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

Memory management



Hi all. Ever since I went from kernel 2.0.36 to 2.2.10 (Still glibc2.0),
my system has been having an interesting memory management.
1. When Iboot the syste up, memory consumed is right where it should be.
Shortly after a startup, memory usage goes from 20M (Right after boot) to
completely fill the physical Ram. top shows a large amount of data in
buffered memory.
Once I run some programs like Netscape, memory usage goes up (as it
should) but upon closure of application it does not completely return to a
normal state as it was before, instead a lot of memory is being cached.
Here is a record of top at the moment, just running X (12x10@24) and 2
aterms with a xchat:

CPU states:  6.2% user,  1.3% system,  0.0% nice, 92.6% idle
Mem:   47180K av,  45904K used,   1276K free,  10496K shrd,    980K buff
Swap: 110808K av,  13732K used,  97076K free                 15292K cached

Note the large amount of memory cached. 
Now I start Netscape.

CPU states:  5.8% user,  1.1% system,  0.0% nice, 93.1% idle
Mem:   47180K av,  45936K used,   1244K free,  12628K shrd,    980K buff
Swap: 110808K av,  14804K used,  96004K free                 11816K cached

I don't know what to make of it. Is the system keeping some apps in memory
for better performance at startup? This is a problem because of the
swapping that takes place while running programs.
Please help to fix this (a full upgrade to slink would be out of the
question, because I simply dont have time to reinstall the system and
recompile apps that need recompiling.)
This has happened with 2.2.10 and 2.2.11 kernels. Even if I revert back to
2.0.36, this problem still persists.
Any ideas?
Andrew

-------------------------------------------------------------------------
 Andrei S. Ivanov                  
 Scorpio@hushmail.com
 c680789@showme.missouri.edu
 UIN 12402354                      
 http://scorpio.myip.org    <--All the pages bundled together.
---------------------------------------------------------------------------


Reply to: