev/dsk/c0t2d0s1 32,17 16 262944 253216
/dev/dsk/c0t0d0s4 32,4 16 293344 283312

   I'm looking forward to your replies.


Here are the replies:

From: SW <>

Please check your /etc/system see if you have setup shared memory.
My users running big imaging analysis applications on Solaris2.5,
which needs an entry in the file like the following line:

set shmsys: shminfo_shmmax = 512000000

And you need to reboot the machine after you added the line.

Good luck!

From: Joseph S D Yao <>

Go into format / partition, and make sure that no partitions overlap!

Also, I don't remember exactly, but ISTM that Sun's swap schema is one
of the ones that interleaves partitions. In that case, your swap size
would be (number of partitions) X (size of smallest partition). You
might want to double-check that (a) the swap partitions are indeed the
same size [without overlapping others], and (b) Sun's swap does in fact
interleave as I have said.

Joe Yao - Joseph S. D. Yao
(temporarily relocated)
From: (Mr Rene Occelli)


The entry looks correct.

Can you send me the output of swap -s

From: Matt Reynolds <>

  Make sure /dev/dsk/c0t0d0s1 and /dev/dsk/c0t0d0s4 don't overlap at
all. (Do a "format" - "partition" - "print" to see the partition

  It shouldn't work if they overlap, but you might find it just
coincidentally works for a while before hitting a part of the shared


# Matthew Reynolds,                | Email:     #
# Aston Space Geodesy              | Web:  #
# Aston University, Birmingham,    | Phone: (+44) 121-359-3611 x4552 #
# B4 7ET. England, Earth.          | Fax:   (+44) 121-333-3389       #
From: (Brad Young )

> Here is the output of "swap -l": > swapfile dev swaplo blocks free > /dev/dsk/c0t0d0s1 32,1 16 262944 252016 > /dev/dsk/c0t2d0s1 32,17 16 262944 253216 > /dev/dsk/c0t0d0s4 32,4 16 293344 283312

Having two swap partitions on a single spindle, as above with c0t0, can = cause performance problems. Think of all the extra travel the write heads for = the c0t0 spindle has to do when writing to two separate areas on the disk.

Good Luck, Brad ************************************************************************ From:

Here's the equivalent on my 2.5.1 1000E server :

# swap -l swapfile dev swaplo blocks free /dev/md/dsk/d4 85,4 8 525160 464336 /dev/md/dsk/d5 85,5 8 1026136 965440

# grep swap /etc/vfstab swap - /tmp tmpfs - yes - /dev/md/dsk/d4 - - swap - no - /dev/md/dsk/d5 - - swap - no -

So, your entries look fine to me. ************************************************************************ From: Casper Dik <casper@holland.Sun.COM>

I'm not sure how much memory your tool needs, but it seems that almost none of your swap is in use.


This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:12:31 CDT